[Global Issues in Water Policy] Economic Modeling of Water Volume 3 ||

Economic Modeling of Water

GLOBAL ISSUES IN WATER POLICYVOLUME 3

Series Editors

Ariel Dinar

José Albiac

Eric D. Mungatana

Víctor Pochat

Rathinasamy Maria Saleth

For further volumes:http://www.springer.com/series/8877

Glyn WittwerEditor

Economic Modeling of Water

The Australian CGE Experience

EditorDr. Glyn WittwerCentre of Policy StudiesMonash UniversityWellington Road11th Floor, Menzies Building 11EClayton, VIC [email protected]

ISSN 2211-0631 e-ISSN 2211-0658ISBN 978-94-007-2875-2 e-ISBN 978-94-007-2876-9DOI 10.1007/978-94-007-2876-9Springer Dordrecht Heidelberg New York London

Library of Congress Control Number: 2012934055

© Springer Science+Business Media Dordrecht 2012This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifi cally the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfi lms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. Exempted from this legal reservation are brief excerpts in connection with reviews or scholarly analysis or material supplied specifi cally for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work. Duplication of this publication or parts thereof is permitted only under the provisions of the Copyright Law of the Publisher’s location, in its current version, and permission for use must always be obtained from Springer. Permissions for use may be obtained through RightsLink at the Copyright Clearance Center. Violations are liable to prosecution under the respective Copyright Law.The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specifi c statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use.While the advice and information in this book are believed to be true and accurate at the date of publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors or omissions that may be made. The publisher makes no warranty, express or implied, with respect to the material contained herein.

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

v

By 2030, the OECD predicts that over half the world’s population will be living with water scarcity. By this they mean that these people will be living in a world where water availability, or more correctly the lack of it, limits economic opportunity. To get to the bottom of this issue, one needs to understand how local water supply conditions infl uence water use and what changes in availability mean for local, regional, and national economies.

Before the advent of TERM – The Enormous Regional Model – any discussion about the likely impacts of changes in water policy or changes in supply tended to be based on a combination of some partial analysis coupled with speculative assertions about fl ow on effects to other sectors. As any CGE modeler will tell you, in the complex world we live in, changes in one sector often have counterintuitive implications for other sectors. Assertion making is a risky and unwise business to be in. The recommended approach is to use a model to estimate the likely impacts of a change in one sector on all other sectors. Expect to be surprised.

For many problems, fi ne scale insights are needed. One needs to know which industries in which towns will gain income and which will lose income. It is impor-tant to know how quickly people can adjust. TERM was built to allow such analysis. Think about 50 regions each with 170 sectors. Yes, TERM is enormous, and yes, it is constructed from the “bottom up.” But, by taking a bottom-up approach, the detail that determines the fate of any region and its relationship with all other regions can be captured realistically.

This book shows how such models can be built. But the book does not stop there. Catchments and rivers have little respect for statistical survey boundaries so TERM’s architects have spent considerable time working out how to convert a conventional regional CGE model into one that respects catchment and other biophysical boun-daries. This is no easy task but, once completed, the resultant model is extremely powerful. The power comes from the use of regional boundaries that are consistent with the very same regions that people are arguing about.

TERM-H2O demonstrates the potential of this approach. The boundaries used align with catchment, not statistical division boundaries, and the amount of water available for use in each region described precisely. Objective exploration of the effects of water

Foreword

vi Foreword

scarcity, policy changes, and government expenditure becomes possible. Moreover, because the boundaries used align with catchment boundaries, it is diffi cult for water managers to dismiss the results as irrelevant. Instead, they are given a platform that allows them to examine impacts at local, regional, and national levels.

The development of TERM-H2O and, more importantly, the completion of this book enable others to see how to build such a model. It represents an important breakthrough.

The power of models like TERM-H2O to bring objectivity to complex political issues is important. This is best demonstrated in Chaps. 6 and 7 of this book.

Chapter 6 is about the impacts of a government water entitlement buyback scheme in Australia’s Murray Darling Basin. Water entitlements are traded throughout this region, and in an attempt to resolve over-allocation problems, the government has been purchasing water entitlements and transferring them to a body responsible for making water available for environmental purposes. Many irrigation communities are strongly opposed to this buyback program because they perceive the resultant capital fl ight would destroy their livelihood. TERM-H2O shows that the reverse is the case. Buyback programs increase economic activity in the region.

TERM-H2O has also been used to show that most of the adverse fi nancial impacts experienced by irrigators in recent times can be explained by the severity of the drought not government policy reforms (see Chap. 7 ). Variants of TERM work in urban, as well as rural areas, and in Chap. 8 one can see how models like this can be used to assess the merits of different water infrastructure and demand management options. Powerful insights about the economic wisdom of different investment strategies emerge.

In summary, the power of modeling systems like TERM-H2O has proven to be greater than many people had expected. This power comes from the richness that fl ows from the construct of models whose regions align with catchments rather than broader statistical areas, have hydrological integrity, and allow objective exploration of options at a level of detail and complexity consistent with the way people talk around a dinner table.

Is this approach generally applicable? The answer is a resounding yes – read Chap. 9 .

The world we are living in is changing rapidly and becoming increasingly complex; the approach taken in this book is one that should be applied to all problems. The future will be much better if we explore options carefully and avoid listening to those who make assertions that cannot be shown to be real.

Prof. Mike Young Executive Director, The Environment Institute

The University of Adelaide, Australia

Reference

OECD (2009) Managing water for all: an OECD perspective on pricing and fi nancing. OECD, Paris

vii

Multiregional national CGE modeling took a dramatic turn in 2002 when Mark Horridge devised a new approach to regional representation in TERM (The Enormous Regional Model). The Australian version of this new model became available just in time to undertake modeling of the 2002–2003 drought. The new model was based on a massive master database which had to be aggregated to undertake any simulation. In theory, this implied that the model could be aggregated to focus on any number of issues in the Australian economy. In practice, water issues have dominated the model’s use.

Starting in 2003, various government agencies including the Productivity Commission, the Murray-Darling Basin Authority, and Victoria’s Department of Primary Industries have commissioned studies concerning the Murray-Darling Basin that required use of the model. CSIRO funded a study of rural and urban water usage. The Productivity Commission has also sponsored database develop-ment that has been important in improving the model. Consulting fi rms, including Frontier Economics, Marsden Jacob Associates, and Deloitte Touch Tohmatsu, have subcontracted work to the Centre of Policy Studies requiring TERM.

Two Australian Research Council grants have been instrumental in TERM-H2O development. The fi rst (LP0667466) was undertaken through a linkage with Victoria’s Department of Sustainability and the Environment. The second (DP0986783) pro-vided the resources to bring this volume into being.

A number of people in government departments and consulting fi rms mentioned above have assisted us in various ways in developing TERM-H2O, thereby bringing this volume into being. I thank in particular Michael Vardon for ongoing guidance on database development, and Mike Young, who remains an inspiration for others pursuing water issues in Australia and the rest of the world. Nadya Ivanovna provided invaluable background information for the fi nal chapter.

Glyn Wittwer

Preface

ix

1 Practical Policy Analysis Using TERM ................................................... 1Glyn Wittwer

Part I The TERM Approach

2 The TERM Model and Its Database ....................................................... 13Mark Horridge

3 Introducing Dynamics to TERM ............................................................. 37Glyn Wittwer and George Verikios

Part II Water Modeling

4 Water Resources Modeling: A Review .................................................... 59Marnie Griffi th

5 The Theory of TERM-H2O ...................................................................... 79Peter B. Dixon, Maureen T. Rimmer and Glyn Wittwer

6 Buybacks to Restore the Southern Murray-Darling Basin .............................................................................. 99Peter B. Dixon, Maureen T. Rimmer and Glyn Wittwer

7 The Economic Consequences of a Prolonged Drought in the Southern Murray-Darling Basin ................................... 119Glyn Wittwer and Marnie Griffi th

8 Urban Water Supply: A Case Study of South-East Queensland ........................................................................ 143Glyn Wittwer

Contents

x Contents

9 Applying TERM-H2O to Other Countries ............................................. 163Glyn Wittwer

About the Authors ........................................................................................... 179

Index ................................................................................................................. 183

xi

Note : All authors are at the Centre of Policy Studies

Peter B. Dixon Centre of Policy Studies , Monash University , Melbourne , VIC , Australia

Marnie Griffi th Centre of Policy Studies , Monash University , Melbourne , VIC , Australia

Mark Horridge Centre of Policy Studies , Monash University , Melbourne , VIC , Australia

Maureen T. Rimmer Centre of Policy Studies , Monash University , Melbourne , VIC , Australia

George Verikios Centre of Policy Studies , Monash University , Melbourne , VIC , Australia

Glyn Wittwer Centre of Policy Studies , Monash University , Melbourne , VIC , Australia

Contributors

xiii

Fig. 2.1 The TERM fl ow database ................................................................. 18 Fig. 2.2 TERM production structure ............................................................. 22 Fig. 2.3 TERM sourcing mechanisms ........................................................... 23 Fig. 2.4 Statistical divisions in Australia ....................................................... 32 Fig. 2.5 Aggregating from master database

to policy simulation (watershed) regions ......................................... 33 Fig. 2.6 Producing regional databases for MMRF and TERM ..................... 34

Fig. 3.1 Outline of preparation of dynamic TERM ....................................... 44

Fig. 5.1 Production function for a farm industry ........................................... 83 Fig. 5.2 Data generation procedure for TERM-H2O .................................... 93 Fig. 5.3 Regions available in TERM-H2O .................................................... 95

Fig. 6.1 Price of irrigation water in SMDB ($ per megalitre) ....................... 102 Fig. 6.2 Buyback-induced percentage effects on real GDP:

TERM-H2O result and back-of-the-envelope calculation ............... 103 Fig. 6.3 Demand for irrigation water in SMDB in 2018 ............................... 104 Fig. 6.4 Map of SMDB regions used in the buyback simulation .................. 106

Fig. 7.1 Map of SMDB regions in TERM-H2O ............................................ 131 Fig. 7.2 Macroeconomic outcomes for SMDB

(% deviation from forecast) ............................................................. 132 Fig. 7.3 Downstream processing and farm capital, SMDB

(% deviation from forecast) (Used and available capital in SMDB for the aggregate of meat products, dairy products, wine and other beverages, and fl our and processed cereals) ...................................................... 133

Fig. 7.4 Employment and industry contributions to GDP, Lower Murrumbidgee (% change relative to forecast) .................... 135

List of Figures

xiv List of Figures

Fig. 7.5 Industry contributions to GDP, all SMDB (% change relative to forecast) ......................................................... 135

Fig. 7.6 Price of water in buyback scenario: droughts in 2015 and 2020 ............................................................... 136

Fig. 7.7 MDB farm output: buyback scenario ($m output relative to forecast) ........................................................ 137

Fig. 8.1 Map of south-east Queensland ......................................................... 145 Fig. 8.2 Impact of the Traveston dam project

on south-east Queensland’s labour market ....................................... 155 Fig. 8.3 Impact of project on south-east

Queensland’s GRP, capital and labour ............................................. 155 Fig. 8.4 Impact on south-east Queensland’s

aggregate consumption and investment ........................................... 156 Fig. 8.5 Contribution of trade to overall deviation

in south-east Queensland’s real GDP ............................................... 157

xv

Table 1.1 Irrigation water use and price in the Murray-Darling Basin .......................................................... 6

Table 2.1 Main sets of the TERM model ....................................................... 19

Table 5.1 Farm industries in region d in the input-output data for TERM-H2O ....................................... 81

Table 6.1 Buyback-induced percentage deviations in farm outputs in SMDB regions in 2018 ..................................... 105

Table 6.2 Prices of permanent water rights ($ per ML, 2009 prices) ............ 117

Table 7.1 Water consumption (GL) by crop in the Murray-Darling basin, 2001–02 to 2005–06 ................................. 122

Table 7.2 Impacts of drought by region, 2007–08 relative to no-drought baseline (%) ................................................ 127

Table 7.3 Comparing modeled SMDB outcomes to observed changes ....................................................... 130

Table 7.4 GDP defl ator, income and household spending in Lower Murrumbidgee ................................................ 133

Table 7.5 Data used in irrigation water price regression ............................... 138

Table 8.1 Population growth: south-east Queensland, other mainland capitals and rest of Australia ................................ 144

Table 8.2 Major water supply projects in south-east Queensland ................. 148

Table 9.1 Countries included in the UN survey on water accounts ............... 170

List of Tables

1G. Wittwer (ed.), Economic Modeling of Water: The Australian CGE Experience, Global Issues in Water Policy 3, DOI 10.1007/978-94-007-2876-9_1, © Springer Science+Business Media Dordrecht 2012

Abstract TERM entails a more detailed representation of regional economies than any previous multi-regional national CGE model. This has broadened the array of policy topics covered using a CGE model. In this book, we concentrate on applica-tions of TERM to rural and urban water issues.

Keywords CGE modeling • Major projects • Regional modeling • Water allocation • Water infrastructure

1.1 The Need for Regional Economic Detail

Regional analysts have often yearned for greater regional detail in input-output and CGE models. In some countries, published input-output tables are available for many regions. For example, Okuda et al. ( 2004 ) took advantage of tables published for 30 of China’s 31 provinces and municipalities to devise a multi-regional input-output model. Another approach is that taken by the IMPLAN team in the United States. 1 Using technologies (that is, intermediate and primary factor costs shares) from the national published input-output table and a number of small-region data sources, IMPLAN is capable of devising input-output tables for any state or county within the USA. Okuda et al. ( 2004 ) estimated inter-regional trade matrices, whereas the IMPLAN approach does not. Each approach has its advantages, the former by linking economies together, and the latter by providing input-output detail across small regions.

G. Wittwer (*) Centre of Policy Studies , Monash University , Wellington Rd, Clayton campus , 3800 Clayton , VIC , Australia e-mail: [email protected]

Chapter 1 Practical Policy Analysis Using TERM

Glyn Wittwer

1 See http://implan.com/v4/index.php

http://implan.com/v4/index.php

2 G. Wittwer

There are issues concerning resource constraints in input-output analysis. Public funds poured into a particular region may provide impressive outcomes for that region, but are they are in the national interest? That is, the public funds might be put to better use elsewhere in the economy, though of less benefi t to that particular region. Moreover, not all of the response to a regional stimulus is likely to quantita-tive. In the short term, a stimulus may increase demand for a fi xed stock of housing, driving up prices. This may suit industries that are direct benefi ciaries of the stimulus, but there may also be losers. Employees in industries that are not directly affected by the stimulus may face local cost pressures, particularly if they rent rather than own their housing. This has been the case during Australia’s most recent mining boom that started around 2004. Port Hedland, in particular, has grappled with extremely expensive housing. 2

Some regional governments no longer accept analysis undertaken using an input-output model. 3 The way forward is to combine a methodology that includes resource constraints with small regional representation. To do this, it comprehensively requires detailed linkages between the small regions and the rest of the national economy and suffi cient behavioural theory within the model.

1.2 A CGE Approach to Small-Region Representation

The TERM (The Enormous Regional Model) approach to sub-national regional modeling makes it possible to represent many small regions within a CGE database. In the past, modelers have been dissuaded from representing small regions in a CGE framework by an absence of input-output detail at the regional level. In TERM, by assuming that each industry has the same technology in all regions, there is no need for published regional input-output tables. Differences in technologies that appear in published regional tables, in cases where they exist, may arise from aggregation. For example, agriculture may vary widely between regions in terms of composition, resulting in differences in technologies across regions when represented as a single sector in regional tables. Rather than aggregating national sectors so as to compen-sate for defi ciencies in regional data, the data estimation approach used in TERM starts with many sectors, even more than appear in published national input-output tables.

The creation of small-region databases using the TERM approach has similarities to the IMPLAN approach. But TERM goes further: it includes both small-region representation and inter-regional trade matrices. Moreover, it includes the full theory

2 A Western Australian politician, Eric Ripper, summarised the crisis in a media statement (20 April 2010) downloadable from http://www.pdc.wa.gov.au/media/66775/20100420%20port%20hedland%20housing%20crisis.pdf 3 See http://www.dip.qld.gov.au/resources/guideline/project-assurance-framework/paf-cost-benefi t-analysis.pdf and http://www.dtf.wa.gov.au/cms/uploadedFiles/ecoresearchart2002.pdf

http://www.pdc.wa.gov.au/media/66775/20100420%20port%20hedland%20housing%20crisis.pdf

http://www.pdc.wa.gov.au/media/66775/20100420%20port%20hedland%20housing%20crisis.pdf

http://www.dip.qld.gov.au/resources/guideline/project-assurance-framework/paf-cost-benefit-analysis.pdf

http://www.dip.qld.gov.au/resources/guideline/project-assurance-framework/paf-cost-benefit-analysis.pdf

http://www.dtf.wa.gov.au/cms/uploadedFiles/ecoresearchart2002.pdf

31 Practical Policy Analysis Using TERM

of a CGE model. For example, it is possible to account for the costs to the rest of the economy of public spending in a small region and rigidities associated with short-run impacts.

TERM uses a ‘bottom-up’ approach to regional sub-national modeling. Demands, supplies, prices, and quantities are computed for each region separately in ‘bottom-up’ models. The theory of ‘bottom-up’ models at the regional level is much the same as that which applies at the national level in CGE models such as ORANI (Dixon et al. 1982 ) . This represents an advance on ‘top-down’ models, which compute region-specifi c quantities only for ‘local’ sectors and have no region-specifi c prices.

Chapter 2 , written by the creator of the TERM methodology, contains further details of the TERM approach. Perhaps, Sect. 2.5.1 (‘The false allure of regional input-output tables’) best captures the spirit of the TERM approach to database preparation. The approach used to devise not only sub-national but also small-region representation has opened up the possibility of using a CGE model to analyse rela-tively local issues. For example, a version of TERM showed the dramatic short-run impact of a mining boom on local housing prices (Wittwer and Horridge 2010 ) . Input-output analysis is always in danger of getting short-run local analysis wrong, as adjustments are a combination of price and quantity changes, rather than one or the other.

The applications of TERM in this book are to water issues. In the regional dimension, water use and water requirements vary widely. In irrigation regions, prolonged drought and government policy have had impacts on local economies that are possible to analyse using TERM, without compromising the theory of a national CGE model.

1.3 Modeling Water Scarcity Issues

The fi rst application of TERM was to the Australian drought of 2002–03 (Horridge et al. 2005 ) . Although this study included unprecedented bottom-up representation, it nevertheless used an aggregation of the master TERM database, which at the time included 144 sectors and 57 sectors. The drought study used a 38-sector, 45-region aggregation. The study served its purpose in showing that although agriculture’s share of GDP was little more than 3%, a drought of suffi cient magnitude could have a marked negative impact on national employment and real GDP.

Despite the contribution of the Horridge et al. ( 2005 ) study to regional and national analysis, there was scope to enhance future modeling of drought and water issues. In particular, the authors noted that reduced livestock herd numbers could impact adversely on future incomes in some regions. Therefore, full analysis of drought should extend beyond the short run and include dynamic analysis. Chapter 3 details the inclusion of dynamics in TERM, covering both the usual annual models and, in addition, quarterly models. The latter may be more relevant when modeling disruptions of a relatively short duration.

4 G. Wittwer

Another area for improvement beyond the fi rst application of TERM concerned the treatment of irrigation sectors. Without some theory to account for interaction between dryland and irrigated sectors, such a model will have defi ciencies. For example, the fi rst TERM study did not split farm sectors into dryland and irrigation sectors so that the drought was imposed as shocks to farm productivity rather than irrigation water availability.

Australian researchers have modeled water issues for more than 40 years. Chapter 4 discusses other models (mainly partial equilibrium) used to model water availability and farm productivity. These models started with a single small region. A recurring policy conclusion from early modeling efforts was that water was not allocated effi ciently. Early modelers recommended that the water authorities raise the price of irrigation water. Subsequent model developments led to multi-regional representation and have covered key issues including the impacts of water trading. With the separation of land and water titles, water trading becomes possible. With such trading, the market rather than authorities can determine the price of water.

The main limitation of partial equilibrium models is that they do not have a ready-made link between the irrigation activities and the rest of each regional economy. Many modelers make do with the limitations of partial equilibrium analy-sis and use results cautiously. Indeed, insights from such models have made impor-tant contributions to water policy debate in Australia and elsewhere. But in the absence of modeled interactions between the agriculture and the rest of a regional economy, some lobbyists assume without any empirical support that total impacts are manyfold greater than direct impacts, based on multipliers. The regional detail in a CGE database contains the share of agriculture in each region. This is a fi rst step in keeping the farm contribution in perspective. Typical CGE theory which includes resource constraints and imperfect factor movements between regions keep multipliers relatively small, particularly in the short run.

The theoretical modifi cations included in TERM-H2O, a version of TERM with water accounts, are detailed in Chap. 5 . Concerning the split between irrigation and dryland sectors, the theory of TERM-H2O starts with the premise that it is possible for an individual farm to produce a number of outputs and to switch between out-puts. For example, a rice farmer may switch to wheat during drought when water is scarce. A dairy farmer may move livestock from irrigated pasture to a dryland tech-nology with hand-feeding in drought. This illustrates the usefulness within a model of having irrigable land mobile between irrigated and dryland activities, depending on water availability. From water accounts compiled by the Australian Bureau of Statistics (see Table 7.1 ), there is ample evidence that farm factors are highly mobile between sectors in the short run. The modifi cations to TERM-H2O refl ect observed factor mobility. The farm factor mobility implicit in TERM-H2O is essentially a continuation of the Australian treatment of farm sectors. Powell and Gruen ( 1968 ) devised the CET (constant elasticity of transformation) specifi cation for farm sec-tors, in which farmers allocate given inputs to a combination of outputs in response based on relative output prices. This specifi cation was adapted for use in the ground-breaking ORANI model (Dixon et al. 1982 ) .


Regardless of the theoretical elaborations contained in a particular model, the justifi cation for a model comes from the insights it provides and its application to policy analysis. The most substantial contribution of this book to policy analysis is in the modeling of water buybacks for the environment (Chap. 6 ) and in modeling the regional economic impacts of drought (Chap. 7 ). Analysis in Chap. 6 shows that as water is taken out of production, its price rises. This leads to an increase in the asset value of water, which provides a windfall gain for holders of water rights. At the same time, there is a long-run decline in the value of land. This implies that farmers with relatively valuable holdings of water and less valuable land holdings will do better from buyback than farmers with less water and more valuable land. In this context, rice growers will do better than orchard and vineyard producers. In practice, we would not expect any farmer to be worse off due to buybacks in a process that is voluntary and entails full compensation. Orchard and vineyard pro-ducers, for example, may not be motivated to sell water unless they are able to make water savings over time or are seeking to exit the industry.

Community acceptance of water buybacks in the Murray-Darling Basin has grown, for example, as indicated by the volume of sales since the inception of the process. Modelers have had a role in preventing the process from being derailed by reminding policy advisers that job losses in the Murray-Darling Basin between 2005 and 2010 arose not from buybacks which started during drought but from the drought itself. The drought modeling detailed in Chap. 7 shows that job losses in the basin arising from drought were around 6,000 in the short term. Recalling the impact of the 2002–03 drought on livestock herd numbers, with a dynamic model, we are able to simulate the long-run consequence of drought-induced depressed farm investment. The modeling shows that a decade after the end of drought, basin-wide employment remains 1,500 below what it would have been without the earlier drought.

In communities with a signifi cant agricultural base, seasonal and international market conditions have a substantial impact on collective emotions. While water buybacks may have been blamed for job losses during the depths of drought, com-munity optimism returned with the rain. The dairy industry, despite being hit in some regions by extreme drought and fl oods in the space of a year or so, was upbeat about global demand and domestic supply conditions in 2011 (Gray 2011 ) . The Australian dollar driven by the mining boom is emerging as a much greater issue than water buybacks for irrigation farmers over the 2010s.

Another insight gained from modeling both dryland and irrigation agriculture is that irrigation water availability alone is not a good indicator of relative water scar-city. In Table 1.1 , we see that water availability in 2005–06 was not much greater than that in 2002–03, yet the average trading price was less than one-sixth that of 2002–03. There was severe drought throughout the basin in 2002–03. Since there had not been drought for a number of years, the impact of the drought on water allocations was smaller than in the case of repeated droughts, when reservoir levels continue to fall, thereby prompting cuts in water availability. The price of water in 2002–03 shot upwards as dryland productivity collapsed, and additional water was required on irrigated land to compensate for rainfall defi cits. 2005–06 was not a year of drought; rather, water allocations remained below usual levels due to a

6 G. Wittwer

limited recovery in reservoir storage levels in the wake of the 2002–03 drought. Had water been relatively scarce in 2005–06, the trading price would have been much higher. The extraordinarily high water price in 2007–08 arose from a combi-nation of low dam levels in the middle of a 3-year drought, reduced dryland produc-tivity due to the drought, and high commodity prices driven by the biofuel boom prior to the global fi nancial crisis.

1.4 Urban Water Issues

We can enhance the usefulness of TERM by modifying the theory and database to deal with specifi c issues. TERM-H2O was developed to deal mainly with water allocation issues in the Murray-Darling Basin. In cities, rapid population growth and prolonged drought have brought urban water policies to the forefront of govern-ment planning. Dynamic TERM (TERM-DYN) has been used in various studies (see Sect. 3.3 ), including dam construction (Chap. 8 ). TERM-DYN does not include the modifi cations of TERM-H2O that apply specifi cally to irrigation agriculture and farm factor mobility between irrigation and dryland sectors. TERM-DYN does not include water accounts, as these are most relevant when there is potential for the water price to vary widely between users and when water’s cost share in production is substantial.

Australia is often described as the driest inhabited continent on earth. This does not mean that Australia is the most water-scarce nation on earth. Rather, since Australia is sparsely populated, per capita water availability is not low by global standards. Water scarcity in Australia arises because much of the population is hud-dled in the south of the continent, far removed from the water resources of the north. A comparison between per capita water availability and per capita water use illustrates the point. FAO’s AQUASTAT records water availability as exceeding 20,000 kL per capita, while the Australian Bureau of Statistics records per capita water use in 2004–05 as 923 kL or 642 kL in 2008–09 when drought affected the south-eastern part of the nation (ABS 2010 ) .

At the regional level, water issues have economic signifi cance in urban as well as rural settings. A major urban water infrastructure project may require billions of dollars of expenditure, with prolonged construction and operational phases. The construction phase may boost regional employment, while the operational

Table 1.1 Irrigation water use and price in the Murray-Darling Basin

2002–03 2005–06 2007–08

Total agriculture (GL) 7,150 7,720 3,142 Weighted avg price a ($/mL) 364 57 562 MDB farm output price index (2005–06 = 100) 102 100 130

Source: ABS ( 2009a, b ) a Based on Watermove data for the Goulburn region


phase may bring lasting net benefi ts or lasting net costs, depending on the overall costs of the project and the marginal benefi ts it yields, as discussed in Chap. 8 .

Inevitably, given the large distances that separate much of Australia’s population from substantial water resources, occasional rallying cries emerge from commenta-tors and politicians to bring the water south. In 2005, the then Western Australian Liberal Party Leader and present Premier, Colin Barnett, fought and lost a state elec-tion on plans to channel water along a 3,700 km canal from the Kimberley region to Perth. Other politicians have lost votes on water issues. The fate of the former Victorian State Labor government, led by John Brumby, may have been sealed when the drought ended in 2009–10. The government had approved billions of dollars to build a desalination plant at Wonthaggi that was larger and even more expensive than plants approved in the other mainland states. Each of the plants was built in response to prolonged drought, and only one, that in Perth, became operational while the surrounding region was still in drought. In addition, Victoria built a politi-cally contentious pipeline that connected Melbourne to the water supply of the Murray-Darling Basin on the other side of the Great Dividing Range – opened at a time when both sides of the divide were facing water scarcities. With the return of the rain, voters may have perceived the Wonthaggi desalination plant as an expen-sive white elephant that would drive up Victorian water bills for years to come without any marginal benefi t.

In south-east Queensland, major capital works commenced towards the end of a drought to secure the region’s urban water supply for the future. The projects included pipelines to create a water grid across the region. Two new dams were proposed, Wyaralong and Traveston. The former was completed in 2011, while the latter was blocked by the Commonwealth Government in November 2009 and con-sequently abandoned. Earlier, a referendum held in Toowoomba voted against the use of recycled water in the city despite acute water shortages at the time. By late 2010, seasonal conditions had turned dramatically. Eastern Australia was feeling the effects of a strong La Niña weather event. In January 2011, when a major fl ood event damaged thousands of properties along the Brisbane River, public debate switched to appropriate fl ood mitigation rather than drought management.

From an economic perspective, the failure to price water according to scarcity inevitably leads to misallocation. Council of Australian Government (COAG) reforms in the 1990s did much to improve the allocation of water in the Murray-Darling Basin. Introduced late in the era of the Howard government, the 2007 Water Act sought to improve environmental outcomes through a combination of water infrastructure upgrades and water buybacks from farmers. But reforms in urban water allocation have been patchier. Each of the fi ve mainland state capitals has proceeded with engineering solutions to urban water scarcity at arguably excessive public expense. During the repeated droughts, there was no attempt in any city to adjust marginal prices upwards (and adjust fi xed changes downwards) as a tool of allocation. Equity concerns are the basis of the argument against marginal pricing. Yet, state governments appeared not to consider whether the billions of dollars poured into desalination plants would have provided more equitable outcomes used elsewhere. Moreover, fi xed and variable water prices are rising to cover the costs of

8 G. Wittwer

desalination. Water managers are quick to discourage residents from installing water tanks on the basis of their monetary return, at the same time as shying away from adjusting marginal prices to refl ect scarcity.

1.5 Beyond Australia

The fi nal chapter explores the possibility of devising versions of TERM to deal with water issues in other countries. Our experience in Australia is that collecting data to use in a multi-regional model is a process of gradual improvement. For example, in the fi rst version of TERM, crude estimates of exports and imports by port were devised from the annual reports of port authorities. We subsequently discovered that the Australian Bureau of Statistics (ABS) collected such data. Concerning water statistics specifi cally, the data published by the ABS have improved gradually over time. This has been important in representing the regions of the Murray-Darling Basin at the catchment level.

The philosophy of TERM (Chap. 2 ) applies in other countries as much as in Australia. That is, one should never wait until one is satisfi ed with the quality of readily available data before running a model. That day may never come. It is better to accept that data are limited, and record (in a database-creating computer pro-gram) the assumptions used to make do with such data. Often, the process of run-ning a multi-regional CGE model leads to improvements in the database. The onus is on the practitioner to keep a record of all assumptions used in devising a database. If better data become available, they may supersede some of the assumptions used in earlier database programs.

Water issues are a major part of policy in many nations. Cities around the world need to grapple with turning to recycling, storm-water catchment or desalination, or building a new dam as they respond to increasing urban demands for water. The competition between urban and rural uses for water is pervasive. If the focus of a CGE-based project is to examine such competition, the modeler needs to decide on appropriate regional representation. Next, water trading rules need to be devised within the model. In the Australian context, water is tradable between catchment regions in the southern part of the Murray-Darling Basin, but geographically impos-sible in the north of the basin. In other countries, the legislation required to imple-ment water trading may be absent.

Section 9.3 discusses the possibility of applying a version of TERM-H2O to China. SinoTERM developed by Horridge and Wittwer ( 2008 ) includes representa-tion of the economies of 31 regions, with agriculture represented in much greater detail than in the input-output table published by the National Bureau of Statistics in Beijing. An appropriate regional representation using a version of TERM-H2O in China would concentrate on the most water-stressed part of the country. China has a diverse geography and disparity in relative water scarcity between the water-abundant south and water-stressed regions such as Hebei and Beijing. It would appear appropriate to start with a model that includes regions of Hebei at the prefecture level (but separating cities from irrigation regions), plus Beijing and Tianjin (each


driven by burgeoning urban water demands) and the rest of China. By going to a sub-provincial level, we are already stretching available data on industry production shares, although data tend to be better for agriculture at this level. 4 It would be nec-essary to glean as much information as possible from national and provincial data sources. Yet, within the TERM approach, research resources rather than data are the main constraints on devising such a model. Our experience is that with repeated efforts in compiling data, running a CGE model, and interpreting results, the quality of regional data improves.

Already, versions of TERM have been developed for Brazil, China, Finland, Indonesia, South Africa, and the United States. An important part of model develop-ment has come through the contribution of collaborators from each of these nations. Not only do such collaborators have more ready access to data than the Australians who developed the fi rst version of the model. Collaborators from within the country of interest usually have a keener sense of major regional policy issues than outsiders. Moreover, they are more likely to build and maintain professional relationships with other policy analysts and policy makers within their own country. This strikes at the heart of our model development efforts. We use multi-regional CGE models as tools of analysis of relevant policy questions, rather than to explore theoretical curiosities.

References

ABS (Australian Bureau of Statistics) (2009a) Water use on Australian farms, 2007–08 (and earlier issues). Catalogue 4618.0. Australian Bureau of Statistics, Canberra

ABS (Australian Bureau of Statistics) (2009b) Value of agricultural commodities produced, 2007–08 (and earlier issues). Catalogue 7503.0. Australian Bureau of Statistics, Canberra

ABS (Australian Bureau of Statistics) (2010) Water Account Australia, 2008–09. Catalogue 4618.0. Australian Bureau of Statistics, Canberra

Dixon P, Parmenter B, Sutton J, Vincent D (1982) ORANI: a multisectoral model of the Australian Economy. Contributions to economic analysis 142, North-Holland, Amsterdam

Gray M (2011) Dairy farmers relishing best conditions in years. The Age, May 11. http://www.theage.com.au/victoria/dairy-farmers-relishing-best-conditions-in-years-20110510-1eh64.html. Accessed 12 May 2011

Horridge M, Madden J, Wittwer G (2005) Using a highly disaggregated multi-regional single-country model to analyse the impacts of the 2002–03 drought on Australia. J Policy Model 27:285–308

Horridge M, Wittwer G (2008) SinoTERM, a multi-regional CGE model of China. China Econ Rev 19:628–634

Okuda T, Hatano T, Qi S (2004) An estimation of a multi-regional input/output table in China and the analysis. Paper presented at the EcoMod IO and general equilibrium data, modelling, and policy analysis conference, Brussels, 2–4 Sept

Powell A, Gruen F (1968) The constant elasticity of transformation Frontier and linear supply system. Int Econ Rev 9:315–328

Wittwer G, Horridge M (2010) Bringing regional detail to a CGE model using census data. Spat Econ Anal 5:229–255

4 Moreover, despite legal reforms concerning ownership of property in China over the past decade, the state still owns all land. It may be premature to model water in China as though market mecha-nisms can play a role in its allocation.

Part I The TERM Approach


Abstract TERM (The Enormous Regional Model) provides a strategy for creating a ‘bottom-up’ multi-regional CGE model which treats each region of a single country as a separate economy. This makes it a useful tool for examining the regional impacts of shocks that may be region specifi c. TERM is designed to allow quick simulations with many regions, so allowing for models of large countries with 30–50 provinces, such as USA or China. TERM also offers a standard procedure for preparing a database which requires, in addition to a national input-output or use-supply table, a minimal amount of regional data. More regional data can be used if available.

Keywords CGE modeling • Database development • Demand sourcing • Gravity assumption • Input-output data • Sub-national data

2.1 Introduction

TERM is a framework for CGE (computable general equilibrium) modeling of multiple regions within a single country. It was developed to address two common problems of multi-regional CGE models:

As the number of regions increases, simulations become very slow, or require • large amounts of memory. It is diffi cult to develop a database for such models; published data are usually • quite sparse.

Chapter 2 The TERM Model and Its Database*

Mark Horridge

M. Horridge (*) Centre of Policy Studies , Monash University , Wellington Rd, Clayton campus , Clayton , VIC 3800 , Australia e-mail: [email protected]

* Portions of this chapter draw on Horridge et al. (2005).

14 M. Horridge

TERM offers a solution to both problems:

The database and equation system are structured to allow fast solutions with • small memory needs. An inbuilt automatic system to aggregate regions and/or sectors allows model size to be reduced to speed simulations, while preserving detail that is needed for a particular application. There is a standard procedure for preparing a database which requires, in addi-• tion to a national input-output or use-supply table, a minimal amount of regional data. More regional data can be used if available.

From the outset, the TERM framework has been intended as a template which might be quickly applied to a variety of countries. Thus, the standard version of TERM is fairly simple, avoiding mechanisms which might be specifi c to a particular country or application. Rather the emphasis is on allowing a basic multi-regional model to produce simulation results as soon as possible. Very often, analysis of results reveals shortcomings of the model or data, or suggests priorities for improve-ment. To arrive quickly at this stage is key to the quality of the fi nal model.

TERM builds on the ORANI model (Dixon et al. 1982 ) , which distinguished over 100 sectors, and introduced large-scale computable general equilibrium mod-eling in 1977. In particular, the minimal data requirements for constructing a TERM database scarcely exceed those for a ‘top-down’ multi-regional version of ORANI, described below. In fact, the standard procedure for preparing a TERM assumes that a working ‘top-down’ database has already been prepared and used for simulations. This allows most potential problems with regional data to be noticed and fi xed at an early stage.

2.2 Progress in Australian Regional Economic Modeling

Since ORANI, related models have developed in several new directions. ORANI’s solution algorithm combined the effi ciency of linearised algebra with the accuracy of multi-step solutions, allowing the development of ever more disaggregated and elaborate models. The GEMPACK software developed by Ken Pearson ( 1988 ) and colleagues since the mid-1980s simplifi ed the specifi cation of new models, while cheaper, and more powerful computers allowed the development of computer-intensive multi-regional and dynamic models.

On the demand side, these advances have been driven by the appetite of policy-makers for sectoral, regional, temporal, and social detail in analyses of the effects of policy or external shocks. Since parliamentary representatives are elected by regions, demand for regional detail is particularly strong.

To meet this need, even early versions of ORANI (see Dixon et al. 1978 ) included a ‘top-down’ regional module to work out the regional consequences of national economic changes: national results for quantity (but not price) variables were broken down by region using techniques borrowed from input-output analysis.

152 The TERM Model and Its Database

The name ‘top-down’ refl ects the feature that national results drive regional results and are unaffected by the regional subsystem. Key assumptions are:

For each sector, the technology of production (i.e., cost shares) is uniform across • regions. For commodities that are heavily traded between regions (the ‘national’ com-• modities), each region’s share of national output is fi xed or exogenous. So for these sectors, the percent change in output is uniform across regions. For the remaining, ‘local’, commodities (that are little traded between regions), • output in each region adjusts to meet demand in that region.

Using the top-down technique, from 8 to 100 regions can easily be distinguished. Region-specifi c demand shocks may be simulated, but since price variables have no regional dimension, there is little scope for region-specifi c supply shocks. 1 On the other hand, the ‘top-down’ approach requires little extra data or computer power.

A second generation of regional CGE models adapted ORANI by adding two regional subscripts (source and destination) to many variables and equations. In this ‘bottom-up’ type of multi-regional CGE model, national results are driven by (i.e., are additions of) regional results. Liew ( 1984 ) , Madden ( 1990 ) , and Peter et al. ( 1996 ) describe several Australian examples. Dynamic versions of such models have followed (Giesecke 1997 ) . The best-known example of this type of regional model is the Monash Multi-Regional Forecasting model, MMRF (Adams et al. 2002 ) .

Bottom-up models allow simulations of policies that have region-specifi c price effects, such as a payroll tax increase in one region only. They also allow us to model imperfect factor mobility (between regions as well as sectors). Thus, increased labour demand in one region may be both choked off by a local wage rise and accommodated by migration from other regions. Unfortunately, models like MMRF pose formidable data and computational problems—limiting the amount of sectoral and regional detail. Only two to eight regions and up to 40 sectors could be distin-guished. 2 Luckily, Australia has only eight states, but size limitations have hindered the application of similar models to larger countries with 30–50 provinces and have hitherto prevented us from distinguishing smaller, sub-state regions.

Finer regional divisions are desirable for several reasons. Policymakers who are concerned about areas of high unemployment or about disparities between urban and rural areas desire more detailed regional results. Environmental issues, such as water management, often call for smaller regions that can map watershed or other natural boundaries more closely. Finally, more and smaller regions give CGE models a greater sense of geographical realism, closing the gap between CGE and LUTE (Land Use Transportation Energy) modeling.

1 Such limitations could be partially circumvented: see Higgs et al. ( 1988 ) . 2 More precisely, these second-generation models (like MMRF) become rather large and slow to solve as the product: (number of regions) × (number of sectors) exceeds 300. TERM raises this limit to about 2,500.

16 M. Horridge

The TERM 3 model adds to the ORANI/MMRF tradition by allowing greater disaggregation of regional economies than was previously available. For example, it allows us to analyse effects for each of 57 statistical divisions within Australia—which would be computationally infeasible using the MMRF framework.

2.3 The Structure of TERM

A key feature of TERM, in comparison to predecessors such as MMRF, is its ability to handle a greater number of regions or sectors. The greater effi ciency arises from a more compact data structure, made possible by a number of simplifying assumptions.

2.3.1 Defeating the Curse of Dimensionality

The database for a CGE model consists of matrices of fl ow values dimensioned by commodity, industry, and region. The model will contain quantity and price vari-ables for each of these fl ows, so the number of variables and equations tends to track database size. The computer resources (time and memory) needed to solve the model increase super-proportionately 4 as the size of the database increases. Indeed, a doubling of database size may multiply solution time by three. Sectoral or regional detail may have to be sacrifi ced to reduce computing problems.

To illustrate, the value of intermediate demands in a single-region CGE model (like ORANI) might be represented by a matrix V , with dimensions COM*IND, where, for example:

V (‘Coal’,‘Steel’) = value of Coal used by the Steel industry. With 50 commodities and industries, V would contain 2,500 elements.

In the MMRF framework, V would be dimensioned COM*IND*REG*REG, where the fi rst regional subscript denotes the region of origin of some input and the second regional subscript denotes the region where the input is used. Since MMRF distinguishes eight Australian states, the V matrix would be 64 times bigger than in ORANI—leading to much larger (but just acceptable) solution times. A USA ver-sion of MMRF, distinguishing 50 states, would imply a database which was 39 [=(50/8) 2 ] times larger than the eight-state MMRF, leading to model solution times

3 TERM is an acronym for “The Enormous Regional Model.” 4 A key stage in the model solution process is the solution of an N * N linear equation system, where N is the number of endogenous variables. Using conventional techniques, we can expect that the time for this step will follow the cube of N . GEMPACK’s sparse matrix and automatic substitution techniques reduce this penalty substantially; we assume below that solution time and space require-ments follow N 1.5 .


and memory requirements perhaps 500 times those of the Australian MMRF—which is quite impractical. TERM’s solution to this problem is to restructure the model database so that no matrix contains more than three of the ‘large’ COM, IND, or REG dimensions. For example, instead of the large four-dimensional intermediate input matrix used by MMRF: V(COM, IND, REG, REG), we could instead use two 3-dimensional matrices:

V(COM, IND, REG) = value of commodities used by industries in region of use, and

T(COM, REG, REG) = value of commodities used, by regions of production and use,

which together are 25 times smaller (with 50 sectors and regions), leading to model solution times and memory requirements perhaps 125 times less. The cost is a small loss in generality: the sourcing or trade matrix T encapsulates the assump-tion that all users in a particular region of, say, vegetables, source their vegetables from other regions according to common proportions.

2.3.2 The TERM Data Structure

Figure 2.1 is a schematic representation of the model’s input-output database. It reveals the basic structure of the model, which is key to its effi ciency. The rectangles indicate matrices of fl ows. Core matrices (those stored on the database) are shown in bold type; the other matrices may be calculated from the core matrices. The dimensions of the matrices are indicated by indices (c, s, i, m, etc.) which correspond to the sets of (Table 2.1 ); there, the sets DST, ORG, and PRD are in fact the same set, named according to the context of use.

The matrices in Fig. 2.1 show the value of fl ows valued according to three methods:

1. Basic values = Output prices (for domestically produced goods), or CIF prices (for imports).

2. Delivered values = Basic + Margins. 3. Purchasers’ values = Basic + Margins + Tax = Delivered + Tax.

The matrices on the left-hand side of the diagram resemble (for each region) a conventional single-region input-output database. For example, the matrix USE at top left shows the delivered value of demand for each good (c in COM) whether domestic or imported (s in SRC) in each destination region (DST) for each user (USER, comprising the industries; IND; and four fi nal demanders: households, investment, government, and exports). Some typical elements of USE might show:

USE(‘Wool’, ‘dom’, ‘Textiles’, ‘North’): domestically produced wool used by • the textile industry in North. USE(‘Food’, ‘imp’, ‘HOU’, ‘West’): imported food used by households in West. •

18 M. Horridge

Fig. 2.1 The TERM fl ow database


USE(‘Meat’, ‘dom’, ‘EXP’, ‘North’): domestically produced meat exported from • a port in North. Some of this meat may have been produced in another region. USE(‘Meat’, ‘imp’, ‘EXP’, ‘North’): imported meat re-exported from a port in • North.

As the last example shows, the data structure allows for re-exports (at least in principle). All these USE values are ‘delivered’: they include the value of any trade or transport margins used to bring goods to the user. Notice also that the USE matrix contains no information about regional sourcing of goods.

The TAX matrix of commodity tax revenues contains an element corresponding to each element of USE. Together with matrices of primary factor costs and produc-tion taxes, these add to the costs of production (or value of output) of each regional industry.

In principle, each industry is capable of producing any good. The MAKE matrix at the bottom of Fig. 2.1 shows the value of output of each commodity by each industry in each region. A subtotal of MAKE, MAKE_I, shows the total production of each good (c in COM) in each region d.

TERM recognises inventory changes in a limited way. First, changes in stocks of imports are ignored. For domestic output, stock changes are regarded as one desti-nation for industry output (i.e., they are dimension IND rather than COM). The rest of production goes to the MAKE matrix.

The right-hand side of Fig. 2.1 shows the regional sourcing mechanism. The key matrix is TRADE, which shows the value of inter-regional trade by sources (r in ORG) and destinations (d in DST) for each good (c in COM) whether domestic or imported (s in SRC). The diagonal of this matrix ( r = d ) shows the value of local usage which is sourced locally. For foreign goods ( s = ‘imp’), the regional source subscript r (in ORG) denotes the port of entry. The matrix IMPORT, showing total entry of imports at each port, is simply an add up (over d in DST) of the imported part of TRADE.

The TRADMAR matrix shows, for each cell of the TRADE matrix, the value of margin good m (m in MAR) which is required to facilitate that fl ow. Adding together the TRADE and TRADMAR matrix gives DELIVRD, the delivered (basic + margins)

Table 2.1 Main sets of the TERM model

Index Set name Description Typical size

s SRC (dom,imp) Domestic or imported (ROW) sources 2 c COM Commodities 40 m MAR Margin commodities (trade, road, rail, boat) 4 i IND Industries 40 o OCC Skills 8 d DST Regions of use (destination) 30 r ORG Regions of origin 30 p PRD Regions of margin production 30 f FINDEM Final demanders (HOU, INV, GOV, EXP) 4 u USER Users = IND union FINDEM 44

20 M. Horridge

value of all fl ows of goods within and between regions. Note that TRADMAR makes no assumption about where a margin fl ow is produced (the r subscript refers to the source of the underlying basic fl ow).

Matrix SUPPMAR shows where margins are produced (p in PRD). It lacks the good-specifi c subscripts c (COM) and s (SRC), indicating that, for all usage of margin good m used to transport any goods from region r to region d, the same proportion of m is produced in region p. Summation of SUPPMAR over the p (in PRD) sub-script yields the matrix SUPPMAR_P which should be identical to the subtotal of TRADMAR (over c in COM and S in SRC), TRADMAR_CS. In the model, TRADMAR_CS is a CES aggregation of SUPPMAR: margins (for a given good and route) are sourced according to the price of that margin in the various regions (p in PRD).

TERM assumes that all users of a given good (c,s) in a given region (d) have the same sourcing (r) mix. In effect, for each good (c,s) and region of use (d), there is a broker who decides for all users in d whence supplies will be obtained. Armington sourcing is assumed: the matrix DELIVRD_R is a CES composite (over r in ORG) of the DELIVRD matrix.

A balancing requirement of the TERM database is that the sum over user of USE, USE_U, shall be equal to the sum over regional sources of the DELIVRD matrix, DELIVRD_R.

It remains to reconcile demand and supply for domestically produced goods. In Fig. 2.1 , the connection is made by arrows linking the MAKE_I matrix with the TRADE and SUPPMAR matrices. For non-margin goods, the domestic part of the TRADE matrix must sum (over d in DST) to the corresponding element in the MAKE_I matrix of commodity supplies. For margin goods, we must take into account both the margin requirement SUPPMAR_RD and direct demands TRADE_D.

At the moment, TERM distinguishes only four fi nal demanders in each region:

(a) HOU: the representative household (b) INV: capital formation (c) GOV: government demand (d) EXP: export demand

For many purposes, it is useful to break down investment according to destination industry. The satellite matrix INVEST (subscripted c in COM, i in IND, and d in DST) serves this purpose. It allows us to distinguish the commodity composition of investment according to industry: for example, we would expect investment in agri-culture to use more machinery (and less construction) than investment in dwellings.

Similarly, another satellite matrix, HOUPUR, allows us to distinguish several household types with different budget shares. Both satellite matrices enforce the assumption that import/domestic shares and commodity tax rates are uniform across household (or investor) types: For example, we assume that the tax rate on cigarettes is the same for rich and poor, as is the share of imports in cigarette consumption.

Missing from Fig. 2.1 is an account of how factor incomes and tax revenue accrue to regional households and governments. Such data would be needed to convert the TERM data scheme into a complete SAM. Australian versions of TERM typically


assume that wage income generated in region A accrues to households in region A, while capital income goes into a national ‘pot’ which is shared between regional households. Similarly, tax revenue accrues to a national authority which distributes it between regions. Such assumptions might be inappropriate if TERM were applied to other countries. For example, in the USA, some taxes accrue directly to state governments, and wage income generated in Washington, DC may well be spent by households in Maryland or Virginia. Hence, the generic version of TERM enforces no default system: users must devise their own mapping of income to agents, appropriate to a particular country. Their decisions may be infl uenced by the chosen level of regional detail. For example, some Australian versions of TERM distinguish 57 ‘statistical division’ regions which do not entirely corre-spond with administrative regions—so regional government incomes are not modeled in these versions.

2.4 The TERM Equation System

The equations of the TERM model are broadly similar to those of other CGE models. Producers choose a cost-minimising combination of intermediate and primary factor inputs, subject to production functions which are structured by a series of CES ‘nesting’ assumptions, illustrated in Fig. 2.2 . Two high-level aggregates, of primary factors and of intermediate inputs, are each demanded in proportion to industry output (Leontief assumption). The primary factor aggregate is a CES composite of capital, land, and a labour aggregate—the latter being itself a CES composite of labour by skill group. The aggregate intermediate input is again a CES composite of dif-ferent composite commodities, which are in turn CES composites of commodities from different sources—as described in detail in the next section. Industry outputs are transformed into commodity outputs via a CET mechanism that is calibrated from the MAKE matrix of Fig. 2.1 .

2.4.1 TERM Sourcing Mechanisms

Figure 2.3 illustrates the details of the TERM system of demand sourcing. Although the fi gure covers only the demand for a single commodity (vegetables) by a single user (households) in a single region (North), the same diagram would apply to other commodities, users, and regions. The diagram depicts a series of ‘nests’ indicating the various substitution possibilities allowed by the model. Down the left side of the fi gure, boxes with dotted borders show in upper case the value fl ows associated with each level of the nesting system. These value fl ows may also be located in Fig. 2.1 . The same boxes show in lower case the price (p....) and quantity (x....) variables associated with each fl ow. The dimensions of these variables are critical both to the usefulness of the model and to its computational tractability; they are indicated by

22 M. Horridge

subscripts c, s, m, r, d, and p, as explained in Table 2.1 . Most of what is innovative in TERM could be reconstructed from Figs. 2.1 and 2.2 .

At the top level, households choose between imported (from another country) and domestic vegetables. A CES or Armington specifi cation describes their choice—as pioneered by ORANI and adopted by most later CGE models. Demands are guided by user-specifi c purchasers’ prices (the purchasers’ values matrix PUR is found by summing the TAX and USE matrices of Fig. 2.1 ). Two is a typical value for the elasticity of substitution.

Demands for domestic vegetables in a region are summed (over users) to give total value USE_U (the ‘_U’ suffi x indicates summation over the user index u).

KEY

Inputs orOutputs

FunctionalForm

CES

CES

Leontief

CESCES

Labourtype O

Labourtype 2

Labourtype 1

CapitalLabourLand

PrimaryFactors

ImportedGood G

DomesticGood G

ImportedGood 1

DomesticGood 1

Good GGood 1

CES 0.50.15

1.5

0.35

Intermediate

CET

up to Good GGood 2Good 1

ActivityLevel

0.5 STOCKS

up to

Fig. 2.2 TERM production structure


North Middle South

Vegetables

Trade Road Rail

North Middle South

Vegetables toHouseholds in

North

DomesticVegetables

ImportedVegetables

PUR_S(c,u,d)ppur_s(c,u,d)xhou_s(c,d)

CES

CES

Leontief

CES

DELIVRD(c,s,r,d)pdelivrdr(c,s,r,d)

xtrad(c,s,r,d)

PUR(c,s,u,d)ppur(c,s,u,d)xhou(c,s,d)

TRADE(c,s,r,d)pbasic(c,s,r)xtrad(c,s,r,d)

TRADMAR(c,s,m,r,d)psuppmar_p(m,r,d)xtradmar(c,s,m,r,d)

SUPPMAR(m,r,d,p)pdom(m,p)

xsuppmar(m,r,d,p)

add over sourceand commodities

Region where road margin is produced

Origin of vegetables

DomesticVegetables

add overusers

USE_U(c,s,d)pdelivrd_r(c,s,d)

xtrad_r(c,s,d)

user-specificpurchasers' values

origin-specific

deliveredprices

not user-specificdelivered values

c = "Vegetables"u = "Hou"d = "North"

RoadTRADMAR_CS(m,r,d)

psuppmar_p(m,r,d)xtradmar_cs(m,r,d)

σ =0.1 - 0.5

σ=2

σ = 0.2 - 5

Fig. 2.3 TERM sourcing mechanisms

24 M. Horridge

The USE_U matrix is measured in ‘delivered’ values—which include basic values and margins (trade and transport), but not the user-specifi c commodity taxes.

Moving down, the next level treats the sourcing of USE_U between the various domestic regions. The matrix DELIVRD shows how USE_U is split between origin regions r. Again, a CES specifi cation controls the allocation; substitution elasticities range from 5 (merchandise) to 0.2 (services). The CES implies that regions which lower production costs more than other regions will tend to increase their market share. The sourcing decision is made on the basis of delivered prices—which include transport and other margin costs. Hence, even with growers’ prices fi xed, changes in transport costs will affect regional market shares. Notice that variables at this level lack a user (u) subscript—the decision is made on an all-user basis (as if wholesalers, not fi nal users, decided where to source vegetables). The implication is that, in North, the proportion of vegetables which come from South is the same for households, intermediate, and all other users.

The next level down shows how a ‘delivered’ vegetable from, say, South is a Leontief composite of basic vegetable and the various margin goods. The share of each margin in the delivered price is specifi c to a particular combination of origin, destination, commodity, and source. For example, we should expect transport costs to form a larger share for region pairs which are far apart, or for heavy or bulky goods. The number of margin goods will depend on how aggregated is the model database. Under the Leontief specifi cation, we preclude substitution between road and retail margins, as well as between road and rail. For some purposes, it might be worthwhile to construct a more elaborate nesting which accommodated road/rail switching.

The bottom part of the nesting structure shows that margins on vegetables pass-ing from South to North could be produced in different regions. The fi gure shows the sourcing mechanism for the road margin. We might expect this to be drawn more or less equally from the origin (South), the destination (North), and regions between (Middle). There would be some scope ( s = 0.5) for substitution since truck-ing fi rms can relocate depots to cheaper regions. For retail margins, on the other hand, a larger share would be drawn from the destination region, and scope for sub-stitution would be less ( s = 0.1). Once again, this substitution decision takes place at an aggregated level. The assumption is that the share of, say, Middle, in providing road margins on trips from South to North, is the same whatever good is being transported.

Although not shown in Fig. 2.3 , a parallel system of sourcing is also modeled for imported vegetables, tracing them back to port of entry instead of region of production.

2.4.2 Other Features of TERM

The remaining features of TERM are common to most CGE models, and in particular to ORANI, from which TERM descends. Industry production functions are of the


nested CES type: Leontief except for substitution between primary factors and between sources of goods. Exports from each region’s port to the ROW face a con-stant elasticity of demand. The composition of household demand follows the linear expenditure system, while the composition of investment and government demands is exogenous. A variety of closures are possible. For the short-run simulation, we might hold fi xed industry capital stocks and land endowments, whilst allowing labour to be fully mobile between sectors within a region and partially mobile between regions. At the regional level, we may link household consumption to regional factor incomes.

2.4.2.1 National and Regional Macro Closures

Closure fl exibility in TERM applies separately at the national and regional levels. For example, we may wish to impose a balance of trade constraint at the national level, without, however, enforcing balanced trade for each region. We might stipu-late that regional consumption C

r follows wage income W

r via a rule like:

λ= ,r r rC F W

where F r is a regional propensity to consume and l is a slack variable which adjusts

to satisfy the national balance of trade constraint. Similarly, we might relate government spending in each region to that region’s

GDP, while holding fi xed national government spending.

2.4.3 Comparison with the GTAP Model

GTAP, a well-known model of the world economy, has a fairly similar structure to TERM. The ‘regions’ of GTAP, however, are countries or groups of countries, while in TERM, they are regions within a single country. In GTAP, regional trade defi cits must sum to zero (the planet is a closed system), while in TERM, a national trade defi cit is possible. There are also differences in data structures: GTAP has a far more detailed representation of bilateral trade taxes than does TERM, refl ecting the freer trade that is usually possible within a nation. TERM can accommodate commodity tax rates that vary between regions (North might tax whisky more than South), but it does not allow for regional tax discrimination (such as a tax, in North, that applied only to whisky from West). Inter-regional labour movements, a rarity in GTAP, are usual in TERM. Finally, TERM has a more detailed treatment of transport margins. While GTAP identifi es how much each country contributes to world shipping sup-ply, the TERM data structure shows how much each region contributes to supply of transport between all separate pairs of source and destination regions (the matrix SUPPMAR in Fig. 2.1 ).

26 M. Horridge

2.5 Gathering Data for 144 Sectors and 57 Regions

As formidable as the computational demands of regional CGE models are the data requirements—which usually far exceed what is available.

2.5.1 The False Allure of Regional Input-Output Tables

Newcomers to regional CGE often assume that the natural starting point is a pub-lished set of regional input-output tables. However, even if such tables are available, they may suffer from serious defi ciencies:

They typically distinguish far too few sectors to support serious CGE modeling. • The regional coverage may be too coarse, or incomplete (China without Tibet) or • inconsistent (regional tables for different dates, or with different formats). They are typically not designed for use by CGE modelers. •

Enlarging on the last point, regional IO modelers produce tables for their own purposes, which are different to those of CGE modelers. A central purpose of much regional IO analysis is to estimate the ‘multiplier’: the amount by which employment in some region will increase if autonomous demand (e.g., a construction project) increases by, say, a million dollars. For this purpose, it is suffi cient if the regional table shows, in a single row, the value of imports (from other regions or the rest of the world) used by each sector and fi nal demander. Indeed, regional IO tables are often presented in this form. Such tables tell us, for example, how much locally produced gasoline is used by the transport sector, but not how much gasoline (from any source) is used. They cannot therefore support the varied range of applications for which CGE models are designed. For example, we could not easily estimate by how much an increased gasoline tax would increase transport costs.

A regional IO specialist needs some criterion to choose between alternative tech-niques of constructing regional IO tables. The criterion will refl ect the planned use of the tables. For example, Flegg et al. ( 1995 ) and Flegg and Tohmo ( 2008 ) refer to the effect (of different methods) on estimated multipliers as a way to compare tech-niques. For a CGE model, other criteria are important, and so different techniques of generating regional tables may be preferred.

2.5.2 The TERM Data Strategy

By contrast, TERM offers a strategy, depicted in Fig. 2.6 , to estimate its database from very limited regional data. We describe below some key features of this strategy as applied to the Australian TERM model with base year of 1997. 5

5 A more detailed description is given in Wittwer and Horridge ( 2010 ) . However, that paper describes a later edition of the Australian TERM database, which distinguishes more sectors


(a) The process starts with a national input-output table and certain regional data. The minimum requirements for regional data are very modest: the distribution between regions of industry outputs and of fi nal demand aggregates. This dis-tribution can be conceived as a set of regional shares, which may in turn be based on value data, or on physical units (e.g., tonnes of wheat) or on numbers employed. This fl exibility (regarding units) greatly increases the amount of data which may be used. Additional regional detail, such as region-specifi c technologies or consumption preferences may be added selectively, when available.

(b) The process is automated so that additional detail can easily be added at a later stage.

(c) The database is constructed at the highest possible level of detail: 144 sectors and 57 regions. Aggregation (for computational tractability) takes place at the end of the process, not at the beginning. Perhaps surprisingly, the high level of disaggregation is often helpful in estimating missing data. When aggregated, the model database displays a richness of structure that belies the simple mechanical rules that were used to construct its disaggregated parent. For example, even though we normally assume that a given disaggregated sector has the same input-output coeffi cients wherever it is located, aggregated sectors display regional differences in technology. Thus, sectoral detail partly compen-sates for missing regional data. 6

Our technique of combining a national IO table with limited regional data to produce a detailed inter-regional table bears many similarities to methods devel-oped over several decades by regional IO modelers. Indeed, published regional input-output tables may well be in part constructed rather than observed. Unfortunately, the method of construction may be poorly documented or unrepeat-able. The TERM data programs are downloadable and may easily be customised to suit particular needs. They will appeal to the modeler who would prefer to construct a multi-regional database using known assumptions, rather than rely on data con-structed somehow by others.

Of course, published regional input-output tables may well form part of the inputs to the TERM data process. But they should certainly not constrain the degree of regional or sectoral detail that we aim for.

2.5.3 The National Input-Output Database

As shown in Fig. 2.6 , the TERM data process starts from the 1997 Australian input-output tables, distinguishing 107 sectors. Our fi rst step was to convert these tables

(172 rather than 144) and many more regions (206 rather than 57). The greater regional detail relies on census data that show employment by sector and small region. The later edition also benefi ts from more detailed (four-digit) merchandise trade data for 60 ports. 6 This point is enlarged below in Sect. 2.5.4.1 Region-Specifi c Technology and Output Mix.

28 M. Horridge

to the fi le format of ORANI-G, a standard single-country CGE model. Next, working at the national level, we expanded the 107 sectors to 144. In choosing to split sectors, we hoped to avoid infelicities of classifi cation that have caused problems in the past (such as the lumping together of exports of sugar, cotton, and prawns) and also to split up sectors which showed regional differences in input mix or sales pattern. For example, we split up electricity generation according to the fuel used (which differs among Australian regions) and added considerable agricultural detail. The interests of one collaborator led to a remarkably detailed treatment of the wine and grape sectors, which were divided according to quality (some regions produce high-quality wine for export, others a cheaper brew for local drinking).

The main source for the sectoral split was unpublished ABS commodity cards data. Such data provide a split of sales for approximately 1,000 commodities to 107 industries, plus fi nal users. 7 However, the cards data do not always provide a desir-able split from the 107 industries to the eventual 144 sectors of the disaggregated database. For example, there are signifi cant sales of sugarcane to the other food products sector (107-sector aggregation). We allocated all sugarcane sales to refi ned sugar and zero sales to the seafood and other food products in our 144-sector disag-gregation. When the intermediate sales split was less obvious, we used activity weights of the purchasing sectors for the split.

The 144-sector national database has an independent value for our modeling work (e.g., it forms the bulk of the MONASH database). For TERM purposes, it was converted to a simpler format prior to the addition of regional detail.

2.5.4 Estimates of the Regional Distribution of Output and Final Demands

The next step was to obtain, for each industry and fi nal demander, an estimate of each statistical division’s share of national activity (these shares are the R001, R002, etc., of Fig. 2.6 ). To develop a full input-output table for each region, we required estimates of industry shares (i.e., each region’s share of national activity for a given industry), industry investment shares, household expenditure shares, international export and import shares, and government consumption shares.

The main data sources for the industry split were:

AgStats data from ABS, which details agricultural quantities and values at the • SD level Employment data by industry at the SD level prepared by our colleague Tony • Meagher from ABS census data and surveys Published ABS manufacturing census data (state level) • State yearbooks (for mining, ABS 1301, and for grapes and wine, ABS 1329.0) •

7 Commodity sales details are now downloadable from http://abs.gov.au/AUSSTATS/[email protected]/DetailsPage/5215.0.55.0012005-06?OpenDocument (accessed 19 October 2011)

http://abs.gov.au/AUSSTATS/[email protected]/DetailsPage/5215.0.55.0012005-06?OpenDocument

http://abs.gov.au/AUSSTATS/[email protected]/DetailsPage/5215.0.55.0012005-06?OpenDocument


Our sectoral split included a split of electricity into generation by fuel type plus a distribution sector. We relied on the internet sites of various electricity and energy agencies for capacity levels, on which shares of national activity were based.

Manufacturing, mining, and services data disaggregated at the statistical division level were in quantities rather than values. These were adjusted to fi t state account sector aggregates (ABS 5220.0 ( 2002 )), as wages and industry composition vary between states. Industry investment shares are similar to industry activity shares for most sectors. Exceptions include residential construction input shares, set equal to ownership of dwellings investment shares in each statistical division.

Published ABS data (Tables 4 and 5, ABS 6530.0 ( 2005 )) provide suffi cient commodity disaggregation for the task of splitting regional consumption aggregates into commodity shares. Such data also provide a split between capital city regions and other regions within each state.

In compiling international trade data by region, we fi rst gathered trade data by port of exit or entry. For this task, we used both unpublished ABS trade data avail-able for each state and territory plus the annual reports of various ports authorities. Queensland Transport’s annual downloadable publication Trade Statistics for Queensland Ports gives enough data to estimate exports by port of exit with reason-able accuracy for that state. For other states, port activity is less complex, with most manufacturing trade passing through capital city ports and regional ports specialis-ing in mineral and grain shipments.

State accounts data provide aggregated Commonwealth and state government spending in each region (ABS 5220.0 ( 2002 )). Employment numbers by statistical division for government administration and defence provide a useful split for these large public expenditure items. For other commodities, population shares by statis-tical division were used to calculate the distribution of Commonwealth and state government spending across regions.

By applying these shares to the national CGE database, we were able to compute the USE, FACTOR, and MAKE matrices on the left-hand side of Fig . 2.1 . None of these matrices distinguish the source region of inputs.

2.5.4.1 Region-Specifi c Technology and Output Mix

By default, applying regional output shares to a national dataset leads to industry technologies that do not vary by region. That assumption would be very crude, were it not for the fact that very many sectors are distinguished during data construction . For example, it seems reasonable to assume that bananas are grown in the same way in those (few) regions that they grow. It would be less reasonable to assume that ‘agriculture’ had the same technology.

Within Australia, some regions generate electricity with black coal (which is internationally traded), some with brown coal (which is too bulky to ship). The dif-ference is important as brown coal emits far more CO

2 . To accommodate the regional

technology difference, we distinguished separate brown-coal and black-coal elec-tricity sectors—which each had uniform technology over regions. However, any one

30 M. Horridge

region used only one of the technologies. Prior to simulation, we could aggregate together brown-coal and black-coal electricity sectors to produce a single electricity sector which burns brown coal in some regions, black coal in others.

A related strategy is used to capture regional variation in crop mix. During the data-building process, we usually distinguish a large number of crops, each of which is a single product industry. Thus, we avoid complications by assuming a diagonal MAKE matrix. Prior to simulation, we may aggregate together various agricultural industries, while leaving the associated commodities separate. The effect is that the input technology of ‘agriculture’ varies by region, as does the crop mix. Moreover, inputs such as land and labour can be switched between crops, facilitat-ing the analysis of land use change.

2.5.5 The TRADE Matrix

The next stage was to construct the TRADE matrix on the right-hand side of Fig. 2.1 . For each commodity, either domestic or imported, TRADE contains a 57 × 57 sub-matrix, where rows correspond to region of origin and columns correspond to region of use. Diagonal elements show production which is locally consumed. As shown in Fig. 2.6 , we already know both the row totals (supply by commodity and region) and the column totals (demand by commodity and region) of these submatrices. For Australia, hardly any detailed data on inter-regional state trade are available. We used the gravity formula (trade volumes follow an inverse power of distance) to construct trade matrices consistent with predetermined row and column totals. In defence of this procedure, two points should be noted:

Wherever production (or, more rarely, consumption) of a particular commodity • is concentrated in one or a few regions, the gravity hypothesis is called upon to do very little work. Because our sectoral classifi cation was so detailed, this situ-ation occurred frequently. Outside of the state capitals, most Australian regions are rural, importing ser-• vices and manufactured goods from the capital cities and exporting primary products through a nearby port. For a given rural region, one big city is nearly always much closer than any others, and the port of exit for primary products is also well defi ned. These facts of Australian geography again reduce the weight borne by the gravity hypothesis.

For a particular commodity, the traditional gravity formula may be written:

( ) ( ) ( ) ( )2, ( )· ( )· ,* · *, / , ,V r d r d V r V d D r d r d= ¹λ μ

where V ( r , d ) = value of fl ow from r to d (corresponding to matrix TRADE in Fig. 2.1 ) V ( r ,*) = production in r (known) V (*, d ) = demand in d (known) D ( r , d ) = distance from r to d


The l ( r ) and m ( d ) are constants chosen to satisfy

( ) ( ) ( ) ( )= =å å, *, and , ,* .

r dV r d V d V r d V r

For TERM, the formula above gave rather implausible results, especially for service commodities. Instead, we set

( ) ( ) ( ) ( ), / *, ,* / , ,μ ¹k

V r d V d V r D r d r d

where K is a commodity-specifi c parameter valued between 0.5 and 2, with higher values for commodities not readily tradable. Diagonal cells of the trade matrices were set according to

V ( d , d )/ V ( d ,*) = locally supplied demand in d as share of local production = MIN{ V ( d ,*)/ V (*, d ),1} × F

where F is a commodity-specifi c parameter valued between 0.5 and 1, with a value close to 1 if the commodity is not readily tradable.

The initial estimates of V ( r , d ) were then scaled (using a RAS procedure) so that

( ) ( ) ( ) ( )= =å å, *, and , ,* .

r dV r d V d V r d V r

Transport costs as a share of trade fl ows were set to increase with distance:

( ) ( ) ( ), / , ,T r d V r d D r dμ

where T ( r , d ) corresponds to the matrix TRADMAR in Fig. 2.1 . Again, the con-stant of proportionality is chosen to satisfy constraints derived from the initial national IO table.

All these estimates are made with the fully disaggregated database. In many cases, zero trade fl ows can be known a priori. For example, ABS data indicate that rice is grown in only four of the 57 statistical divisions. At a maximum sectoral disaggregation, the load born by gravity assumptions is minimised.

2.5.6 Aggregation

Even though TERM is computationally effi cient, it would be slow to solve if a full 144-sector and 57-region database were used. The next stage in the data procedure is to aggregate the data to a more manageable size. This stage is automated and effortless. The aggregation choice is application specifi c. For example, to analyse the effects of drought, we might choose a sectoral aggregation that retained detail in the agricultural and agriculture-related sectors, while grouping manufacturing and service industries broadly.

Similarly, the regional aggregation will be tailored to the simulation. For exam-ple, to analyse water shortage, we might aggregate the original 57 regions of Fig. 2.4 to a smaller number of regions (see Fig. 2.5 ) that approximately outline the

32 M. Horridge

watershed of the Murray-Darling river which feeds most of Australian agriculture. To analyse, say, the spread of dengue fever, a different regional aggregation would be appropriate.

As Fig. 2.6 shows, the TERM data process supports some other models used at CoPS, including the single region ORANI-G and MONASH models. By aggregat-ing TERM’s 57-region database down to the eight Australian states, we obtain the kernel of the MMRF database. MMRF is still frequently used, since it incorporates features that TERM lacks, such as emissions modeling. TERM is needed when sub-state detail is required, especially if supply-side shocks must be imposed which differ among regions within a state.

2.6 Conclusion: Applications and Developments of TERM

The TERM framework, developed originally for comparative static analyses of Australian issues, has been extended in several directions:

Fig. 2.4 Statistical divisions in Australia


It has been applied to several other countries, including Brazil, China, Finland, • Indonesia, South Africa, Poland, USA, and Japan. An Italian version is planned. Dynamic or multi-period versions of TERM have been constructed for Australia, • Brazil, and Finland. The South African and Brazilian versions distinguish several household types, to • focus on income distribution. One Brazilian version drives a large micro-simula-tion database, distinguishing 100,000 households. Another Brazilian version focuses on land use, dividing land into four main types • in each region. The aim is to analyse whether increased export and biofuel demand for crops is compatible with preserving Amazon rainforests. TERM’s capacity to model region-specifi c supply-side shocks in small regions • has proved useful in modeling the effects of natural disasters such as earthquakes, droughts, or crop diseases.

The web page http://www.monash.edu.au/policy/term.htm links to a number of these versions.

The wide applicability of the TERM framework rests on two key features: the ability of the model to solve quickly without compromising regional or sectoral detail and a method to construct a database rapidly, so that the fi rst simulations can be run soon, maximising the scope for experience to suggest ways to improve the model or data.

Fig. 2.5 Aggregating from master database to policy simulation (watershed) regions

http://www.monash.edu.au/policy/term.htm

34 M. Horridge

TRADE MATRIX144 commodity

x [dom,imp]x 57 origin regionx 57 dest region

ABS 1997107 sector

Input-Output Tables

107 sectorHAR databasefor ORANI-G

Data to split107 sectors to 144

144 sectorORANI-Gdatabase

144 sectorMONASHdatabase

40 sector 8 regionMMRF database

144 sectorsingle region TERM

database

40 sector 8 regionTERM database

40 sector 30 regionTERM database

144 sector57 destination

TERM database

144 sector57 origin 57 destTERM database

TERM modelMMRF model

Data to split users by57 dest regions

R001 R002 R003 etc

aggregation

aggregation

programs

programs

programs

simplif

programs

programs

supplyby

programs

additional data:state gov accounts

emissionscap stocks & RoR

population, employment

Distance matrix (gravity); imports by

port of entry

demandby

Fig. 2.6 Producing regional databases for MMRF and TERM


References

ABS (Australian Bureau of Statistics) (2002 (and previous issues)) Australian National Accounts: state accounts, catalogue 5220.0, Canberra

ABS (Australian Bureau of Statistics) (2005) Household expenditure survey, Australia: summary of results, catalogue 6530.0, Canberra

Adams P, Horridge M, Wittwer G (2002) MMRF-Green: a dynamic multi-regional applied general equilibrium model of the Australian economy, based on the MMR and MONASH models. Prepared for the regional GE modelling course, 25–29 Nov 2002

Dixon P, Parmenter B, Vincent D (1978) Regional developments in the ORANI model. In: Sharpe R (ed) Papers of the meeting of the Australian and New Zealand section regional science asso-ciation, third meeting, Monash University, Melbourne, pp 179–188

Dixon P, Parmenter B, Sutton J, Vincent D (1982) ORANI: a multisectoral model of the Australian economy. North-Holland, Amsterdam

Flegg A, Tohmo T (2008) Regional input–output models and the FLQ formula: a case study of Finland. Discussion paper 0808, Department of Economics, University of the West of England

Flegg A, Webber C, Elliott M (1995) On the appropriate use of location quotients in generating regional input–output tables. Reg Stud 29:547–561

Giesecke J (1997) The FEDERAL-F model. CREA paper no. TS-07, Centre for Regional Economic Analysis, University of Tasmania, November

Higgs P, Parmenter B, Rimmer R (1988) A hybrid top-down, bottom-up regional computable general equilibrium model. Int Sci Rev 11:317–328

Horridge M, Madden J, Wittwer G (2005) The impact of the 2002–03 drought on Australia. J Policy Model 27:285–308

Liew L (1984) “Tops-Down” versus “Bottoms-Up” approaches to regional modelling. J Policy Model 6:351–368

Madden J (1990) FEDERAL: a two-region multi-sectoral fi scal model of the Australian economy. PhD thesis, University of Tasmania, Hobart

Pearson K (1988) Automating the computation of solutions of large economic models. Econ Model 7:385–395

Peter M, Horridge M, Meagher GA, Naqvi F, Parmenter B (1996) The theoretical structure of MONASH-MRF Centre of Policy Studies working paper. http://www.monash.edu.au/policy/ftp/workpapr/g-140.pdf . Accessed 8 Mar 2011

Wittwer G, Horridge M (2010) Bringing regional detail to a CGE model using census data. Spat Econ Anal 5:229–255

http://www.monash.edu.au/policy/ftp/workpapr/g-140.pdf



Abstract The massive master database of TERM needs to be aggregated before it can be used for any simulation. There is demand for moving to dynamic TERM simulations and rewards from doing so due to additional insights that arise from the infl uence that a dynamic baseline may have on a policy simulation. This chapter covers a number of issues concerning dynamic modeling with TERM. We start by outlining the motivations for moving from comparative static to dynamic regional modeling. Following that, we provide an overview of how we go about making a version of TERM dynamic. This includes details of how to vary the time intervals within a dynamic model. Next is an explanation of using the master database of TERM to prepare variable aggregation versions of dynamic TERM. The chapter also outlines how recursive dynamic models are run. RunDynam (specialist software) is a very useful tool for the dynamic CGE practitioner.

Keywords Aggregation procedure • Dynamic software • Dynamic modeling • Quarterly modeling

3.1 Motivations for Moving to Dynamic Modeling

Comparative static CGE modeling has made an important contribution to policy analysis. The approach compares a scenario with a base case without defi ning the base case. Invaluable insights have arisen from this approach. For example, output and job losses suffered by formerly heavily protected import-competing sectors

G. Wittwer (*) • G. Verikios Centre of Policy Studies , Monash University , Wellington Rd, Clayton campus , Clayton , VIC 3800 , Australia e-mail: [email protected] ; [email protected]

Chapter 3 Introducing Dynamics to TERM

Glyn Wittwer and George Verikios

38 G. Wittwer and G. Verikios

may be substantially offset by gains in other sectors of an economy when import tariffs are cut. These gains come about through resource movements and real exchange rate effects that boost the international competitiveness of trade-exposed sectors not directly affected by tariff cuts.

There are a number of reasons we may wish to move from comparative static to dynamic CGE modeling. Changes in the base case may matter. For example, it was not possible to include escalating import competition from China in the 1990s in comparative static simulations of trade reforms in manufacturing. This is not to say that dynamic modeling at the time routinely included China’s emergence in the baseline but rather that there is provision within dynamic baselines for such changes. That is, although policy simulation outcomes depend to some extent on the underly-ing forecast, we often fall short of an accurate forecast.

In the context of the study on water buybacks in Chap. 6 , changes in rainfall conditions impact on the policy simulation. A year of drought will raise the price of irrigation water even if there is no shortfall in water allocations, as discussed in Sect. 7.4 . This is because drought-induced dryland productivity losses, together with reduced water availability on irrigated land from rainfall, will raise the value of the marginal product of water. This implies that if we are running a buyback simula-tion with a baseline that includes recurrent droughts, the economic costs of the buyback scheme will rise relative to a baseline that does not include droughts. Similarly, by raising the economic costs of buyback, an increased frequency of expected drought will raise the price that the Commonwealth pays for permanent water. Consequently, the calculation of national welfare (from a purely economic perspective without considering the environmental benefi ts), the cost of full com-pensation to the Commonwealth, and the impact on regional outcomes relative to forecast will change as baseline rainfall (implied mainly by dryland productivity) and water availability change.

A second reason arising from the buyback study of Chap. 6 for a dynamic approach is that the pace of policy enactment matters. The more time given to farmers to sell entitlements to the Commonwealth, the more time they will have to utilise water-saving and other primary factor-saving technologies, thereby reducing the impact of reduced water availability on irrigation output.

Another reason is that a dynamic model allows us to update the database. CGE modelers may be frustrated by a relatively dated CGE database. If we start with a 2006 database, but wish to start our policy simulation in 2012, we can ascribe as shocks to the model observed changes in baseline macroeconomic variables, inter-national trade conditions, estimates of productivity growth at the industry level, and, if relevant, consumer taste changes over time. Dixon et al. ( 2000 ) contain a detailed explanation of the historical simulation procedure used to estimate changes in unob-served variables over time. Updated databases will contain updated sectoral weights with which to run a policy simulation.

A common use of dynamic simulations is to deal with climate change scenarios. Whereas we can deal with most policy scenarios in a dynamic framework with a 10- to 20-year simulation, climate change scenarios typically run out to the year 2050 or 2100. There is a tendency towards ‘straight line’ projections of all variables

393 Introducing Dynamics to TERM

in the baseline when simulating very long runs. TERM-H2O does not include greenhouse gas accounts. The preferred tool at the Centre of Policy Studies for greenhouse gas scenarios is MMRF-Green (Adams et al. 2003 ) . As observations of climate change appear to indicate wider and more frequent variations between dry years and wet years, there is still a role for TERM-H2O in climate change analysis focussed on impacts on agriculture. Given that technological change plays an increasing role with time, it follows that assumed rates of technological change are critical in determining the impacts of dynamic scenarios run over many decades.

3.2 Making TERM Dynamic

Following the methodologies detailed in Dixon and Rimmer ( 2002 , pp. 2–10), TERM-H2O used in Chaps. 6 and 7 and TERM-DYN used in Chap. 8 are dynamic. As Dixon and Rimmer ( 2002 ) explain, their model includes three types of inter-temporal links. These are physical capital accumulation, accumulation of fi nancial assets/liabilities and lagged adjustment processes.

3.2.1 Physical Capital Accumulation

At the heart of the theoretical modifi cations for dynamics is the linking of annual investment fl ows to capital stocks. Investment makes up a substantial proportion of economic activity on the expenditure side and capital rentals a substantial propor-tion on the income side in the economy. Typical of most CGE models, capital accu-mulates in dynamic versions of TERM according to:

+ = − +, 1 , , ,(1 ) ,j t j t j t j tK K D I (3.1)

where K

j,t is the quantity of capital available to industry j in year t

I j,t is the quantity of investment (new capital) in industry j in year t

D j,t is the rate of depreciation

The expected rate of return in industry j determines its level of investment in a given period. An investment supply curve shows the required rate of return for an additional dollar of investment, which depends on the rate of growth of industry j ’s capital stock.

3.2.2 Accumulation of Financial Assets/Liabilities

Another important link for Australia (and other heavily indebted nations) is that between current account fl ows (i.e., the trade balance and net interest payments to


foreigners on existing debt) and net foreign liabilities. This link feeds into net disposable income and the consumption function that links household spending to disposable income.

Given that Australia’s net foreign liabilities are a substantial share of GDP, moving the model from one year to the next without imposing any economic updating shocks will still have impacts, as annual interest payments on net foreign liabilities accumulate. This effect, referred to by Dixon and Rimmer ( 2002 ) as ‘momentum’, is an example of an initial condition (i.e., a substantial foreign debt) that matters in a dynamic framework but is not part of comparative static analysis.

3.2.3 Lagged Adjustment Processes

Lagged adjustment processes entail period-by-period partial adjustments. Following Dixon and Rimmer ( 2002 ) , investment in TERM-H2O involves such a process. Partial adjustments in investment eliminate inconsistencies between levels of invest-ment and rates of return on one hand and the theory of investment behaviour on the other hand.

TERM-DYN, as we refer to the dynamic TERM variant without water accounts, includes a theory of sluggish labour adjustment at the regional level (Wittwer et al. 2005a ) . The regional labour market adjustment mechanism, in levels, is given by:

1

–1

1 1 .r r r r

t t t tr r r r

t t t t

W W EMP LS

Wf Wf EMPf LSfa−⎛ ⎞ ⎛ ⎞ ⎛ ⎞ ⎛ ⎞

− = − + −⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ (3.2)

The interpretation of ( 3.2 ) is that if the deviation shock weakens the labour mar-ket in region r and period t relative to forecast, real wages r

tW in deviation will fall gradually relative to forecast r

tWf . In addition, there will be an initial enlarged gap between labour market demand r

tEMP and supply rtLS , relative to forecast levels

rtEMPf and r

tLSf . In successive years, the gap between demand and supply will gradually return to forecast through a further decline in real wages. The speed of labour market adjustment is governed by a , a positive parameter.

The regional labour supply equation is:

( )( )

( )( )

γ γ

γ γ=∑ ∑

r rrt tt

r q q q qt t t t t

q q

W WfLS

LSf W S Wf Sf

(3.3)

The deviation in regional labour supply from forecast depends on the deviation in regional relative to national real wages from forecast. In ( 3.3 ), ( )q q

t tq

W S∑ γ is a

measure of labour responsiveness to real wages summed across all regions, where g is a positive parameter and q

tS is the share of region q in national employment. Should the deviation in real wages from forecast fall in a particular region relative to the deviation in real wages nationally, this equation implies that labour supply in


the particular region will fall, while in other regions it will rise. Combining ( 3.2 ) and ( 3.3 ), adjustment in the labour market in a given region will initially occur via a combination of additional unemployment and lower real wages. Unemployment will eventually return to forecast rates, with lower real wages. As real wages fall relative to the base case, the region’s labour supply will also fall. Within this theory, long-run labour market adjustment occurs as a combination of inter-regional labour migration and changes in regional real-wage differentials.

TERM-H2O does not include lagged adjustment processes in the labour market. Rather, as the focus is on the Murray-Darling Basin which accounts for a small share of national economic activity, national employment is assumed to be exoge-nous with real wages equal across regions. Such an assumption might overstate inter-regional labour movements in some scenarios. TERM-H2O is run routinely with more than 20 bottom-up regions. Labour is substitutable with water in irriga-tion sectors, and the price variation in water may be extreme. Were we to introduce a sluggish labour market to TERM-H2O at the regional level, with inter-regional wage differentials, adjustments induced by wide variations in the price of water could create modeling diffi culties. A simplifi ed labour market theory is one conces-sion we have chosen in order to ease the solution process in running TERM-H2O.

3.2.4 Another Potential Dynamic Link in TERM-H2O: Reservoirs

We have not implemented dynamic links between water fl ows and reservoir stocks in TERM-H2O. Instead, we keep annual water allocations exogenous. While it may seem an obvious move in a model that contains water accounts, we have opted to avoid an overly elaborate link between rainfall and dam levels. This is more in the domain of hydrological models. That is, we regard optimal period-by-period supply of water subject to seasonal conditions as the task of water managers. There are many policy issues that remain for economic modelers to explore in the context of water policy without entering into the issue of timing of water supply. Signifi cant insights into the interaction between regional annual rainfall and water policy sce-narios have arisen without a theory of optimising water supply. These include the impact of baseline droughts on the national welfare outcomes and costs to the Commonwealth of water buybacks.

3.2.5 Fiscal Accounts

Both TERM-H2O and TERM-DYN contain fi scal accounts covering Commonwealth and state revenues and expenditures. They can become extremely important in pol-icy simulations, most notably when analysing the impacts of indirect tax scenarios.


A classic example arising from MMRF-Green (Adams et al. 2003 ) concerns greenhouse gas scenarios.

Suppose we impose a carbon tax in a model by fi xing the national government’s budget balance via an endogenous shift in some other tax rate. In a comparative static model, this will result in a markedly different national welfare outcome from imposing the tax without an offsetting tax cut elsewhere. The apparent welfare loss will be much greater if we allow the budget surplus to grow with the new tax. Tax trade-offs are a signifi cant part of the carbon tax plan announced by the Gillard Government in July 2011 (Thompson 2010 ) .

In a dynamic CGE model, we measure welfare as a terminal condition. That is, we calculate the net present value of deviations in public and private consumption from forecast summed across all years of the simulation and subtract the net present value of the change in net foreign liabilities relative to forecast in the fi nal year of the simulation. Such an approach reduces the burden on fi xing the budget balance. In the case of imposing a carbon tax without an offsetting reduction in some other tax, any increase in the budget surplus relative to forecast will be refl ected in a reduction in net foreign liabilities relative to forecast in the fi nal year of the simula-tion. There may still be losses in international competitiveness that are exacerbated by not fi xing the budget balance, but they form only one part of the welfare calculation.

The main data sources for the fi scal accounts are taxation revenue data (ABS 2007a ) and government fi nancial statistics (ABS 2007b ) . Fiscal accounts are based on the nine governments of Australia (one federal and eight state and territory gov-ernments). We have devised a computer program that links fi scal accounts to the master database. In doing so, we need to distinguish between data that are already in the input-output structure of the core CGE database and data from fi scal accounts that provide additional information. For example, for most commodities in the core database, indirect taxes on household consumption are mainly in the form of GST (the goods-and-services tax). This is a federal tax, but the revenues are distributed to the states and territories. The fi scal accounts in TERM-H2O and TERM-DYN include Commonwealth-to-state transfers. In the case of motor vehicles, we have fi scal data on state and territory stamp duties. In allocating taxation revenue to the Commonwealth from indirect taxes on motor vehicles, we fi rst subtract state and territory government stamp duties, which we have already added to state and terri-tory government revenues.

There are dangers in this approach to fi scal accounting. We need to align fi scal data with appropriate commodities or industries. For example, motor vehicle duties reported in fi scal data might be spread over a number of transport commodities in the input-output table. If we exclude one or more of these commodities, our residual method may let us down by calculating negative motor vehicle tax revenues to the Commonwealth.

In Chap. 2 , we state that the best way to improve a TERM database is to run simulations. This is as much so with the fi scal accounts as other aspects of the data-base. If our simulations do not concern fi scal detail within the model, we might never discover errors in the fi scal accounting of the model. Once fi scal detail is


important in a particular scenario, we may discover the need to amend the links between the fi scal accounts and input-output database.

3.3 Preparing TERM for Dynamic Simulations

The TERM philosophy outlined in Chap. 2 is to prepare a massive master data-base, with small region representation and more sectoral detail than is available in published national input-output tables. This database will be far too large for sim-ulations. The key to applying the model to a wide variety of applications is to devise a variable aggregation procedure, thereby tailoring the dimensions of the database to a manageable size for an individual application. For comparative static versions of TERM, the variable aggregation procedure entails mapping the sec-tors and regions of the master database to target sectors and regions. The aggrega-tion will maintain separate representation of sectors and regions that are the focus of a particular study while aggregating those of little interest in that study.

It follows that if variable aggregation is a useful tool for comparative static stud-ies, then it will also be useful in dynamic studies. Indeed, Chaps. 6 and 7 are studies using dynamic TERM-H2O which concentrates on farm sectors in regions of the Murray-Darling Basin. The regional and sectoral disaggregations are detailed enough to analyse water buybacks and drought in the basin. Chapter 8 uses an entirely different aggregation of the master database to analyse water issues in pre-dominately urban south-east Queensland. More generally, TERM-DYN has been used to analyse a wide variety of topics in studies undertaken for clients at the Centre of Policy Studies. These include hypothetical disease outbreaks and weeds control scenarios in agriculture (Wittwer et al. 2005a, b, 2006 ) . Analyses of infra-structure projects have included channel deepening in the bay of Melbourne (Pricewaterhouse Coopers 2007 , appendix A), broadband rollouts, upgrades at docking ports and road-building scenarios. Other scenarios have included (almost simultaneously) a hypothetical outbreak of Moko disease that wiped out half the banana crop in Far North Queensland and the impacts of actual cyclone Larry 1 (that wiped out 90% of all crops) on the same region. TERM-DYN has also been applied to gambling tax scenarios and several dam construction scenarios, including that of Chap. 8 . More recently, a quarterly version of TERM-DYN provided a tool for analysis of the January 2011 fl ood in Brisbane. 2

Each study requires a separate regional and sectoral aggregation of the master database. Here, we go through the steps taken to transform comparative static TERM into a dynamic model. First, we outline the steps taking to devise a dynamic CGE version of TERM. Then we explain how we go about preparing an aggregation of the master database of TERM for dynamic simulations (Fig . 3.1 ).

1 See http://www.ga.gov.au/image_cache/GA8708.pdf 2 Some projects undertaken for clients remain confi dential.

http://www.ga.gov.au/image_cache/GA8708.pdf


3.3.1 The Initial Database and Conditions

A dynamic model adds several modules to a comparative static framework. No lon-ger is a database that contains only fl ows suffi cient for our purposes. In TERM, we calculate capital stocks based on typical year after-tax rates of return, using capital rentals as a starting point. This implies that we need typical year rates of return for each industry, plus an estimate of initial capital stocks. In adding equations that link fl ows to stocks, we need depreciation rates in the equations concerning investment fl ows and capital accumulation. We also add net foreign liabilities (NFLs) to the database. Therefore, we need interest rates when we link the balance of trade and NFL accounts.

Essentially, we are creating a dynamic master database that is even larger than that of the TERM comparative static model outlined in Chap. 2 . In addition to the core CGE database, we have capital stocks for each sector and region and deprecia-tion rates for each sector.

We need to update the year-to-year baseline in dynamic simulations. To do so, we need the usual array of macro and sectoral shocks used in a national model. In practice, a dynamic regional modeling will have slightly less elaborate shocks than

RegionalConvert national macro to regional macro shocks

Initial conditions1. Initial multi-regional CGE database2. Add capital stocks, net foreign liabilities3. Capital stocks reflect typical year rates-of-return4. Typical year rates-of-return by industry5. Sectoral depreciation rates6. Interest rates7. Stock-flow accumulation equations

National dynamic1. Macro shocks, up dating and forecasting2. Industry technological changes3. Consumer taste changes4. Observed trade volumes and prices

Fig. 3.1 Outline of preparation of dynamic TERM


a national model that follows the MONASH methodology, elaborated by Dixon and Rimmer ( 2002 ) . For example, macro shocks might include regional GDP, aggregate consumption and aggregate investment, but exclude import and export volumes. The sectoral shocks typically will include primary factor technological shocks and consumer taste changes.

Given that dynamic modeling is more computationally demanding than com-parative static modeling, it will be impossible to run the model without aggregating the dynamic master database fi rst. We require an aggregation procedure that enables us to prepare a model for dynamic simulations that is suffi ciently small to run. Section 3.5 outlines the aggregation procedure.

3.3.2 Running a Dynamic Model

In dynamic applications, there are three sets of year-by-year simulations. The base-line forecast simulation takes forecasts of regional (statewide) economic growth from forecasting agencies. Target macro variables including regional real GDP, aggregate employment, aggregate consumption and aggregate investment are usu-ally endogenous. In the forecast run, we make them exogenous through swaps with various shift terms. For example, a statewide-level technological change term becomes endogenous so as to meet the state-level real GDP forecast.

The baseline forecast scenario is rerun with a policy closure to avoid lineariza-tion errors. That is, variables that are usually exogenous in policy simulations, such as statewide technological change used to target regional real GDP, are made exogenous and shocked by the endogenous change of the initial baseline simula-tion. The rerun simulation therefore includes a lot of shocks to various shifters that were endogenous in the forecast run so as to accommodate forecast targets. This enables us to include forecasts over time while running the model with a closure that is suitable for policy analysis. The policy simulation includes all the shocks of the policy rerun simulation plus additional shocks specifi c to the scenario. Dixon and Rimmer ( 2002 , pp. 233–274) detail the various closures (i.e., combinations of endogenous and exogenous variables) used in dynamic simulations.

3.4 Altering the Time Interval of a Dynamic Model

Dynamic CGE models typically represent time as discrete periods of 1 year, and this is also true of TERM-DYN and TERM-H2O. Dynamic CGE models with annual periodicity are well suited to analysing events that last for a year or more. For events that last much less than a year and whose effects are major, a model with quarterly periodicity is more appropriate. For example, TERM-DYN modeling of the impacts


of January 2011 fl ood in Brisbane used quarterly dynamics. 3 Although an annual model would have dealt reasonably with the reconstruction of damaged buildings, the city centre ground to a halt for a week and did not resume business as usual for several weeks. A quarterly model was more appropriate for dealing with the severe short-term disruption to service activities in the city centre. The character of the disruption was quite different in a quarterly model than it would have been in an annual model. The initial disruption was larger and shorter than as an annual model would have depicted it. In a quarterly model, it is possible to start the reconstruction phase with its accompanying employment stimulus in the second quarter rather than assume that it only starts in the second year.

3.4.1 The Treatment of Time

Annual CGE models, such as TERM-DYN, are commonly solved recursively so that each period (year) is solved sequentially, and periodic solutions are computed in the order 0, 1, 2, 3,…. 4 For such models, time is treated as comprising discrete intervals, and economic variables are assumed to change at the end of each interval. An alternative treatment of time is where it is assumed to change continuously.

Discrete-time annual models take the form

( ),Y G X= (3.4)

where Y and X are the levels of the endogenous and exogenous variables in a period. Computations are then carried out according to

( ) ,Y G X XΔ = Δ′ (3.5)

where D refers to changes from one period to the next. If we have changes from one year to the next for the exogenous variables, D X , then to model quarterly changes, we must divide these changes by four. Besides dividing exogenous changes by four, there are three other changes required to move from annual to quarterly periodicity:

(a) Equations must be added that handle quarterly accumulation of stock variables. (b) The base data for the initial values of lagged variables must be altered, e.g., if

the data for a price variable in the base and lagged years are 1 and 0.96, then we should use 1 and 0.99 for the quarterly model.

(c) Parameter values in equations describing partial adjustment mechanisms must also be altered.

3 Another example where quarterly modeling is important is the outbreak of an infectious disease; these tend to begin and end within a year. See, for example, Dixon et al. ( 2010 ) and Verikios et al. ( 2011 , 2012 ). 4 The exceptions are inter-temporal models that compute results simultaneously for all time periods (e.g., McKibbin and Wilcoxen 1999 ; Malakellis 2000 ) .


3.4.2 Physical Capital Accumulation

Physical capital accumulation in TERM-DYN has been briefl y described in stylised terms in Sect. 3.2.1 ; here, we are more specifi c in order to describe the move from annual to quarterly capital accumulation.

We rewrite Eq. 3.1 as

1 ,t t tj j j jKE KB D I⎡ ⎤= − +⎣ ⎦ (3.6)

where tjKB and t

jKE are the quantity of capital available for use in industry j at the beginning and end of year t , respectively, t

jI is the quantity of new capital created (i.e., investment) for industry j during year t , and jD is the rate of depreciation in industry j , treated as a parameter.

Beginning-of-year capital is defi ned as capital at the end of the previous year, 1t

jKE − , or

1 11 ,t t tj j j jKB KB D I− −⎡ ⎤= − +⎣ ⎦

(3.7)

where 1tjKB − and −1t

jI are the values of depreciated capital and investment in the initial solution, i.e., the year t − 1. With t

jKB , jD and tjI representing annual values,

tjKE in ( 3.6 ) will grow at an annual rate.

To defi ne a quarterly rate of accumulation (signifi ed here by superscript q rather t ) with no change in the values of any variable in ( 3.6 ), we add the following equation to the model:

1 .q q qj j j jKE KB D I⎡ ⎤= − +⎣ ⎦

(3.8)

In deriving ( 3.8 ), we create quarterly values for depreciation rates, qjD , and invest-

ment, qjI , that ensure new capital accumulates at a quarterly rate. Thus, /4q t

j jD D= and /4q t

j jI I= . Existing capital in ( 3.8 ), qjKB , is defi ned similarly to t

jKB :

1 11 ,q q qj j j jKB KB D I− −⎡ ⎤= − +⎣ ⎦

(3.9)

where 1qjKB − and −1q

jI are the values of depreciated capital and investment in the initial solution, i.e., quarter q − 1. The initial solution has now been reinterpreted as the previous quarter rather than the previous year. That is, if our initial solution in the annual model is data for 2010, then the initial solution for the quarterly model is 2010:4. This change is effected by applying quarterly values for depreciation rates and investment.

Recall from Sect. 3.2.1 that in an annual model, the expected rate of return in industry j determines its level of investment in a given year, t

jI . This relationship still holds in the quarterly model, and so Eq. 3.6 remains in the model. Thus, the relation-ship between the rate of return on capital and investment is an annual one even though the periodicity of the model is now quarterly. This assumes that fi rms still make investment plans over a 1-year time horizon, but only the fi rst quarter of those yearly


plans are implemented in the current period (quarter). Thus, tjKE and t

jI are never realized: they are only planning variables. Further, t

jKE and tjI are reevaluated every

quarter, and their value will change on a quarter-by-quarter basis.

3.4.3 Accumulation of Financial Assets/Liabilities

Just as Eq. 3.9 is the quarterly version of ( 3.1 ), so too must we modify other stock-fl ow relationships in the model. In the dynamic TERM models, net foreign liabilities in the present quarter will be equal to the previous quarter’s net foreign liabilities plus quar-terly interest payments plus the previous quarter’s balance of trade defi cit.

There is a small complication. We retain the fl ows in the database on an annual basis. Unlike our treatment of capital, in which depreciation is set on a quarterly basis, we retain annual interest payments in the link between the balance of trade and net foreign liabilities, as the entire expression is annual. Consequently, we calculate the ordinary change in net foreign liabilities as one quarter of the difference between the level of net foreign liabilities and the lagged level of net foreign liabilities.

3.4.4 Lagged Adjustvment Processes

In moving from annual to quarterly periodicity, any parameters controlling lagged adjustment responses must be modifi ed to exhibit quarterly responses. An example of such a parameter is a in Eq. 3.3 , which controls the speed of labour market adjustment. This parameter is usually calibrated so that the employment deviations of a shock to the economy are approximately zero after about 5 years. We wish this condition to continue to hold in the quarterly model. This requires that a is divided by four so that the employment deviations of a shock to the economy are approxi-mately zero after about 20 quarters.

3.5 Variable Dynamic Aggregation

In addition to the core database and parameters of a comparative static model, a dynamic model requires data and parameters as outlined in Sect. 3.3 . Here, we dis-cuss the aggregation of the master dynamic database for practical use.

3.5.1 The Core Database

The procedure for aggregating the core database is the same as for aggregations of TERM’s comparative static database. First, we require a mapping from the commodities, industries and regions of the master database to those of the aggregation


required for a particular simulation. Then we run a program prior to aggregation that matches database weights with coeffi cients that require weighted aggregation. Coeffi cients not matched to database weights in this program are aggregated without weights by default. For example, most of the core CGE database is comprised of matrices designated in (millions of) dollar values, which are aggregated without weights. The parameters in the core database are aggregated using weights. For example, we aggregate household expenditure elasticities using a matrix of values of household commodity purchases.

3.5.2 Model Sets

The specifi c preparations for a highly mechanised dynamic aggregation procedure include modifi cation of set statements within the model code. Dynamic models typically include many subsets of industries and commodities used in an array of equations. In dynamic TERM, subsets of commodities are designated as traditional exports, tourism-related sectors and sectors related to various forms of taxation, such as gambling and motor vehicles. In addition, there is a subset of exogenous investment industries: there are usually public industries for which investment does not follow a rate-of-return rule, as applies to other industries. While it would be possible to go through the TERM model code and change the sectors manually in each of these particular sets, the task is time-consuming. An alternative is to defi ne subsets of the commodity and industry sets using parameters. This is an approach fi rst used by Mark Horridge in the ORANIG teaching model. 5

To illustrate how we proceed with parameter-based sets, we use the original example from the TABLO code of ORANIG:

Coeffi cient (parameter)(all,c,COM) IsIndivExp(c); Read IsIndivExp from fi le BASEDATA header “ITEX”; Set TRADEXP # Individual export commodities # = (all,c,COM: IsIndivExp(c)>0.5); Write (Set) TRADEXP to fi le SUMMARY header “TEXP”;

The parameter IsIndivExp designates commodities as either exports with their own export demand curve (i.e., IsIndivExp >0.5) or commodities that are part of the grouped exports without individual export demand behaviour ( IsIndivExp < =0.5). In a dynamic aggregation program, we can aggregate the parameter IsIndivExp using values of exports as weights. Weighted aggregation implies that if a commod-ity is assigned as an individual export but ends up with a small value share in an aggregated sector of which the other components are grouped exports, the new sector will be part of the grouped exports. That is, the aggregated sector will have a below-threshold parameter defi ning individual exports. With unweighted aggregation,

5 ORANIG is downloadable from http://www.monash.edu.au/policy/oranig.htm

http://www.monash.edu.au/policy/oranig.htm


any aggregated commodity including an individual export among the original commodities would become an individual exporter regardless of the value share weight of the initial individual exporter in the aggregated entity.

In sets containing a single element in the master database, either an unweighted aggregation is appropriate or the parameter defi ning the set should apply for any value greater than 0. Otherwise, we may end up with an empty set, if that element ends up in an aggregated sector in which the parameter is below the assigned threshold.

One subset that can cause problems in new dynamic databases is the set of exogenous investment industries. Some industries do not behave well if subjected to the usual endogenous investment behaviour of the model. In particular, industries with an unusually high or low investment-to-capital rental ratio are troublesome. Sorting out a suitable combination of endogenous and exogenous investment indus-tries is to some extent a matter of trial and error. In many aggregations, troublesome industries do not emerge because they have been aggregated into a relatively broad sector in which they are not disruptive.

3.5.3 Baseline Year-to-Year Shocks

There are various ways of assigning baseline shocks in the GEMPACK/RunDynam suite of programs we use. The easiest way to prepare shocks for dynamic aggrega-tion is to put all shocks into a header array fi le. This way, we can use weights to aggregate to the sectors and regions of choice. The following line comes from the baseline shocks fi le in RunDynam (an outline of RunDynam follows in Sect. 3.6 ):

shock <year> xinvitot = select from fi le FinalShk.har header “YSHK”;

In this line, FinalShk.har is a fi le specifi c to a particular aggregation that includes shocks as a weighted aggregation of those in the master shocks fi le. In this example, xinvitot is the percentage change in the investment relative to the previous period. Elsewhere, a set of exogenous investment industries is defi ned, using parameters as described above. Only those sectors of xinvitot assigned as exogenous will be shocked, hence the words ‘select from’ in the line above.

3.5.4 Parameters Concerning Investment Behaviour

Some parameters within the investment/capital module of the dynamic model vary between industries. For example, motor vehicles and information technology equip-ment have relatively high rates of depreciation. Housing and rail transport have rela-tively low rates of depreciation. Having assigned depreciation rates for all industries in the master database, we aggregate to the sectors of choice using capital returns from the core database as weights. Other parameters may not vary between industries.


Since the models include a combination of parameters that rarely vary between industries and others that do, we fi rst run a program that defi nes parameters at either the master database level of industry representation, for parameters that vary between sectors, or at the aggregated level for parameters that tend to be uniform across sectors. An example is the adjustment parameter that determines the speed at which rates of return move back to normal level. We defi ne this parameter in dynamic TERM models as an industry vector, though for most applications, a scalar would be suffi cient. Then we use weights to aggregate those parameters such as depreciation that were not assigned a value at the aggregated level in the program.

3.5.5 Automating the Aggregation Procedure

In order to automate the aggregation of dynamic TERM, we use a suite of GEMPACK-based programs. A relatively complex task is broken into a series of small programs and commands. For example, using the mappings from the master database to the aggregated industries, commodities and regions, we fi rst aggregate the core database. Next, we aggregate the parameters defi ning the numerous subsets in the model. Aggregation of the fi scal module follows. Then the investment-capital module is aggregated followed by the baseline year-by-year shocks.

The programs are run in a batch fi le which helps trace errors. Errors typically arise from inconsistent mappings. For example, if the new industry list contains a sector called ‘agriculture’, but the fi le which maps the full list of industries to new sectors refers to ‘agricultures’, we will get an error during the aggregation procedure.

Another potential source of error in dynamic aggregations might arise from over-lapping sets. Suppose we have an equation for communications export demands and another equation for tourism export demands. In the master database, these two equations clearly involve different sectors. But in aggregation, we might defi ne an aggregated services sector that involves both communications and tourism sectors. In such a case, we will end up with structural singularity. That is, we will have two equations solving for one variable until we turn off one of the two equations by making a shifter in one of the equations endogenous.

Preparation of the dynamic master database and shocks, plus aggregation programs and command lines, takes time. But once the procedure is operating, it is reproducible and rapid. When the procedure is working well, the main task of the aggregator is to prepare mappings to the aggregation of choice and test the new aggregation for possible bugs by using it to run a dynamic simulation.

Finally, although preparation of the TERM-H2O database requires a variant of the aggregation procedure, there is a tendency to use only one or two aggregations for all scenarios. There is a good theoretical reason for not running TERM-H2O with overly aggregated regions. This is that TERM-H2O allows water trading between regions in the southern part of the Murray-Darling Basin but not in the northern part of the basin. If we aggregate the regions, we implicitly free up mobility


of water by allowing movement over a larger region. If we are examining policy impacts in both the southern and northern part of the basin, it is most appropriate to have the catchments across the basin represented individually within TERM-H2O. Otherwise, we will overstate the mobility of water in the northern basin: it consists of tributaries fanned out over large distances between which water trading is not feasible. Chapters 6 and 7 , which concentrate on impacts on the southern part of the basin, use a database with relatively aggregated regions in the northern part of the basin. On the other hand, the work commissioned by the Murray-Darling Basin Authority examined the impacts on both the southern and northern part of the basin (Wittwer 2010 ) . This required disaggregation of the northern basin into catch-ment regions. Increases in computing memory have ensured that running a version of TERM-H2O with 23 bottom-up regions is manageable.

3.6 RunDynam: Software to Automate Dynamic Simulations

In Sect. 3.3.2 , we state that each year of a dynamic simulation entails running the model three times. For a 20-year run, that implies that we need 60 simulations. In turn, the solution fi le of the baseline run for a given year includes shocks that we will use in the rerun and policy modes for that year. For example, we may target export volumes for certain commodities in the baseline or business-as-usual fore-cast run. Since we want export volumes to be endogenous in the policy run, we do a closure swap between export volumes and export demand shifters in the initial rerun. The appropriate shocks to give to the export demand shifters are provided by the solution for the relevant shifter variable in the forecast run. Dynamic simula-tions may contain an elaborate combination of shocks to shifter variables, twist variables (factor or source), taste variables and technological change variables. Since the dynamics used in the Monash school of CGE modeling are recursive, the policy simulation uses the data fi les updated in the simulation for the previous year as an input. In turn, for a given year, the data fi les updated in the simulation become the input fi les for the following year.

Early methods of automating dynamic simulations involved the use of batch fi les that included command lines defi ning each simulation. The data in fi les for the policy simulation for year t + 1 were updated fi les generated by the policy simula-tion for year t . The shock fi les for the two simulations that use the policy closure (i.e., the rerun baseline and policy simulations) required as inputs the solution fi le from the baseline run for the same year. The solution fi les from the baseline run were converted to a form that was readable in the GEMPACK software. The use of batch fi les made dynamic simulations possible and reproducible.

There are several disadvantages in the batch fi le approach to dynamic simulations. It is a time-consuming process to set up simulations. Moreover, archiving of dynamic simulations and making a particular simulation run available to other users is a dif-fi cult task. Both seasoned practitioners and newer modelers could see the advan-tages of dynamic simulations being undertaken with a greater degree of automation.


This is the purpose of the RunDynam software created by the GEMPACK team at the Centre of Policy Studies. 6 RunDynam simplifi es the task of running dynamic simulations by automatically creating the command fi les that read in updated fi les from the previous year’s simulation. The process of reading in shocks from solution fi les of the baseline simulation for a given year has also been automated.

The essential ingredients of dynamic simulations using RunDynam are:

The executable fi le or its equivalent that runs the model • The various database and sets fi les for the initial year • A valid closure for the baseline and a separate policy closure, and any year-to-• year variations that might refl ect the policy scenario, such as exogenous invest-ment and outputs associated with the specifi c policy A ‘DIN’ fi le that points to the fi les dealing with lagged relationships in policy • and forecast Specifi cation of the number of periods (and period length) in the dynamic • simulation

In dynamic TERM models, a combination of the methodology outlined in Sect. 3.5 and RunDynam makes the task of preparing a new aggregation of the data-base with a dynamic baseline and policy closure relatively straightforward. Consistent with the development of GEMPACK and related software over several decades, RunDynam has arisen out of Ken Pearson’s vision to make complex computing tasks available to practitioners without specialist computing skills (Horridge and Pearson 2011 ). 7

One of the objectives of software upgrades to GEMPACK and RunDynam is to allow CGE practitioners more time to analyse their results and explain them in terms of the ascribed shocks, theory, data and closure of the model. RunDynam includes a built-in version of AnalyseGE, designed for this task ( Pearson et al. 2010 ) .

A recent enhancement to RunDynam is the conversion of the time-series solution fi le to a form readable in VIEWHAR, another GEMPACK facility that provides ready access to multi-dimensional data arrays. 8 This has a particular advantage in a multi-regional dynamic CGE model because multi-regional solution variables often entail three or more dimensions. That is, region and time are two dimensions: we can view a scalar variable in the regional and time dimension at once, but not vector variables or matrices. A VIEWHAR-readable version of the dynamic solution pro-vides us with a convenient way of viewing slices of vector or matrix variables across regions and time. Moreover, we can write TABLO-based programs to process the VIEWHAR-readable solution fi le. An example of this concerns a study undertaken for the Murray-Darling Basin Authority (Wittwer 2010 ) . The client requested that

6 See http://www.monash.edu.au/policy/gprdyn.htm 7 Mark Horridge and more recently Michael Jerie have played major roles in ongoing development and upgrades to GEMPACK software. Earlier contributors include George Codsi and Jill Harrison. 8 See http://www.monash.edu.au/policy/gpwingem.htm

http://www.monash.edu.au/policy/gprdyn.htm

http://www.monash.edu.au/policy/gpwingem.htm


regional results be reported by natural resource management (NRM) regions. This required several steps. First, a top-down module was added to TERM-H2O to pro-vide regional impacts at the statistical local area (SLA) level. 9 A TABLO-based program then used the VIEWHAR-readable solution fi le and a mapping from SLAs to NRMs to generate regional solutions for the latter.

References

ABS (Australian Bureau of Statistics) (2007a) Taxation revenue, Australia, 2005-06. ABS, Canberra. http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/5506.02005-06?OpenDocument . Accessed 8 Feb 2011

ABS (Australian Bureau of Statistics) (2007b) Government fi nancial statistics, Australia, 2005-06. ABS, Canberra. http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/5512.02005-06 ? OpenDocument. Accessed 8 Feb 2011

Adams P, Horridge M, Wittwer G (2003) MMRF-GREEN: a dynamic multi-regional applied gen-eral equilibrium model of the Australian Economy. Based on the MMR and MONASH models. Centre of Policy Studies working paper. http://www.monash.edu.au/policy/ftp/workpapr/g-140.pdf . Accessed 8 Feb 2011

Dixon P, Rimmer M (2002) Dynamic general equilibrium modelling for forecasting and policy: a practical guide and documentation of MONASH. North-Holland, Amsterdam

Dixon P, Menon J, Rimmer M (2000) Changes in technology and preferences: a general equilib-rium explanation of rapid growth in trade. Aust Econ Pap 39:33–55

Dixon P, Lee B, Muehlenbeck T, Rimmer M, Rose A, Verikios G (2010) Effects on the U.S. of an H1N1 epidemic: analysis with a quarterly CGE model. J Homel Secur Emerg Manage 7, article 75

Horridge M, Pearson K (2011) Solution software for CGE modelling. Centre of Policy Studies working paper. http://www.monash.edu.au/policy/ftp/workpapr/g-214.pdf . Accessed 24 Mar 2011

Malakellis M (2000) Integrated macro-micro-modelling under rational expectations: with an appli-cation to tariff reform in Australia, Contributions to economics. Physica-Verlag, Heidelberg

McKibbin W, Wilcoxen P (1999) The theoretical and empirical structure of the G-Cubed model. Econ Model 16:123–148

Pearson K, Hertel T, Horridge M (2010) AnalyseGE: software assisting modellers in the analysis of their results. http://www.monash.edu.au/policy/ftp/ange/phh.pdf . http://www.monash.edu.au/policy/ftp/workpapr/g-214.pdf . Accessed 24 Mar 2011

Pricewaterhouse Coopers (2007) Economic analyses of the port of Melbourne. http://www.dtf.vic.gov.au/CA25713E0002EF43/WebObj/PortofMelbourne-EconomicAnalyses/$File/Port%20of%20Melbourne%20-%20Economic%20Analyses.pdf . Accessed 8 Feb 2011

Thompson J (2010) Gillard reveals carbon price scheme. http://www.abc.net.au/news/stories/2011/07/10/3265732.htm . Accessed 11 July 2011

Verikios G, McCaw J, McVernon J, Harris A (2012) H1N1 infl uenza and the Australian macro-economy. J Asia Pacifi c Econ 17:22–51

Verikios G, Sullivan M, Stojanovski P, Giesecke J, Woo G (2011) The global economic effects of Pandemic Infl uenza. Centre of Policy Studies working paper. http://www.monash.edu.au/ policy/ftp/workpapr/g-224.pdf . Accessed 21 Oct 2011

9 ABS prepares census data at the SLA level (1400+ regions). These aggregate to 200+ statistical sub-divisions (SSDs): TERM-H2O represents SSDs in the Murray-Darling Basin with relatively aggregated regions elsewhere.

http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/5506.02005-06?OpenDocument

http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/5512.02005-06




http://www.monash.edu.au/policy/ftp/ange/phh.pdf



http://www.dtf.vic.gov.au/CA25713E0002EF43/WebObj/PortofMelbourne-EconomicAnalyses/$File/Port%20of%20Melbourne%20-%20Economic%20Analyses.pdf



http://www.abc.net.au/news/stories/2011/07/10/3265732.htm





Wittwer G (2010) The regional economic impacts of sustainable diversion limits. http://www.mdba.gov.au/fi les/bp-kid/1472-regional-economic-impacts-wittwer.pdf . Accessed 18 Feb 2011

Wittwer G, Vere D, Jones R, Griffi th G (2005a) Dynamic general equilibrium analysis of improved weed management in Australia’s winter cropping systems. Aust J Agric Resour Econ 49:363–377

Wittwer G, McKirdy S, Wilson R (2005b) The regional economic impacts of a plant disease incur-sion using a general equilibrium approach. Aust J Agric Resour Econ 49:75–89

Wittwer G, McKirdy S, Wilson R (2006) Analysing a hypothetical Pierce’s disease outbreak in South Australia using a dynamic CGE approach. Centre of Policy Studies working paper. http://www.monash.edu.au/policy/ftp/workpapr/g-162.pdf . Accessed 8 Feb 2011

http://www.mdba.gov.au/files/bp-kid/1472-regional-economic-impacts-wittwer.pdf



Part II Water Modeling

59

Abstract Australian economists were modeling irrigation water scenarios many years before substantial reforms started taking effect. Early modelers recommended that the water authorities raise the price of irrigation water. A recurring theme of later modeling is that water trading plays an important role in improving allocative effi ciency. The eventual COAG reforms included the separation of land and water titles. With this separation, water trading became easier. With such trading, the mar-ket rather than authorities determine the price of water. Modeling has shown that other institutional aspects of water management have also hindered optimal water allocations in the past.

Keywords COAG reforms • Factor mobility • Institutional constraints • Partial v. general equilibrium modeling • Water trading

4.1 Issues and Modeling of Water Resources in Australia

Australia has the driest and most variable rainfall of any inhabited continent, and water has long been perceived as a constraint to development. Initially, development of water resources, especially for irrigation, was based not so much on economic considerations but to achieve social and political objectives, such as ‘development of the nation’ (Quiggin 2001 ; Watson 2003 ) .

Gradually, however, problems emerged: the bodies responsible for storing and distributing water were not covering costs; use of water for irrigation was increasing rapidly, to the detriment of the environment; and ineffi ciencies arising from the existing allocation of water both within agriculture and between agriculture and

M. Griffi th (*) Centre of Policy Studies , Monash University , Clayton , VIC , Australia e-mail: marnie.griffi [email protected]

Chapter 4 Water Resources Modeling: A Review

Marnie Griffi th

G. Wittwer (ed.), Economic Modeling of Water: The Australian CGE Experience, Global Issues in Water Policy 3, DOI 10.1007/978-94-007-2876-9_4, © Springer Science+Business Media Dordrecht 2012

60 M. Griffi th

other uses were becoming increasingly evident. According to Edwards ( 2003 ) , a turning point in the debate was a publication by Davidson ( 1969 ) , which questioned the benefi t of large-scale irrigation works. Another seminal paper was Randall ( 1981 ) , which characterised Australia as transitioning from a ‘developing’ to a ‘mature’ water economy. This transition occurs as attention turns from developing water resources and associated infrastructure to meet an unchecked demand to real-locating existing supplies.

Over the past few decades, Australia has embraced rapid and broad-based reform of management of water resources, most notably through the landmark Council of Australian Governments’ (COAG) water reform program (COAG 1994 ) and its update in the National Water Initiative (COAG 2004 ) . Policy and related modeling work, often based in the southern Murray-Darling Basin, have focused on pricing access to the storage and distribution infrastructure. It is diffi cult to determine appropriate prices for water, but it was widely acknowledged that pre-reform prices were too low and also lacking in the inter-regional consistency necessary to avoid distortions in allocation (Hall et al. 1993 ) .

Modeling has also examined the implications of limiting water use, particularly for irrigation. In 1995, the Murray-Darling Basin Ministerial Council placed a ‘Cap’ on water use in the Basin, restricting any future diversions to levels commensurate with a 1993–94 level of irrigation development. While Randall ( 1981 ) noted the maturation of the Australian water economy, for Brennan and Scoccimarro ( 1999 ) , the Cap was the true signal that Australia had entered a mature water economy phase. Engineering-based models such as the Goulburn Simulation Model (built in the REALM software) are used to monitor adherence to the Cap (Perera et al. 2003 ) as well as hydrological implications of other policy changes.

Reallocation of water to higher valued uses has been another modeling issue. Initially, focus was on a reallocation within irrigation, but more recently, work has included reallocating to competing uses such as the environment and urban demands (Young et al. 2006 ) . Whereas the Cap restrained the potential for further increases in diversions, recent times have seen a push to reallocate a much more generous amount of water to the environment, starting from the Living Murray initiative of 2004 and reiterated in the Water for the Future initiative. 1

Modeling has examined the potential of water markets to achieve a more effi cient distribution of water. The years following the 1994 COAG package saw a focus on developing models to predict the direction of trade, the traded price of water, and the gains from trade. There are a number of potential third party impacts arising from water trade, often linked to the diffi culty in assigning complete property rights to water. These are summarised by Heaney et al. ( 2006 ) as impacts on reliability of irrigation water supply, timeliness of irrigation water delivery, storage and delivery charges, and water quality. Models have been developed to estimate the costs of these externalities and to analyse the effectiveness of remedies.

1 See http://www.environment.gov.au/water/policy-programs/lmi/index.html and http://www. environ-ment.gov.au/water/australia/index.html

http://www.environment.gov.au/water/policy-programs/lmi/index.html

http://www.environment.gov.au/water/australia/index.html

http://www.environment.gov.au/water/australia/index.html

614 Water Resources Modeling: A Review

Ongoing reforms of property rights have also concerned analysts. Initially, the focus was on separating water rights from land title and formalising property rights so that they are clearly specifi ed in terms of volume, reliability, transferabil-ity, and quality. More recently, property right reform has expanded to encom-pass rights over time. A particular area of interest has been the capacity sharing concept for dams and also distribution channels (Dudley and Musgrave 1988 ; Hughes 2010 ) .

Modeling has also examined the impact of water reforms on regional communities. While there was political and academic support for reforms, concerns were raised over impacts on irrigators and local communities (NRE and NSW Agriculture 1999 ) .

Finally, variability and uncertainty in water supplies and the impacts of climate change are modeling issues. The focus on potential climate change impacts on the water resources of the Murray-Darling Basin is relatively recent – Adamson et al. ( 2009 ) note that the 1994 COAG reform documents do not consider potential impacts of climate change.

4.2 Pre-COAG Modeling

Economic modeling in the 1970s and 1980s demonstrated that the existing allocation of irrigation water in Australia was highly ineffi cient. There were isolated instances of water trading in the 1960s and 1970s (as a response to drought) and a general ability to trade allocations from 1983–84 in NSW and 1987–88 in Victoria (Alaouze 1991 ) . However, prior to the 1994 COAG reform agenda, which formalised a com-mitment to limiting water availability and encouraging water markets, modeling efforts tended to assume that ineffi ciencies in water allocation would be solved via set water prices rather than through markets.

In one of the fi rst Australian modeling exercises, Guise and Flinn ( 1970 ) developed a spatial equilibrium model to investigate the optimal allocation of water across seasons and between fi ve irrigation areas and a hydroelectric use in the Yanco irrigation area, part of the Murrumbidgee Irrigation Area (see Fig. 6.4 ). They found that the allocation of water at that time was far from optimal due to underpricing.

Briggs Clark et al. ( 1986 ) modeled the response of irrigators in the Murrumbidgee and Coleambally irrigation areas to changes in water prices, to analyse the potential for pricing to achieve a more effi cient allocation of water. The two regions were both split further by farm size, to give four areas in total. An optimisation model was formulated to maximise regional gross margins, subject to constraints including land type, water, labour, feed, irrigation channel capacity, and crop rotations. Various levels and timings of irrigation were considered – thus farmers were able to reduce water use by either changing crop areas or reducing water applications.

They found, for the range of prices assessed, that the short-run demand for water is inelastic. Price increases were not likely to lead to substantial changes in cropping areas or overall water use, though there was some switching from irrigation to

62 M. Griffi th

dryland pastures, and some reduction in irrigation applications, lowering yields. Despite that, the effects on regional gross margins were signifi cant (a 40% decline with an increase in water charges from $8/ML to $28/ML).

In retrospect, the top end price considered by Briggs Clark et al. ( 1986 ) in deriving a demand curve seems remarkably low, even after adjusting for infl ation. But this is an administrative price rather than a market price – the modeling aimed to assess the ability of higher water charges to change water demand; transferable water rights were given only passing mention. Temporary water market prices have since topped $1,000/ML in times of extreme scarcity, but there is no way this could have taken the form of an administrative price hike. In addition, the environment for water allocation in this period was one of comparatively plentiful water. The water scarcity starting from just over a decade later, caused by drought and also further irrigation diversions, was simply not anticipated at this time.

By the late 1980s, there was a growing sense of urgency about the severity and extent of salinity in Australia, both dryland and irrigation. Quiggin ( 1988 ) modeled how different institutional structures (property rights) impact on farm land use decisions and hence salinity outcomes. Farm land in six regions of the Murray River system was allocated between four irrigated crops (stone fruits, citrus, grapes, and pastures) and a dryland activity. Downstream salinity and water availability were determined by upstream decisions. In the fi rst scenario, each of the blocks optimised privately; the second involved a system-wide optimum. The global optimisation sce-nario was implemented as a dynamic program, with downstream regions taking the place of later time periods usually seen in dynamic programming. Global optimising involved lower levels of water use and salinity, but higher total profi ts (by almost one-third), though this involved a transfer in income from upstream areas to downstream areas. These modeled costs of salinity were higher than other contemporaneous esti-mates, which focused on the cost at existing production patterns, implying that oppor-tunity costs of salinity may be greater than visible costs involving yield reductions.

Trade allows water to be distributed effi ciently in a spatial sense in the current season, but due to highly variable seasonal conditions, true effi ciency in water allocation must also allow farmers to allocate water inter-temporally. Traditionally, Victoria has had a conservative policy for allocating water in reservoirs inter- temporally. Prior to 2007, all available water was allocated for use up to the point of 100% of the water right; after that, no more water was allocated until the following year’s water right was also ensured (initially assuming average future infl ows and, later, the very lowest infl ows). Any water available after that was allocated to the current season as ‘sales’ water. Until recently, Victorian irrigators were not able to carry over water to the following irrigation season, and certainly not able to draw down from expected future allocations. 2

2 Water allocation policies in Victoria have changed substantially since. The water right and sales system was converted to high-reliability water shares and low-reliability water shares from 1 July 2007 (also unbundled were a delivery share and a water use licence). Carry over has been permitted in some form since 2006–2007.


Alaouze ( 1991 ) developed a two-period model to derive optimal inter-temporal water management when Victorian irrigators are allowed to carry over water to the next irrigation year or overdraw water from the next irrigation year. This proposal was designed to provide a low-cost alternative to the capacity sharing concept, which has high information costs. Farmers can choose to carry over up to their water right plus sales to the next period but may only draw forward up to the water right. The decision is taken subject to uncertainty in second-period commodity prices and second-period sales water allocation.

The optimal inter-temporal allocation of water depends on storage losses, the interest (discount) rate, water entitlement, fi rst-period output price, fi rst-period soil moisture, and fi rst-period sales allocation. Factors associated with drought bring forward consumption of water – that is, irrigators would choose to use the overdraw option. The author concluded that optimal inter-temporal distribution was unlikely to match the centrally determined allocation policy that Victoria followed and that allowing inter-temporal transfers of water would improve profi tability, especially in times of drought. Research in a similar vein was continued by Brennan ( 2006, 2008 ) , described below.

4.3 Modeling Water Markets: Benefi ts and Direction of Trade

Following the COAG reform package in the mid-1990s, water markets have been the focus of economic modeling of water resources. Initially, there was much interest in predicting the direction and benefi ts of trade. The other major compo-nents of the COAG package were a move to full cost-recovery prices and sustain-able water use, involving some reallocation of water from irrigation to the environment.

In one of the fi rst modeling exercises of water trading in Australia, Hall et al. ( 1993 ) formulated a model of irrigated agriculture of 18 regions in the southern Murray-Darling Basin to estimate the effects of water pricing and, in particular, water trading. In addition, the impacts of a ‘water bank’, buying water for environ-mental or urban purposes, were assessed. The model is a short-run annual spatial equilibrium model, built around a directed network representation of the river system, which tracks the implications of various policies for river salinity, as well as farm income and water use. Land use is constrained by irrigation land availability, crop rotations, and regional feed pools, as well as by water.

One of their principal fi ndings was an increase in total farm gross margins of around 4.6% as a result of the introduction of trade. They noted that this gain may be higher in the long run as more adjustment opportunities are available and as other pressures come to bear. These pressures might include worsening salinity and deterio-rating storage and delivery infrastructure. The authors also found that increasing water delivery charges from the then-current charges to charges estimated to cover costs would have little impact on water use, as the higher charges still lie below opportunity cost of water use.

64 M. Griffi th

The Victorian Department of Primary Industries and New South Wales Agriculture, and later Primary Industries and Resources South Australia, also devel-oped short-run spatial equilibrium models through the late 1990s and the early 2000s (Eigenraam et al . 2006 ) . A joint application was applied to 13 Victorian regions and 9 NSW regions of the southern Murray-Darling Basin to analyse likely impacts on irrigators of pricing reform and environmental fl ows through lower irrigation water availability in NSW (NRE and NSW Agriculture 1999 ) . The impact of higher water charges is likely to be minimal due to inelastic demand and a low-cost share, coupled with the marginal value products of water in different uses remaining signifi cantly above base prices. Environmental fl ows, on the other hand, are likely to result in marked reductions in irrigation output. While inter-state water markets offer few advantages when water is at historical average levels, they are an effective means of minimising costs of water scarcity due to, for example, environmental fl ows.

Eigenraam et al . ( 2003 ) assessed the effects on different regions and sectors of the southern Murray-Darling Basin of returning water to the environment. They found relatively large costs associated with these reductions in water available for irrigation. However, there was no report of the partly offsetting impacts of factor movements towards dryland farming, which are likely in such a scenario.

To summarise, early modeling efforts indicated that water charges set by the water authorities were far below opportunity cost of water use. Any foreseeable increase was not likely to have a signifi cant impact on water allocation. These fi nd-ings pointed to the introduction of water markets as a means of making better use of water. However, there are many potential problems which impede the effi cient dis-tribution of water via a water market. While initial work concentrated on predicting the potential reallocation of water due to trade, and the gains that this would entail, attention quickly turned to also considering the problems that would have to be overcome to implement water markets successfully (Brennan 2008 ) .

4.4 Modeling Water Markets: Externalities and Institutional Constraints

There are many externalities involved with water use which are not captured by water rights. Heaney et al. ( 2006 ) grouped water market-related externalities as follows: reliability of supply, arising, for example, in changes to return fl ows and conveyance losses; timeliness of delivery, arising from congestion of infrastructure; storage and delivery charges, which in the extreme case lead to stranded assets; and water quality, most notably salinity. Modeling literature has aimed to assess how important these externalities are in the context of the gains from trade and the effectiveness of policies to remedy the situation.

Changing the spatial and temporal pattern of water demand, as occurs with water trade, can affect the reliability of supply and timeliness of delivery for other water users in various ways. For example, in the case of open channel, it is


often the case that the further the water has to be carried, the larger the evaporative losses. This implies that more water is involved in delivering a given quantity of allocation, impacting on general security of supply. Another example is differences in return fl ows. If, for example, a fl ood irrigator with high return fl ows sells water to a drip irrigator with low return fl ows, downstream users may experience lower water availability. When trade was fi rst introduced, previously dormant rights 3 suddenly acquired value, and the result was an increase in water use without any new rights or licences being issued – mostly confi ned to NSW. This imposed a one-off externality on other irrigators.

In the case of timeliness of deliveries, if trade moves water into an area in which deliveries approach channel capacity constraints, existing irrigators may not receive water when required. This could also occur with trade within a district if the buyer calls on water in peak times and the seller ordered water in off-peak times. Assessing the impacts of water markets on security of supply and delivery of water requires a model capable of simulating the system in much more detail through time and space than is typically considered in a standard economic model. For example, most eco-nomic models are annual or, at best, seasonal, yet congestion in the delivery system tends to occur in peak demand periods lasting a few days or a week.

Engineering-based models of water allocation tend to include this detail. The most commonly used models of this type in Australia are REALM (Victoria and South Australia), IQQM (NSW and Queensland), and the MSM-BIGMOD suite, from the Murray-Darling Basin Authority. In these models, the focus shifts from the ‘demand’ side of the system, focused on the value of water use across sectors and regions, to the supply and (physical) allocation domains of the system.

The Goulburn Simulation Model (GSM) is an application of the general REALM framework to the Goulburn system in Northern Victoria (Perera et al. 2003 ) . Its purpose is to simulate the performance of the storage and delivery system in terms of deliveries to key demands (mostly irrigation), storage volumes, and compliance with other targets such as river fl ows (Zaman et al. 2009 ) .

The basic elements are nodes and carriers: nodes include both supply (mostly reservoirs) and demand (irrigation and urban) centres, while the carriers (e.g., rivers) connect the nodes. There are 20 storages and 58 demand nodes in the GSM. The system is simulated via a linear program which minimises the cost of delivering water given supplies, demands, and the capacity constraints of carriers. Cost is imposed by ‘penalties’ associated with the use of a carrier.

Irrigation demands are based primarily on rainfall and evapotranspiration, transformed into node-level irrigation demands via another model, the Program for Regional Demand Estimation (PRIDE) (SKM 1998 ) . Irrigation demands are also modifi ed to refl ect historical usage. Water allocation volumes and capacity in the delivery system to transport the water determine whether these irrigation demands can be met.

3 That is, rights to water which had not previously been used.

66 M. Griffi th

The GSM is usually run on a monthly time step from 1891–92 to the present. Simulations give an indication of how much allocations and deliveries might vary for a full range of hydroclimatic conditions. Simulations are long run in the sense that volumes in the storages carry over to the next year; they are short run in the sense that node crop areas and permanent water entitlements are fi xed.

Zaman et al. ( 2009 ) extended the GSM with a temporary water trade model. Monthly estimates of the total traded quantity and pool price are derived econometri-cally from historical trade data, starting from 1994–95. Their modeling framework is able to estimate the impact that the physical delivery infrastructure and operational rules have on system security and the ability for water to trade as desired. They fi nd that trade does have an impact on system security, with allocations during drought likely to be lower than otherwise as a result of trade. They also fi nd that trade shifts the location of bottlenecks in the delivery system, with improvements in timeliness of deliveries for some irrigators and deteriorations for others. The modeling also has longer-term implications for planning infrastructure renewal.

One solution to this might be to defi ne rights over the storage and distribution infrastructure, so that irrigators own an exclusive right to access a share of the storage and delivery capacity, defi ned over some period of time. To demonstrate the costs associated with centralised decisions regarding the operation of storages in Victoria, Brennan ( 2008 ) ran three simulations: the fi rst uses the current storage rules, described in the context of the Alaouze ( 1991 ) paper above, and no trade from the Goulburn system to the NSW Murray; the second uses the current storage rules and allows trade, that is, spatial but not temporal equilibrium; and the third scenario also allows irrigators to make their own storage decisions, that is, temporal plus spatial equilibrium. Goulburn system irrigators lose as a result of trade with the centralised storage system, as the impacts of lower reliability due to increased uptake outweigh profi ts from trade. Earlier, the same author concluded that improving property rights to enhance farmer management of inter-temporal risks associated with water availability, should be resolved before the enactment of other policies, such as environmental fl ows or expansions in the spatial scope of trade (Brennan 2006 ) .

The provision of access to the water storage and delivery infrastructure is a joint decision. When one irrigator decides to sell his or her entitlement to water, and hence no longer contributes to the upkeep of the infrastructure, charges to all remain-ing irrigators must increase to cover costs. An increase in costs associated with water use makes it more likely that other irrigators in the area will also sell their entitlement. In the extreme case, it is no longer economic to provide water to the remaining irrigators, leading to the situation known as stranded assets.

Heaney et al. ( 2004 ) developed a model to assess how different charging regimes impact on the effi ciency of trade. Economic welfare is measured via water and fi xed factor rents. Activities include irrigated crops, irrigated pasture, and horticulture, which all compete for water, and dryland crops and pastures which compete with the irrigated crops for land. The model incorporates detailed consideration of cost of delivery. They compared the introduction of inter-regional trade with two charge-recovery mechanisms: in the fi rst, fi xed charges are increased to offset lower


revenue in areas losing water; in the second, variable charges are increased to offset this loss. They found that the fi xed charge revenue recovery route leads to a welfare gain, but relying on changing variable charges to even out revenues leads to lower basin welfare.

In terms of the impact of water markets on water quality, salinity has been the most prominent issue, though waste and nutrients are also of concern. Irrigation can raise saline groundwater tables. The impacts may be localised to the catchment but may also impose broader-scale externalities if water with relatively high salt concen-tration enters the river system (as a return fl ow), increasing the salinity of water for downstream users. This is exacerbated in situations of low fl ow, as there is less water to dilute the salt entering the river system.

In recent times, there has been a decrease in the sense of urgency associated with the salinity problem in Australia. This might be due to an initial exaggeration in the degree of the problem but might also be related to the very dry conditions through the 2000s (Pannell and Roberts 2010 ) .

Ultimately, Heaney et al. ( 2006 ) argue that although third-party effects to water markets may exist, (a) it may cost more to address them than leave them, and (b) it does not preclude pursuing market reforms as the costs may be minor relative to the gains from trade.

Some commentators have suggested that there are other reasons water markets are not effi cient allocators of water, including institutional constraints on trade and high transaction costs. However, Brennan ( 2006 ) found that differing regional prices for water refl ect hydrological rather than institutional constraints and argued that water markets in Northern Victoria are effi cient.

4.5 Recent Advances in Water Allocation Modeling

Dramatic changes in the environment for water allocation have occurred over the past two decades. In order to undertake policy-relevant work, modelers require a good understanding of the behavioural responses of farmers to changes in the operating environment (Adamson et al. 2007 ; Appels et al. 2004 ) . When the COAG reforms were developed, for example, Goulburn-Murray Water in Victoria’s north-east aimed to allocate 100% of water entitlements in 97 years out of 100. Before 2002–03, allocations had never fallen below 100%. Since 2002–03, water allocations have fallen below 100% in 5 of 8 years due to drought. This has forced changes in irrigating industries beyond what was conceived at the time: for example, dairy industries once reliant on irrigated perennial pastures have turned to purchasing feed requirements, and the rice industry declined in drought to the extent that the Deniliquin rice mill closed for 3 years. Constraints embodied in the short-run models developed to support policy through the 1990s, such as limiting changes to crop areas and assumptions of fi xed capital, are now less appropriate.

68 M. Griffi th

Although ‘sustainable’ use of water has been on the policy agenda for decades, the solidifi cation of plans via CSIRO’s work on sustainable diversion limits, and the release of the Guide to the Murray-Darling Basin (MDBA 2010 ) plan in late 2010, involving relatively large volumes of water, has meant that the trade-off between water for irrigation and water for the environment has become highly topical. In addition, the Commonwealth government has been pursuing a policy of buying back water entitlements for the environment since 2007–08. The water is managed by the Commonwealth Environmental Water Holder, and as of 1 July 2011, hold-ings amounted to just over 1,000 GL of water of varying degrees of security. Given that total Basin entitlements are around 14,000 GL, the Commonwealth is now a signifi cant holder of water entitlements.

In 2003, as part of the fi rst step in restoring the River Murray, the Commonwealth, New South Wales, Victorian, South Australian, and ACT governments pledged funds to recover 500 GL of water for environmental purposes. Qureshi et al. ( 2007 ) assess how the impact of this water purchase varies according to the mechanisms used and seasonal circumstances. They also track the implications for salinity. Their modeling framework includes an explicit consideration of differing seasonal condi-tions (allocations and effective rainfall); a crop water production function, which allows some defi cit irrigation as a response to water scarcity; and fl exibility in areas of annual crops in response to seasonal conditions. The authors fi nd that if trade is limited, environmental fl ows are costly to implement. Costs could be reduced both by targeting acquisition from low-value areas and by structuring environmental pur-chases such that more water is bought in water-plentiful years and less in drier years. Freer trade increases total returns to agriculture even with the 500 GL buy-back. If trade is free, it makes relatively little difference whether the water is pur-chased pro rata or targeted. There is a net increase in salinity costs as a result of trade, but these costs are modest relative to the gains from trade.

In the Murray-Darling Basin plan of late 2010 (MDBA 2010 ) , the Murray-Darling Basin Authority proposed returning 3,000–4,000 GL of water per year to the Basin for environmental purposes. Grafton and Jiang ( 2011 ) model the economic impacts on Murray-Darling Basin irrigators of reductions in water availability ranging from 3,000 to 4,400 GL, based on allocations for a ‘normal’ year, 2000–01. They also model impacts in a dry year, based on 2005–06. In a normal year, reducing irrigation water availability across the Basin by 3,000 GL results in a fall in profi ts of 10%; returning 4,000 GL implies a 17% fall in profi ts. Impacts vary signifi cantly by region, with the Murrumbidgee and the Murray regions the hardest hit, with reduced water use in the order of 69–73% and 37–85%, respectively. 4

Brennan ( 2006 ) develops a modeling framework which ties short-run decisions to the long-run environment. The model is applied to Northern Victoria and considers

4 In Chap. 6 , these two regions experience substantial terms-of-trade gains from buybacks, even without compensation. The differences arise because Grafton and Jiang ( 2011 ) report only production-based income in these two regions: net water sales are an additional income source for farmers not reported in the study.


three industries, horticulture, dairying, and annual crops, each with variable capital requirements and profi ts. Optimal investment in each industry depends on the shape of the reservoir yield density function. This is partly determined by policy, for example, environmental fl ow regulations, and the relative costs and returns of each industry. Two sources of intra-seasonal risk are considered: rainfall on the farm, affecting irrigation water requirements, and rainfall in the catchment, affecting irrigation water availability.

A factor which impedes the ability of models to capture real-life behaviour is the degree of uncertainty surrounding water. A group at the University of Queensland has developed models based on state-contingent theory, able to explicitly consider the substantial degree of uncertainty involved with water. Their aim is to provide insight into behavioural responses to changes in climate or policy. Adamson et al. ( 2007 ) model the Murray-Darling Basin as a directed network. They consider three states of nature: average precipitation, occurring with a probability of 0.5; dry, in which infl ows to all 18 catchments are reduced by 40% relative to average, occurring with a probability of 0.2; and wet, in which infl ows to all catchments are increased by 20%, occurring with a probability of 0.3. Two optimisation concepts are simulated: (a) the system is optimised sequentially as water moves downstream and (b) whole-of-basin welfare is maximised. In the global solution relative to the sequential solution, area irrigated and water use are lower (replaced by dryland activities); water use varies more with state of nature, and salinity is signifi cantly lower.

Extending the above, Adamson et al. ( 2009 ) look at the potential impact of climate change on the uncertain nature of drought and the implications of this for irrigation industries in the Murray-Darling Basin. Climate change is likely to be associated with more frequent and severe drought in the basin. Two scenarios for climate change are simulated: a uniform downward shift in the distribution of infl ows and a change in the probability distribution of infl ows towards more fre-quent drought: relative to above, the probability of the wet state is decreased to 0.1 and the dry state is increased to 0.4. More frequent drought has more severe impacts on social welfare than the proportional reduction case. Water markets, as encapsu-lated in the global as opposed to the sequential solution, might offer some means of reducing the impacts of climate change.

Connor et al. ( 2009 ) assess potential impacts of climate change-driven reduc-tions in water availability for irrigation in the lower Murray. They use a two-stage optimisation approach where both long-term capital investment decisions and short-run adjustments to water conditions are modeled. The study includes three climate change scenarios, ranging from mild to severe. The model includes four states of nature for water allocations, with the probability of being in each state varying with the scenario. Two types of water yield trade-off are considered: a trade-off in which impacts are limited to the current season and one in which yields will be reduced beyond the current season if a given water threshold is breached. They fi nd that mild climate change tends to see increased defi cit irrigation and investment in more effi cient irrigation technology; severe climate change sees a shift from perennial to annual crops, which have lower costs in the case of severe water allocation shortfalls.

70 M. Griffi th

4.6 Findings from Partial Models

In summary, partial models of water resource pricing and allocation abound in Australia and overseas. Many of these models deal with the reallocation of water within agriculture. Others extend to urban, environmental, and other uses of water. Much of the modeling work in Australia has focused on water markets (Adamson et al. 2007 ) . Research has found substantial gains from the introduction and broad-ening of water markets, and markets have been noted to be an effective way to miti-gate the effects of lower water availability due to reclaiming of water for environmental fl ows or climate change (Adamson et al. 2009 ; Eigenraam 1999 ) .

In partial modeling work, the main variables of interest are regional farm income/gross margins, water use, crop areas/product, and sometimes related variables such as salinity. Welfare implications are necessarily judged in terms of these variables, rather than broader quantities such as total regional consumption, income, or employment. To estimate the impact that reforms to the water sector might have on regional economic welfare and even the broader national welfare, a general equilibrium approach is required.

4.7 CGE Modeling of Water Resources

Indirect effects of a policy change can be signifi cant and are very diffi cult to predict analytically, even in sign (Roe et al. 2005 ) . A partial economic model is unable to provide this information, but a CGE model, which incorporates constraints and feedbacks between different economic sectors and agents, allows for a more com-plete welfare assessment. This may be particularly important in cases where there are political problems gaining acceptance, as is often the case in water resource management. For example, CGE studies of the reallocation of water from agricul-tural to urban uses have tended to fi nd a net welfare benefi t, but a negative outcome for rural communities (Goodman 2000 ) .

CGE models offer a number of advantages in modeling water resources. Policies and projects involving water tend to be relatively large scale, for example the con-struction of a new reservoir or a dramatic reduction in availability of water due to drought. Hence, they are likely to lead to changes in general prices and incomes (Berck et al. 1990 ) . The value of water changes substantially depending on time and place (Berck et al. 1990 ) : CGE models are able to capture this as they explicitly model equilibrium prices as water availability and demands change.

A number of water-focused CGE models have been developed, in varying degrees of scope and detail. Most of the models reviewed in this section were formulated to evaluate policy in the face of increasing water scarcity. Model scenarios include: a deliberate withdrawal of water to alleviate a drainage situation (Berck et al. 1990 ) , lower expected water supplies (Lofgren et al. 1996 ) , a redirection of water away from agriculture to recreational purposes (Seung et al. 1998, 2000 ) , the potential for water trading to improve allocative effi ciency (Diao et al. 2005 ) , and as an alternative


to more storages (Goodman 2000 ) or desalination plants (Gomez et al. 2004 ) , the interaction between groundwater and surface water systems (Diao et al. 2008 ) , the interaction between water and other economic policies (Hassan and Thurlow 2011 ; Roe et al. 2005 ) , potential impacts of climate change on water availability and demand (Calzadilla et al. 2010), and the potential for water use effi ciency improve-ments to alleviate scarcity (Calzadilla et al. 2011 ) .

Berck et al. ( 1990 ) model water withdrawal as a solution to the San Joaquin Valley’s salinity problem. Water is fi xed in aggregate, and requirements per acre are fi xed, but yields may change via varying labour inputs. Withdrawing water leads to a decrease in low-value irrigated cotton and grains, and an increase in livestock output as land is diverted to dryland pasture. The authors fi nd that a 50% cut in water availability leads to a dramatic fall in agricultural value added (almost 30%) but a reduction in Valley GDP of only around 3%.

Lofgren et al. ( 1996 ) report on an agriculture-focused CGE model of Egypt. They assess the scenario in which water availability declines by 20% between 1990 and 2020, while availability of land remains the same. This results in land rentals declining as the price of water rises. Although worsening water scarcity potentially may diminish agricultural production and employment, such losses may be offset by high enough levels of investment and productivity growth.

Seung et al. ( 1998 ) develop a CGE model of the Walker River Basin (north-western Nevada and north-eastern California) to evaluate the impacts of reallocat-ing water from irrigation to recreational values. Fifteen per cent of total existing irrigation diversions are moved to recreational uses, with direct impacts only on the irrigated hay sector. Landowners are compensated for the loss of water rights, and higher recreational spending is simulated via a random utility model, which is used to estimate how recreation-related trade and services expenditure increase with basin water levels. The authors fi nd that the compensation to water right holders and increased recreational expenditure are not suffi cient to offset the impacts of decreased hay production on the regional economy.

Seung et al. ( 2000 ) combine a dynamic CGE model (run over 6 years) with a recreation demand model to analyse temporal impacts of a reallocation of water from irrigation to wetland-based recreation in Churchill County, Nevada. This county is located in a desert in which dryland farming is not possible. The recreation demand model is based on a participation model which links use to acres of wetland and number of ducks. As with Seung et al. ( 1998 ) , they fi nd that the benefi ts of increased recreation are not suffi cient to offset reduced agricultural output.

Goodman ( 2000 ) looks at the potential for water transfers to provide a lower cost solution than new storages to growing demand for water in south- eastern Colorado. They fi nd that there are some relatively small benefi ts to increased storage for both urban and rural users, but that these are not enough to outweigh the costs involved. Transfers of water give greater benefi ts to both groups, at a much lower cost.

Answering a similar question, Gomez et al. ( 2004 ) apply a CGE model of the Balearic Islands with heightened hydrologic features to evaluate whether a water market may provide a more effi cient solution to water scarcity than building a water

72 M. Griffi th

desalination plant. Water sales increase agricultural income in times of drought. Moving resources from irrigated agriculture to non-irrigated agriculture also partly compensates for lower irrigated production. For the level of drought assessed here (40% of normal water availability), a water market would negate the need for a desalination plant. Desalination is an expensive option because of capital costs but also because it would lead to higher energy prices throughout the Balearic Islands economy.

Diao et al. ( 2005 ) model the economic gains resulting from the introduction of water markets in Morocco. The model is regional and includes seven water districts, divided further into perimeters (20 in total) within each district, refl ecting the diffi culty in transporting water between perimeters. There is substantial variability in the average product of water both within and between irrigation perimeters, suggest-ing potential gains from water trading. The introduction of water markets leads to an increase in water district GDP of around 8%, with an increase in fruit and vegetables, and some decline in grains and fodder. The water market price is higher than the average returns to water pre-reform in 16 out of 20 perimeters. There is a large cor-relation between returns to water and the percentage of total water supply traded.

Diao et al. ( 2008 ) analyse conjunctive ground and surface water use in Morocco, whereby surface and ground water are treated as two components of the same system. An urban water demand sector is also added. Morocco relies on groundwater for a high proportion of both drinking water and irrigation. Groundwater also acts as a stabilising buffer against more volatile surface water supplies. As most economic models of groundwater use have been partial, the benefi ts that groundwater plays in releasing surface water to other (e.g., urban) uses have been ignored.

Roe et al. ( 2005 ) argue that the macro-institutional setting (e.g., trade policies) will have an impact on the effectiveness of water policies. They build on Diao et al. ( 2005 ) , linking a farm-level model to the CGE model of Morocco. The farm model is stand-alone (or top-down) rather than fully integrated into the CGE model, so for example, price inputs to the micro- (farm-) level model from the CGE model are treated as exogenous. They show that economic reform outside of agriculture can affect the productivity of water in agriculture and that the sequencing of policy reforms can be important.

Letsoalo et al. ( 2007 ) use a CGE model of South Africa to assess the potential of higher water charges to bring about a triple dividend: a simultaneous combination of lower water use, higher economic growth, and a more even distribution of income. Water charge increases are those currently proposed by government, not necessarily optimal taxes. Because of the interest in poverty reduction, consumers are differen-tiated into a number of groups. Poverty reduction is measured as real consumption increasing in the three poorest household groups by race. Three potential means of recycling revenue are simulated: lower direct taxation on capital and labour, lower household sales tax, and lower sales tax on food. All scenarios provide a reduction in water use as the direct effect of higher water charges outweighs any indirect effects as a result of revenue recycling. To achieve an economic dividend, the water charge can be levied on any of the agricultural sectors and some mining sectors, including other mining and coal but excluding gold.


Berrittella et al. ( 2005a, b ) develop a water-focused CGE model at the global scale (GTAP-W). The authors note that as most water around the world is used for agricul-ture, and agriculture is highly traded, a global model will offer additional insights to the more usual river basin model. Since food and related markets such as textiles are highly traded, a global model is necessary to estimate the impacts on regions after adjustments in these global markets occur (Calzadilla et al. 2010 ) . The fi rst applica-tion uses GTAP-W to analyse the impact of reduced water supply in water-scarce countries (Berrittella et al. 2005a ) . The main fi ndings are global welfare falls with the reduction in water availability; there is an increase in virtual water imports in coun-tries experiencing a decline in water availability, though their overall trade balance may still improve; and industrial sectors (and hence countries strong in these) gain from lower water availability. In a second scenario examining a more severe drop in water availability in North Africa, there is a larger increase in water rents in that region, but also in other water-constrained regions. Global welfare falls sharply rela-tive to scenario one, although there are also regions which win relative to scenario one.

The second paper explores the impacts of higher water charges (Berrittella et al. 2005b ) . Despite water scarcity, water supply tends to be subsidised. The authors fi nd that an increase in water prices reduces water demand in many regions, though not in Western Europe, which is relatively insensitive to water prices and which produces and exports water-intensive products. In a variant in which only agricul-tural water is taxed rather than water use in all sectors, there is not a signifi cantly greater welfare loss. If water charges are increased only in water-scarce countries, welfare losses are reduced.

Calzadilla et al. ( 2011 ) use GTAP-W to estimate potential water savings and regional welfare implications of an improvement in water use effi ciency in the irriga-tion sector. They fi nd that the modeled improvements to irrigation effi ciency would have a marked impact on crop production, water use, and welfare. Adjustments in international markets are important, reducing water savings from between 12% and 21% per region down to 5–10% per region. They also fi nd that irrigation effi ciency changes comparative advantage across regions, and thus there are winners and losers in each scenario.

Hassan and Thurlow ( 2011 ) assess the effects of various water policies on water use and allocation, rural livelihoods, and GDP in South Africa. At the moment, water is distributed amongst and within 19 water management areas centrally rather than via markets. A more market-based method of water allocation is being debated. However major, micro-economic reforms in other areas are also ongoing, and a CGE framework is required to assess the net impacts. The model developed in this paper takes as a starting point the micro-macro feedback framework developed for Morocco in Roe et al. ( 2005 ) , but macro- and micro-components are fully integrated. They highlight the trade-offs between South Africa’s industrialization strategy and its goal of alleviating poverty and increasing employment. For example, water trading between irrigated sectors moves water to higher value crops. This raises agricultural output, thereby expanding exports and farm employment. Moving water from rural to urban regions, on the other hand, raises food costs, thereby adversely affecting the urban poor.

74 M. Griffi th

In summary, CGE models of water resources have offered many insights that partial models are not capable of providing. In small sub-national regions and nations alike, the database prepared for a CGE model provides an initial contribu-tion of irrigation sectors to the overall economy. Factor mobility within farm sectors reduces the impacts estimated from database weights as water availability falls.

There are important indirect effects associated with water policy reform. In some cases, this reinforces the gains from water reform, as in Diao et al. ( 2005 ) , where the direct gains from introducing a water market are heightened by higher real wages and consumption and an expansion of international trade. In other cases, indirect effects can offset and even reverse direct effects (e.g., Hassan and Thurlow 2011 ) , where changes to pricing and distribution of water lead to negative impacts on agri-culture and rural communities, but positive impacts for urban areas. Impacts are often reported in terms of factor prices but can also include impacts on other inter-mediate inputs. For example, in Gomez et al. ( 2004 ) , a desalination plant would increase energy prices through the economy.

Impacts of water reform depend on degree of factor mobility between sectors and regions. For example, in Seung et al. ( 1998 ) , the impacts of reallocating water from hay production to environmental uses are more severe with more fl exible capi-tal and labour. If water can be traded between regions and sectors, other farm factors follow, effectively making them more fl exible. Impacts also depend on the institu-tional environment and other policy reforms. For example, in Roe et al. ( 2005 ) , the introduction of water market reform without trade reform worsens existing distortions.

Policy formulation may entail complicated trade-offs. For example, Hassan and Thurlow ( 2011 ) identify water policies strategies which aid poor rural households but disadvantage poor urban households. Letsoalo et al. ( 2007 ) also fi nd that it is possible but diffi cult to identify policies which increase overall economic growth, benefi t the environment, and benefi t all types of poor households.

4.8 Conclusion

A major issue for modelers in moving to a CGE framework is to recognise the tensions between small-region representation, typical in depicting a river catchment region; the broader economy; and international conditions. Partial equilibrium models may cover one or several catchment regions. They have provided invaluable insights, particularly in the context of water market reforms. However, they are limited in the insights they can provide concerning the non-agricultural part of the economy in each region or interactions between irrigation regions and the rest of a national economy.

A multi-regional CGE model potentially may combine small-region detail with the feedback effects of a general equilibrium approach. Preparing such a model that includes water accounts requires theoretical and database modifi cations to a standard CGE model. In the case of TERM-H2O, Chap. 5 outlines these modifi ca-tions. In addition to models with a sub-national focus, there is considerable interest


in global water issues. Such interest has resulted in the development of GTAP-W. While it can model international conditions better than a multi-regional, sub-national CGE model, a model with an international focus necessarily lacks detail. As is evident in Chaps. 6 and 7 , a multi-regional (sub-national) model such as TERM-H2O is helpful in examining the regional economic implications of national policy.

References

Adamson D, Mallawaarachchi T, Quiggin J (2007) Water use and salinity in the Murray-Darling Basin: a state-contingent model. Aust J Agric Resour Econ 51:263–281

Adamson D, Mallawaarachchi T, Quiggin J (2009) Declining infl ows and more frequent droughts in the Murray-Darling Basin: climate change, impacts and adaption. Aust J Agric Resour Econ 53:345–366

Alaouze CM (1991) Intertemporal water transfers and drought. Aust Econ Pap 30:114–127 Appels D, Douglas R, Dwyer G (2004) Responsiveness of demand for irrigation water: a focus on

the southern Murray Darling Basin. Productivity Commission, Melbourne Berck P, Robinson S, Goldman G (1990) The use of computable general equilibrium models to

assess water policies, CUDARE working papers, Paper 545, University of California, Berkeley

Berrittella M, Hoekstra AY, Rehdanz K, Roson R, Tol R (2005a) The economic impact of restricted water supply: a computable general equilibrium analysis, research unit sustainability and climate change FNU-93. Centre for Marine and Climate Research, Hamburg

Berrittella M, Rehdanz K, Roson R, Tol R (2005b) The economic impact of water pricing: a com-putable general equilibrium analysis, research unit sustainability and climate change FNU-96. Centre for Marine and Climate Research, Hamburg

Brennan D (2006) Water policy reform in Australia: policy lessons from the Victorian seasonal water market. Aust J Agric Resour Econ 50:403–423

Brennan D (2008) Missing markets for storage and the potential economic costs of expanding the spatial scope of water trade. Aust J Agric Resour Econ 52:471–485

Brennan D, Scoccimarro M (1999) Issues in defi ning property rights to improve Australian water markets. Aust J Agric Resour Econ 43:69–89

Briggs Clark J, Menz K, Collins D, Firth R (1986) A model for determining the short term demand for irrigation water. Discussion paper 86.4, Bureau of Agricultural Economics, Canberra

Calzadilla A, Rehdanz K, Tol R (2011) Water scarcity and the impact of improved irrigation management: a computable general equilibrium analysis. Agric Econ 42:305–323

COAG (1994) Water reform framework. http://www.environment.gov.au/water/publications/action/pubs/policyframework.pdf . Accessed 18 Feb 2011

COAG (2004) Intergovernmental agreement on a national water initiative. Council of Australian Governments, Canberra

Connor J, Schwabe K, King D, Kaczan D, Kirby M (2009) Impacts of climate change on lower Murray irrigation. Aust J Agric Resour Econ 53:437–456

Davidson B (1969) Australia wet or dry? The Physical and Economic Limits to the Expansion of Irrigation. Melbourne University Press, Melbourne

Diao X, Roe T, Doukkali R (2005) Economy-wide gains from decentralized water allocation in a spatially heterogenous agricultural economy. Environ Dev Econ 10:249–269

Diao X, Dinar A, Roe T, Tsur Y (2008) A general equilibrium analysis of conjunctive ground and surface water use with an application to Morocco. Agric Econ 38(2):117–135

Dudley NJ, Musgrave WF (1988) Capacity sharing of water reservoirs. Water Resour Res 24(5):649–658

Edwards G (2003) Water policy: setting the scene. Aust Econ Rev 36:193–202

http://www.environment.gov.au/water/publications/action/pubs/policyframework.pdf


76 M. Griffi th

Eigenraam M (1999) Economic assessment of water market reform using the water policy model. Victorian Department of Natural Resources and the Environment, Melbourne

Eigenraam M, Crean J, Wimalasuriya R, Jayasuriya R (2003) Economic analysis of environmental fl ow scenarios in the Murray river system. NSW Agriculture and Victorian Department of Primary Industries, Orange and Melbourne

Eigenraam M, Wimalasuriya R, Stoneham G (2006) DPI water policy model. Economics and Policy Research Branch, Department of Primary Industries, Melbourne

Gomez CM, Tirado D, Rey-Maquieira J (2004) Water exchanges versus water works: insights from a computable general equilibrium model for the Balearic Islands. Water Resour Res 40:W10502. doi: 10.1029/2004WR003235

Goodman DJ (2000) More reservoirs or transfers? A computable general equilibrium analysis of projected water shortages in the Arkansas River Basin. J Agric Resour Econ 25:698–713

Grafton RQ, Jiang Q (2011) Economic effects of water recovery on irrigated agriculture in the Murray Darling Basin. Aust J Agric Resour Econ 55:487–499

Guise J, Flinn J (1970) The allocation and pricing of water in a river basin. Am J Agric Econ 52:411–421

Hall N, Poulter D, Curtotti R (1993) ABARE model of irrigation farming in the Southern Murray–Darling Basin. ABARE, Canberra

Hassan R, Thurlow J (2011) Macro-micro feedback links of water management in South Africa: CGE analyses of selected policy regimes. Agric Econ 42(2):235–247

Heaney A, Thorpe S, Klijn N, Beare S, Want S (2004) Water charges and interregional trade in the Southern Murray Darling Basin. In: Establishing Australian water markets symposium, Melbourne

Heaney A, Dwyer G, Beare S, Peterson D, Pechey L (2006) Third-party effects of water trading and potential policy responses. Aust J Agric Resour Econ 50(3):277–293

Hughes N (2010) Defi ning property rights to surface water in complex regulated river systems: generalising the capacity sharing concept, In: 54th annual AARES conference, Australian Agricultural and Resource Economics Society, Adelaide

Letsoalo A, Blignaut J, de Wet T, de Wit M, Hess S, Tol R, van Heerden J (2007) Triple dividends of water consumption charges in South Africa. Water Resour Res 43:W05412. doi: 10.1029/2005WR004076

Lofgren H, Robinson S, Nygaard D (1996) Tiger or turtle? Exploring alternative futures for Egypt to 2020. TMD discussion paper no. 11, Trade and Macroeconomics Division, International Food Policy Research Institute, Washington, DC

MDBA (2010) Guide to the proposed basin plan: overview. Murray Darling Basin Authority, Canberra

NRE and NSW Agriculture (1999) Structural adjustment pressures in irrigated broadacre farming in the Southern Murray-Darling Basin. ABARE, Canberra

Pannell D, Roberts AM (2010) Australia’s National action plan for salinity and water quality: a retrospective assessment. Aust J Agric Resour Econ 54(4):437–456

Perera C, James B, Kularathna M (2003) Computer software tool for sustainable water allocation and management – REALM, in MODSIM 2003 international conference on modelling and simulation, Townsville

Quiggin J (1988) Murray river salinity: an illustrative model. Am J Agric Econ 70:635–645 Quiggin J (2001) Environmental economics and the Murray–Darling river system. Aust J Agric

Resour Econ 45:67–95 Qureshi ME, Connor J, Kirby M, Mainuddin M (2007) Economic assessment of acquiring water

for environmental fl ows in the Murray Basin. Aust J Agric Resour Econ 51:283–303 Randall A (1981) Property entitlements and pricing policy for a maturing water economy. Aust

J Agric Resour Econ 25:195–220 Roe T, Dinar A, Tsur Y, Diao X (2005) Feedback links between economy-wide and farm-level

policies: with application to irrigation water management in Morocco. J Policy Model 27:905–928


Seung CK, Harris TR, MacDiarmid TR, Shaw WD (1998) Economic impacts of water reallocation: a CGE analysis for the Walker River Basin of Nevada and California. J Region Anal Policy 28:13–34

Seung CK, Harris TR, Eglin JE, Netusil NR (2000) Impacts of water reallocation: a combined computable general equilibrium and recreation demand model approach. Ann Region Sci 34:473–487

SKM (1998) PRIDE documentation. Sinclair Knight Merz, Melbourne Watson A (2003) Approaches to increasing river fl ows. Aust Econ Review 36:213–224 Young M, Proctor W, Qureshi ME, Wittwer G (2006) Without water: the economics of supplying

water to 5 million more Australians. Water for a healthy country fl agship report, CSIRO and Centre of Policy Studies, Monash University, Melbourne

Zaman A, Malano H, Davidson B (2009) An integrated water trading-allocation model, applied to a water market in Australia. Agric Water Manage 96:149–159


Abstract TERM-H2O is a dynamic variant of TERM with agricultural detail adapted to include regional water accounts. This chapter describes: (a) the technol-ogy assumptions in TERM-H2O for farm industries, (b) the creation of an input-output database for farm industries including inputs of water, (c) the derivation of input demand functions for farm industries and their calibration using farm sector input-output data, and (d) assumptions concerning mobility of capital, land, owner-operator labour and hired labour between farm industries. A key assumption is that irrigable land in TERM-H2O can be used either as an input of irrigated land or an input of dryland, with the division determined endogenously via water availability.

Keywords Dynamic modeling • Farm factor mobility • Input demand functions • Satellite water accounts • Water trade modeling

5.1 Background

This chapter details the theory and database preparation of TERM-H2O. As outlined in Chap. 4 , many models of irrigation cropping exist. Our aim is to devise a dynamic CGE model that represents small regions as separate economies and is capable of tracking observed changes in water use and farm industry outputs. To this end, we modify TERM (outlined in Chaps. 2 and 3 ) to include a detailed representation of farm industries emphasising requirements for irrigation water.

P. B. Dixon • M. T. Rimmer • G. Wittwer (*) Centre of Policy Studies , Monash University , Wellington Rd, Clayton campus , Clayton , VIC 3800 , Australia e-mail: [email protected] ; [email protected] ; [email protected]

Chapter 5 The Theory of TERM-H2O

Peter B. Dixon , Maureen T. Rimmer and Glyn Wittwer

80 P.B. Dixon et al.

5.1.1 Why CGE? Why Dynamics?

The need for a CGE model that represents small regions in the Murray-Darling Basin has become evident during the debate over water buybacks. These are pur-chases of irrigation water entitlements by the Australian government from farmers. The government’s motivation is to leave more water in Australia’s major river system and thereby address environmental concerns.

There are several reasons for adopting a general rather than a partial equilibrium approach for looking at policy issues such as buyback. First, there is major interest in the effect of agricultural/environmental policies on regional employment. There is a belief, at least outside of the economics discipline, that multipliers on reductions in agricultural output will result in total regional job losses that are manyfold greater than the direct farm impacts. We show in Chap. 6 that overall regional impacts may be in fact less than the direct reduction in agricultural employment arising from water buybacks.

Second, agricultural/environmental policies change factor prices. This causes reallocations of resources between agricultural activities and between agriculture and the rest of the economy. Only a general equilibrium approach is adequate to capture these inter-industry effects.

Third, income and consumption effects of agricultural environmental policies can sometimes play a major role in determining overall regional effects on employ-ment. For example, under the current water buyback policy of the Australian gov-ernment, farmers are being fully compensated for water removed from production. Spending by farmers arising from buyback receipts should be included in any assessment of the effects of the policy. Capturing the effects of these expenditures requires a general equilibrium framework which takes account of the regional origin of consumption goods purchased by farm households.

In addition to a general equilibrium perspective, TERM-H2O includes dynamics (Dixon and Rimmer 2002 ). This allows us to distinguish between short-run and long-run responses to policy changes. For example, any factor that is fi xed in the short run will respond to model shocks by short-run changes in prices and sluggish adjustments in quantities. This applies to fi xed farm capital and housing. Sluggish adjustments mean that short-run effects on regional employment of policy changes are likely to be smaller than long-run effects.

Another advantage of dynamics is that we can track the performance of TERM-H2O against observation as a way of testing whether the model’s results are believable. We have found that much of the modeled adjustment to changes in water availability occurs by reallocating water through water trading and movements of farm factors between activities. This is consistent with observed changes in water usage.

5.2 Theoretical Modifi cations to Farm Industries: Input-Output Structure

TERM-H2O contains 35 industries, including 17 farm and 10 irrigation sectors, producing 28 commodities in up to 46 regions in a bottom-up framework. In the application reported in Chaps. 6 and 7 , the regional dimension is somewhat

815 The Theory of TERM-H2O

aggregated. Substantial theoretical modifi cations have been made to standard CGE models such as ORANI (Dixon et al. 1982 ) and TERM (Horridge et al. 2005 ) to represent irrigation sectors in a multi-regional CGE framework.

An effective way to gain an introductory understanding of a CGE model such as TERM-H2O is to look at key aspects of the input-output structure. An industry’s column of an input-output table shows its cost structure and a commodity’s row shows its sales structure. The input-output table also serves another role. It depicts the economy’s initial situation, that is, it provides the initial solution of the model. In modeling undertaken in this chapter, we start with a 2005–06 database.

Table 5.1 is a representation of TERM-H2O’s input-output structure for the farm sector in a region. The columns refer to farm industries. In TERM-H2O, industries are defi ned by region, irrigation status and main product. Examples of farm indus-tries are: Lower Murrumbidgee/irrigated/cereal, Lower Murrumbidgee/dry/cotton and Rest of NSW/dry/other livestock. While each farm industry produces just one product, all agricultural products are produced by several industries. For example, Rest of NSW/irrigated/cotton, Queensland/irrigated/cotton and Rest of NSW/dry/cotton all produce cotton.

We adopt the concept of single-product farm industries because it is in line with available data from the Australian Bureau of Statistics (ABS). These data show outputs by commodity and region. Data exist on the area of irrigation by broad activity, which serve as a starting point for splitting irrigated from dryland activity. Section 5.2.3 provides more details on this split. At fi rst glance, it may seem that our single-product approach is in tension with the Australian reality of multi-commodity farm enterprises (e.g. wheat/sheep farms). However, in the theory described in this chapter, a given farm enterprise can be spread across several farm industries. Our model allows for price-induced movements of productive resources between farm industries. We can think of such movements as occurring at the farm level with the farm manager reallocating labour, land and other resources between the production of different commodities.

Table 5.1 Farm industries in region d in the input-output data for TERM-H2O a

Irrigated industries Dryland industries

Inputs Livestock(2 inds)

Annual crops(6 inds)

Perennial crops(2 inds)

Crops(5 inds)

Livestock (2 inds)

Intermediate, N×(R + 1) Y Y Y Y Y Cereal including hay, R – – – – √ Irrigation water √ √ √ – – Capital, general purpose Y Y Y Y Y Capital, specifi c Y – Y – – Hired labour, M Y Y Y Y Y Operator labour Y Y Y Y Y Dryland – – – √ √ Unwatered irrigable land √ √ √ √ √ Output Total Total Total Total Total

Ticks indicate fl ows of particular interest for this chapter. A large √ indicates a potentially large value. A small √ indicates a small value. – indicates a zero or negligible value a Y indicates a potentially substantial entry, for example, all industries use intermediate inputs


The rows of Table 5.1 refer to values of inputs and outputs. The fi rst N×(R + 1) rows (labelled Intermediate) are fl ows of N intermediate inputs to farm industries in region d from R regions of Australia and from abroad (imports). The next R + 1 rows refer to inputs that receive special treatment in TERM-H2O: cereals differentiated by region of supply and irrigation water. Cereals (including hay) are used as feed in dryland livestock industries. Irrigation water is water obtained from an irrigation authority by allocation or trading, or from rain falling on land being used by irri-gated industries. The remaining M + 5 input rows are a disaggregation of value added into returns to: general purpose capital, items such as tractors, sheds and other farm equipment that are readily transferable between farm activities; specifi c capital such as orchards, vineyards, and herds that are not readily transferable; M types of hired labour; operator labour (the farmer and family); dryland; and unwa-tered irrigable land (rent excluding the value of irrigation water).

5.3 Production and Input-Demand Functions for Farm Industries in TERM-H2O

This section describes the production functions adopted in TERM-H2O. It then sets out the implied input-demand functions and shows how these are calibrated using data of the type represented in Table 5.1 .

5.3.1 Production Functions

We consider farm industry (q,d) where q refers to the industry’s irrigation status and crop (e.g. irrigated/rice) and d refers to the industry’s region (e.g. Lower Murrumbidgee). The modifi cations specifi c to TERM-H2O in the structure of (q,d)’s production func-tion are illustrated in Fig . 5.1 . As in other versions of TERM, we assume that output is a Leontief function of intermediate input and primary factor. Intermediate input is a CES combination of inputs of many goods. Each of these goods is a CES combination of the imported and domestic varieties. The domestic variety of good i is a CES com-bination of good i produced in each of the R regions. This relatively standard treat-ment of intermediate inputs is not shown in detail in Fig. 5.1 .

Primary factor is a CES combination of three inputs: land & operator, general purpose capital and hired labour. Hired labour is a CES combination of labour of M different occupations or skill categories. The only part of the structure in Fig. 5.1 that is a departure from earlier versions of TERM, and the only part that receives any further attention in this chapter, is the treatment of land & operator. In earlier versions of TERM, there were no underlying inputs generating land & operator. In TERM-H2O, land & operator is a CES nest of inputs of operator labour (the farmer and family), specifi c capital and total land. There are then several nests below total land.


Leontief

Intermediateinput

Primaryfactor

CES

Land &Operator Hired Labour

Output

Generalpurpose capital

Specificcapital Total land

CES

Labour type1

Labour typeM

CES

Effectiveland Cereal

CES

CES

Cerealregion 1

Cerealregion R

CES

. .. . .

.....

Inputs orOutputs

Functionalform

KEY

Unwateredirrigable land Water

Leontief

Operatorlabour

Irrigable land Dry landIrrigated land

Fig. 5.1 Production function for a farm industry


The fi rst of these nests makes total land a CES combination of effective land and cereal (including purchased hay). This nest is relevant only for dryland livestock indus-tries: for other industries, the use of cereals in our database is negligible, ensuring that total land is simply effective land. For dryland livestock industries, we recognise that cereal (i.e., hay or feed grain) is a substitute for land: a given amount of livestock can be maintained on less land if we use more cereal. We assume that all cereal is domesti-cally produced. As with other domestically produced intermediate inputs, we model the input of cereal as a CES combination of inputs from the R regions.

Effective land is shown in Fig. 5.1 as a CES combination of irrigated land, (unwatered) irrigable land and dry land. For dryland industries, the use of irrigated land is negligible. Thus, for these industries, effective land is a CES combination of only unwatered irrigable land and dry land. For irrigated industries, the use of unwa-tered irrigable land and dry land is negligible. Thus, for irrigated industries, effective land is simply irrigated land. The bottom nest concerns the input of irrigation water. We model this in a Leontief nest with unwatered irrigable land to form irrigated land. Thus, we assume that irrigated land used by industry (q,d) is always fully watered.

Notice that (unwatered) irrigable land appears twice: in the nest below effective land and in the nest below irrigated land. TERM-H2O implies that a signifi cant frac-tion of the available supply of irrigable land in any region is allocated as unwatered irrigable land to the region’s dryland industries when there are shortages of irriga-tion water. With reductions in water availability, TERM-H2O generates increases in the prices of irrigation water and reductions in the rental values of unwatered irri-gable land. This causes dryland industries to increase their demands for unwatered irrigable land. At the same time, irrigated industries suffer cost increases, causing them to reduce their demands for unwatered irrigable land. In this way, unwatered irrigable land is moved from irrigated industries to dryland industries.

5.3.2 Input Demand Functions

In this section, we discuss the details of the land & operator specifi cation in Fig. 5.1 together with the implied input-demand equations for irrigation water, unwatered irrigable land, dry land, operator labour, specifi c capital and cereals.

5.3.2.1 Demands for Land

First, we examine the composition of the effective land input. In deciding its inputs of irrigated land (LNI), unwatered irrigable land (LNUW) and dry land (LND), we assume that farm industry (q,d)

chooses XLN(q,d,k), k {LNI,LNUW,LND}Î

k

to minimise PLN(q,d,k)* XLN(q,d,k)å (5.1)


( )ksubject to XELC(q,d,EL) CES XLN(q,d,k) ,= (5.2)

where XLN(q,d,k) refers to industry (q,d)’s inputs of land of type k (irrigated land,

unwatered irrigable land, or dry land), PLN(q,d,k) is the cost to industry (q,d) of using a unit of land of type k, 1 and XELC(q,d,EL) is a measure of (q,d)’s requirements for effective land. We defi ne the cost of using a unit of effective land to industry (q,d) as:

kPLN(q,d,k)*XLN(q,d,k)

PELC(q,d,EL)XELC(q,d,EL)

=å

(5.3)

Models such as TERM-H2O are computed with equations that are linear in per-centage changes. The percentage change equations arising from ( 5.1 ) to ( 5.3 ) that are included in TERM-H2O are:

[ ]xln(q,d,k) xelc(q,d,EL)

ln(q,d) pln(q,d,k) pelc(q,d,EL) , k {LNI,LNUW,LND}

=- - Îσ (5.4)

and

kpelc(q,d,EL) SLN(q,d,k)* pln(q,d,k),= å (5.5)

where xln(q,d,k), xelc(q,d,EL), pln(q,d,k), and pelc(q,d,EL) are percentage changes in

the variables defi ned by the corresponding uppercase symbols s ln(q,d) is (q,d)’s elasticity of substitution between irrigated land, unwatered

irrigable land, and dry land in the generation of the overall input of effective land and SLN(q,d,k) is the share of k (irrigated land, unwatered irrigable land, or dry land)

in (q,d)’s cost of using effective land, that is:

j

PLN(q,d,k)* XLN(q,d,k)SLN(q,d,k) , k {LNI,LNUW,LND}.

PLN(q,d,j)* XLN(q,d,j)= Îå (5.6)

5.3.2.2 Demands for Effective Land and Cereal

In deciding its inputs of effective land (EL) and cereal (C), we assume that farm industry (q,d)

chooses XELC(q,d,k), k {EL,C}Î

åk

to minimise PELC(q,d,k)* XELC(q,d,k) (5.7)

1 As we will see, in the case of irrigated land used by irrigated industries, this cost includes not only the rent but also the cost of irrigation water.


( )ksubject to XTLOPSK(q,d,TL) CES XELC(q,d,k)/AELC(q,d,k) ,= (5.8)

where XELC(q,d,k) refers to inputs to industry (q,d) of k (effective land or cereal) PELC(q,d,k) is the cost to industry (q,d) of using a unit of input k (effective land

or cereal) 2 XTLOPSK(q,d, TL) is (q,d)’s requirements of total land (a composite of effective

land and cereal or a measure of land input with associated food for maintaining livestock) and

AELC(q,d,k) are variables that can be used to introduce productivity changes. For example, if AELC(q,d, EL) increases by 50%, then for any given input of cereal, industry (q,d) needs 50% more effective land to achieve any given level of total land requirements. Equivalently, if AELC(q,d, EL) increases and effective land input is held constant, then industry (q,d) will need to increase its input of cereal to achieve a given level of total land input. AELC(q,d, EL) can be used in simulations of the effects of changes in the weather.

The percentage change equations arising from ( 5.7 ) and ( 5.8 ) that are included in TERM-H2O are:

j

j

xelc(q,d,k) aelc(q,d,k) xtlopsk(q,d,TL)

tlnd(q,d) pelc(q,d,k) SELC(q,d,j)* pelc(q,d,j)

tlnd(q,d) aelc(q,d,k) SELC(q,d,j)*aelc(q,d,j) , k {EL,C}

- =

é ù- -ë ûé ù- - Îë û

å

å

σ

σ (5.9)

where xelc(q,d,k), pelc(q,d,k), aelc(q,d,k) and xtlopsk(q,d,TL) are percentage changes

in the variables defi ned by the corresponding uppercase symbols s tlnd(q,d) is (q,d)’s elasticity of substitution between effective land and cereal in

the generation of input of total land and SELC(q,d,j) is the share of j (effective land and cereal) in (q,d)’s cost of total

land, that is:

k

PELC(q,d,j)* XELC(q,d,j) ,SELC(q,d,j) j {EL,C}.PELC(q,d,k)* XELC(q,d,k)

= Îå (5.10)

5.3.2.3 Demands for Operator Labour, Specifi c Capital and Total Land

In deciding its inputs of total land (TL), specifi c capital (SK) and operator labour (OP), we assume that farm industry (q,d)

Îchooses XTLOPSK(q,d,k), k {TL, SK, OP}

2 PELC(q,d, EL) is defi ned by ( 5.3 ). PELC(q,d, C) can be defi ned in a standard way in terms of prices of cereal from different regions via the CES specifi cation in the bottom right hand corner of Fig. 5.1 .


åk

to minimise PTLOPSK(q,d,k)* XTLOPSK(q,d,k) (5.11)

( )= ksubject to XLOKH(q,d,LO) CES XTLOPSK(q,d,k) , (5.12)

where XTLOPSK(q,d,k) refers to inputs to industry (q,d) of k (total land, specifi c capital

or operator labour), PTLOPSK(q,d,k) is the cost to industry (q,d) of using a unit of input k, 3 and XLOKH(q,d,LO) is a measure of (q,d)’s total requirements of land and operator

(LO). The percentage change equations arising from ( 5.11 ) and ( 5.12 ) that are included

in TERM-H2O are:

jxtlopsk(q,d,k) xlokh(q,d,LO) (q,d) ptlopsk(q,d,k) S(q,d,j) * ptlopsk(q,d,j) ,

k {TL,OP,SK}

σ é ù= - -ë ûÎ

å (5.13)

where xtlopsk(q,d,k), ptlopsk(q,d,k) and xlokh(q,d,LO) are percentage changes in the

variables defi ned by the corresponding uppercase symbols s (q,d) is (q,d)’s elasticity of substitution between total land, specifi c capital and

operator labour in the generation of the overall input of land & operator and S(q,d,j) is the share of j (total land, specifi c capital or operator labour) in (q,d)’s

cost of land and operator, that is:

= Îå k

PTLOPSK(q,d,j)* XTLOPSK(q,d,j)S(q,d,j) , j {TL,OP,SK}.

PTLOPSK(q,d,k)* XTLOPSK(q,d,k) (5.14)

5.3.2.4 Demand for Irrigation Water and Irrigable Land in Formation of Irrigated Land

Under our Leontief assumption, industry (q,d)’s demands for irrigation water and unwatered irrigable land are proportional to the industry’s demand for irrigated land, XLN(q,d,LNI). The Leontief assumption also allows us to relate the cost to industry (q,d) of using irrigated land to the rental value of unwatered irrigable land and to the price in region d of irrigation water via the equation:

= +PLN(q,d,LNI) PLN(q,d,LNUW) WPH(q,d)* PW(d), (5.15)

where PW(d) is the price or value per unit of irrigation water in region d and WPH(q,d) is the technologically determined use of irrigation water per hectare

of irrigated land in industry (q,d).

3 Movements in PTLOPSK(q,d,OP) are determined by movements in the demand for and supply of owner operators. Percentage movements in PTLOPSK(q,d,TL) are determined according to

k

ptlopsk(q,d,TL) SELC(q,d,k)*(pelc(q,d,k) aelc(q,d,k))= +å .


In ( 5.15 ), we assume that irrigation water is freely movable between irrigation industries in region d, implying that it has the same price throughout a region.

The percentage change form of ( 5.15 ) for inclusion in TERM-H2O can be written as:

= +pln(q,d,LNI) SLNUW(q,d)* pln(q,d,LNUW) SW(q,d)* pw(d), (5.16)

where the lowercase symbols are percentage changes in the variables denoted by the

corresponding uppercase symbols and SW(q,d) is the share of irrigation water in the cost to (q,d) of using irrigated land and SLNUW(q,d) is the share of rents on irrigable land (before application of water)

in the cost to (q,d) of using irrigated land. As mentioned already, for dryland industries, we use variations in AELC(q,d,EL),

appearing in ( 5.8 ), to represent variations in rainfall. In simulations in which cli-matic conditions are ideal in region d, AELC(q,d,EL) is set at one for dryland industries. In simulations representing severe drought conditions, AELC(q,d,EL) may be set as high as 5. For irrigated industries, AELC(q,d,EL) will normally be set at one: under our assumption that WPH(q,d) is determined technologically, variations in climatic conditions affect the quantity of irrigable land allocated to irrigated industries but not the productivity of the land so allocated.

5.3.2.5 Calibrating Input-Demand Equations in TERM-H2O

To implement Eqs. 5.4 , 5.5 , 5.9 , 5.13 and 5.16 in TERM-H2O, we need to specify initial values for share coeffi cients and values for substitution parameters. The ini-tial values for the share coeffi cients are computed from the input-output database represented by Table 5.1 . For dryland industries, SLN(q,d,k) used in ( 5.5 ) is com-puted from the dryland and unwatered-irrigable-land rows of Table 5.1 . For irri-gated industries, SLN(q,d,k) is one for k = LNI and zero otherwise. For dryland industries, SELC(q,d,j) used in ( 5.9 ) is computed from the sum of the entries in (q,d)’s cereal rows and the sum of the entries in its two land rows. For irrigated industries, SELC(q,d,j) is one for j = EL and zero for j = C. For all industries, S(q,d,j) used in ( 5.13 ) is computed from: the operator-labour row, the specifi c capital row, and the sum of the cereal, two land and irrigation water rows. For irrigated indus-tries, SW(q,d) and SLUW(q,d) in ( 5.16 ) are computed from the irrigation-water and unwatered-irrigable-land rows of Table 5.1 . For dryland industries, these two coeffi cients are irrelevant.

For s ln (q,d), the elasticity of substitution between land types appearing in ( 5.4 ), we

adopt a high value, 10.0. As explained earlier, this elasticity plays a role only for dry-land industries. We adopt a high substitution value because whenever irrigable land is used by a dry industry, this land is unwatered and therefore similar to dry land.

For s tlnd

(q,d), the elasticity of substitution in dry livestock production between effective land and cereal appearing in ( 5.9 ), we adopted a value of 3. In simulations,


we fi nd that this value is large enough to induce substantial substitution away from irrigated livestock production towards cereal production to support dryland livestock production when water scarcity worsens.

For s (q,d), the elasticity of substitution between total land, specifi c capital and operator labour appearing in ( 5.13 ), we adopted the low value of 0.5, implying that total land and operator labour usually move closely together. For irrigation indus-tries, this is approximately equivalent to assuming a fi xed amount of operator labour per hectare of land used. This is because for an irrigation industry, input of effective land and total land move closely in line with hectares of irrigable land.

5.4 Region-Wide Constraints and the Determination of Factor Rents and Prices for Water

In this section, we discuss the region-wide constraints applying to operator labour, irrigable land, dryland, general purpose capital and specifi c capital. We also discuss the rental prices of these factors and the price of irrigation water.

5.4.1 Determination of Rents

We assume that each region d has available for year t fi xed amounts of the factors irrigable land (LNIRR), 4 dry land (LND), operator labour (OP) and general purpose capital (K), that is, there is a fi xed amount of each f in the set {LNIRR, LND, OP, K}. For each f, TERM-H2O allocates this fi xed amount between the H(d) farm industries in region d in a price-sensitive way according to the optimization problem

= ¼choose Z(q,d,f), q 1,2, ,H(d)

åq

to maximise PZ(q,d,f)* Z(q,d,f) (5.17)

( )qsubject to ZTOT(d,f) CET Z(q,d,f ) ,= (5.18)

where Z(q,d,f) is the supply of factor f to industry (q,d), ZTOT(d,f) is a measure of the total quantity of factor f available in region d and PZ(q,d,f) is the rental rate for factor f when used by industry (q,d). In Sect. 5.3 ,

PZ(q,d,LNIRR), PZ(q,d,LND) and PZ(q,d,OP) were denoted as PLN(q,d,LNUW), PLN(q,d,LND) and PTLOP(q,d,OP).

4 Irrigable land can be used as irrigated land (LNI) or as unwatered irrigable land (LNUW).


Optimization problems ( 5.17 ) and ( 5.18 ) give TERM-H2O percentage change equations describing the supply of factors to industries. These equations take the form:

Vz(q,d,f) ztot(d,f) (d,f)* (pz(q,d,f R(v,d,f)*pz(v,d,f )), for all (q,d) and f,τ= + -å (5.19)

where z(q,d,f), ztot(q,d,f) and pz(q,d,f) are percentage changes in the variables denoted

by the corresponding uppercase symbols; R(v,d,f ) is industry (v,d)’s share of the total rental value of factor f in region d and t (d,f ) is a positive parameter (transformation elasticity) that refl ects the ease

with which factor f can be moved between industries in region d. With demands specifi ed through the optimization problems set out in the previ-

ous section, and with supplies specifi ed through the optimization problems set out in this section, TERM-H2O determines rental rates via market-clearing equations:

XTLOPSK(q,d,OP) Z(q,d,OP) for all (q,d)= (5.20)

=XLOKH(q,d,K) Z(q,d,K) for all (q,d) (5.21)

=XLN(q,d,LNI) Z(q,d,LNIRR) for irrigated industries q and regoins d (5.22a)

=XLN(q,d,LNUW) Z(q,d,LNIRR) for dry industries q and regions d (5.22b)

=XLN(q,d,LND) Z(q,d,LND) for all (q,d). (5.23)

The total supplies in region d of irrigable land [ZTOT(d,LNNIR)], dry land [ZTOT(d,LND)] and operator labour [ZTOT(d,OP)] are treated as exogenous vari-ables throughout a TERM-H2O simulation. Supplies of general purpose capital [ZTOT(d,K)] are fi xed within a period but are allowed to change from period to period with availability in period t determined by availability in t − 1 modifi ed by deprecia-tion and investment in period t − 1. Investment in period t − 1 is modeled as a function of expected rates of return, refl ecting mainly the rental rate on general purpose capital in region d in period t − 1. Supplies of specifi c capital to industry q in region d are treated in a similar way to supplies of general purpose capital to region d.

5.4.2 Determination of the Price of Irrigation Water

The supply of irrigation water [ZW(q,d)] to irrigation industry q in region d is given by:

= + +ZW(q,d) AW(q,d) TRADE(q,d) NatW(d)* XLN(q,d,LNI), (5.24)

where AW(q,d) is the amount of irrigation water allocated to (q,d) via the irrigation

system,


NatW(q,d) is the amount of irrigation water per hectare supplied to (q,d) through rainfall,

TRADE(q,d) is the net amount of irrigation water obtained by (q,d) from trade with other industries and regions and

XLN(q,d,LNI) is, as defi ned earlier, the amount of irrigated land used by (q,d). The demand for irrigation water [XW(q,d)] by irrigation industry (q,d) is given by:

=XW(q,d) WPH(q,d)* XLN(q,d,LNI). (5.25)

TRADE(q,d) is determined by equating demand and supply:

=XW(q,d) ZW(q,d). (5.26)

If no inter-regional water trade were allowed, then regional water prices [PW(d)] could be determined by imposing the constraint that the sum over q of TRADE(q,d) is zero. However, TERM-H2O allows fl exibly for different water trading regimes by including equations of the form:

= åG

g = 1

PW(d) Dummy(d,g)* PWG(g) for all d (5.27)

and

=å dDummy(d,g)* TRADE(d) 0 for all g, (5.28)

where TRADE(d) is the net acquisition of water by region d through trade, that is,

qTRADE(d) TRADE(q,d).= å (5.29)

G is the number of groups of regions that form separate water trading blocks PWG(g) is the price of irrigation water in trading group g and Dummy(d,g) = 1 if d is in trading group g and zero otherwise. Then, the no-inter-regional-trade case is handled by making G the number of

regions so that ( 5.27 ) and ( 5.28 ) reduce to:

=PW(d) PWG(d) for all d (5.30)

and

=TRADE(d) 0 for all d. (5.31)

If trade is possible between all regions, then G = 1 and ( 5.27 ) and ( 5.28 ) reduce to:

=PW(d) PWG(1) for all d (5.32)

and

=å dTRADE(d) 0. (5.33)


If there are two water trading groups, one consisting of regions 1 to R 1 and the

other consisting of regions R 1 + 1 to R, where R is the number of regions, then ( 5.27 )

and ( 5.28 ) reduce to:

= = ¼ = = + ¼1 1PW(d) PWG(1), d 1 ,R ; PW(d) PWG(2), d R 1, ,R (5.34)

and

= = +

= =å å1

1

R R

d 1 d R 1

TRADE(d) 0 and TRADE(d) 0. (5.35)

5.4.3 The Impacts of Water Trading

In typical applications of TERM-H2O, we start with a database in which the price of irrigation water is equal across all irrigators in all regions. In the usual closure, we permit trading across the entire SMDB and also permit trade between users in other individual regions. But there may be some simulations where we wish to move from a year in which trade between regions or users was not permitted to a year in which it is. We may wish to estimate the impact of water trading. This can be done by including in TERM-H2O an equation of the form:

= - +PW(q,d) PW0(q,d)*(1 U) PWALL. (5.36)

In ( 5.36 ), PW0(q,d) is the initial water price for each irrigation industry in all regions of the SMDB and U is a variable with an initial value of 0. PWALL also has an initial value of 0. To impose water trading, we shock U to 1 and treat PWALL as an endogenous variable. Then, ( 5.36 ) forces the price of water, PW(q,d), to every industry in the trading regions to be equalised at PWALL.

5.5 Database Amendments

Preparation of a suitably disaggregated multi-regional database entailed a num-ber of steps summarised in Fig. 5.2 . The published IO table released by ABS ( 2006a ) distinguishes 109 sectors. Some of the main irrigation crops, namely, grapes, cotton, fruit and vegetables, are represented by a single composite sector (other agriculture) in the published table. Our model requires separate represen-tation of different irrigation activities with differing water requirements. Therefore, the fi rst step in regional database preparation procedure was to split the national database to sectors of interest. In agriculture, we relied on ABS data on the national value of agricultural production plus small area agricultural data to split the national data (ABS 2008a, b ) . Our enlarged national IO database included 172 sectors.


The next step was to choose a suitable base year. We chose to update the IO table to 2005–06. 5 This enabled us to align the national database with the 2006 census, from which we obtained small region data at the three-digit ANZSIC level for services and at the four-digit level for other sectors. ABS census data include employment by industry for each of 1,400+ statistical local areas. These census data supplement small area agricultural output data by representing grapes, fruit trees, rice and cotton as individual sectors, thereby allowing us to enhance versions of TERM dealing with irrigation sectors and water accounts.

The regional master database on which TERM-H2O is based represents the Australian economy at the statistical subdivision level containing 206 regions

Census data onindustries by regions;agricultural output

data

National IO table109 sectors

IO and interregional trade data, 172sectors by 206 regions

National IO table172 sectors (includes

irrgation-relevant sectors)

28 sectors by 46 regions, emphasison agriculture and Murray-Darling

Basin

35 sectors by 46 regions, splitsirrigation and dry-land activities.Estimate water use in irrigation

sectors

Fig. 5.2 Data generation procedure for TERM-H2O

5 A program used to update the national 2001–02 database is downloadable from www.monash.edu.au/policy/archivep.htm TPMH0058.

http://www.monash.edu.au/policy/archivep.htm



(i.e., we aggregate data from the census for 1,400+ regions to 206 regions). Each of these regions has its own input-output structure and inter-regional trade matrices at the 172 sectoral level. The motivation for this level of detail is that it allows us to cover many of the major topics in Australian policy analysis, including water. In the case of water, the statistical subdivisions of the Murray-Darling Basin align quite well with catchment regions. Chapter 2 outlines in more detail the procedure used to create the master database.

We undertake two sets of aggregations from the 172 sector, 206 region master data-base, to prepare the TERM-H2O database. First, we aggregate the master database to 28 sectors, with an emphasis on farm sectors (10 out of 28) and 46 regions, repre-senting the regions of the Murray-Darling Basin at the statistical subdivision level (40 regions), with more aggregated regions for the rest of Australia (6 regions). 6

With the 28 sector, 46 region database, the next step is to split farm sectors into irrigation and dryland sectors. We split dairy cattle, non-dairy livestock, cotton, cereals, fruit, sugar cane (with negligible production in the basin) and other agricul-ture into dryland and irrigated sectors. We assign rice, vegetables and grapes as exclusively irrigation activities. Initially, we use ABS census data and small region ABS farm census data to estimate regional farm outputs by crop or livestock type. Three additional sources provide us with information to split small region outputs into irrigation and dryland technologies. First, we have basin-wide estimates of irri-gation output by type. Second, we have estimates of the volume of water used in producing these outputs ( ABS 2010 ) . Third, we have the total water used in each statistical division within the basin (ABS 2006b , 2008a, b, c, d ) .

From the fi rst additional source, we can estimate a basin-wide split between irriga-tion and dryland outputs. That is, we already know basin-wide farm outputs by type. The dryland outputs are equal to total outputs minus irrigation outputs. From the second source, we can assign water volumes to the irrigation crop and livestock outputs.

The third source allows us to adjust the split in individual regions so as to reconcile with both regional outputs and regional aggregates of water used. For example, the Murray Statistical Division’s actual water use (the third additional source) was higher than estimated by our initial estimate based on the split from the fi rst and second additional sources. The irrigated share of non-dairy livestock production (a large farm sector in the region) was adjusted upwards in Murray so that it was higher than for the total basin. Another example is that dairy production could be split evenly between irrigated and dryland production in a high rainfall region. In a region with lower average rainfall, we increase the irrigated share and decrease the dryland share of dairy production. In practice, we use a RAS procedure so as to provide the best fi t to each of the data sources. We end with 35 industries producing 28 commodities in each region (Fig . 5.3 ).

Finally, we had to account for the value of water in production in the database. The year on which we base TERM-H2O (2005–06) was the most recent of the years with available data in which water was relatively abundant. Given this, in the

6 After further database modifi cations as outlined below, we aggregate TERM-H2O to around 20 regions to reduce solution time and ease presentation of results.


TERM-H2O database, we assigned a relatively low initial unit value to water used by irrigators, $30 per megalitre. 7 This is consistent with average prices reported for temporary trades in Victoria and New South Wales in 2000–01 (ABS 2004 , Fig. 11.1 and Table 21.5). Water’s value is based on the initial price multiplied by the estimated volume of usage. We had to reassign primary factor values in each irrigation industry so as to account for the value of water. Water rights may be embedded in the industry GOS. We adjusted primary factor returns so as to include

34

28

31

30

17

18

1

6

29

5

168

13

4

2

19

3

12

32

915

7

3314

112721

2310

20

2425

26

22

Fig. 5.3 Regions available in TERM-H2O. 1 Rest of NSW; 2 Tamworth + Northern Slopes, NSW; 3 North Central, NSW; 4 Dubbo + Central Macquarie, NSW; 5 Macquarie-Barwon, NSW; 6 Upper Darling, NSW; 7 Bathurst + Orange + Central Tablelands, NSW; 8 Lachlan, NSW; 9 Queanbeyan + Southern Tablelands, NSW; 10 Lower South Coast, NSW; 11 Snowy, NSW; 12 Wagga Wagga + Central Murrumbidgee, NSW; 13 Lower Murrumbidgee, NSW; 14 Albury + Upper Murray, NSW; 15 Central Murray, NSW; 16 Murray-Darling, NSW; 17 Far West, NSW; 18 Rest of Victoria; 19 Mildura + West Mallee, Vic; 20 East Mallee, Vic; 21 Bendigo, North Loddon, Vic; 22 South Loddon, Vic; 23 Shepparton + North Goulburn, Vic; 24 South Goulburn, Vic; 25 South-West Goulburn, Vic; 26 West Ovens-Murray, Vic; 27 Wodonga + East Ovens-Murray, Vic; 28 Rest of Queensland; 29 Toowoomba + Darling Downs, Qld; 30 South West, Qld; 31 Rest of South Australia; 32 Riverland, SA; 33 Murray Mallee, SA; 34 Rest of Australia

7 Since the relative scarcity of water varies widely according to rainfall conditions, the base price relative to which changes in the price of water are reported also varies widely over time. For this reason, despite the theory of TERM-H2O including substitutability between water and other fac-tors based on percentage changes in relative prices, for convenience, we report the ordinary change (i.e., dollars per megalitre) in water price in scenarios.


the cost of water in irrigated sectors. Natural rainfall is also embedded in GOS but requires no adjustment. We distribute 50% of the remaining GOS to each of owner-operators and unwatered irrigable land. A negligible value is assigned to dryland rentals. Given the unit cost of water imposed above, the shares of irrigation water in primary factor costs (the cost of water plus returns to labour capital and land) are around 20% in the non-rice cereals, grapes, vegetables, fruit and other agriculture sectors; 10% in the irrigated cotton sector and around 50% in the rice sector.

5.5.1 Extending TERM-H2O Beyond the Murray-Darling Basin

TERM-H2O represents the entire economy of Australia, with small region coverage in the Murray-Darling Basin and composite regions covering the rest of the nation. Could we use the present version of the model to analyse water issues outside the basin? The answer in theory in yes, but in practice, we would fi rst have to redress database defi ciencies. The ABS ( 2008c ) has provided a split between dryland and irrigation output for the basin. The ABS has no such estimates for regions outside the basin. For example, the South East Statistical Division of South Australia is an impor-tant irrigation region for which we have no direct estimate of the split. Any attempt to devise such a split based on available data would be a treacherous exercise. For exam-ple, the ABS has regional data on the number of hectares of crop irrigation, but a stumbling block is that these appear to be irrigable rather than irrigated hectares. We require several leaps of faith to convert such data to values of irrigation output.

With time and effort, we could improve the water accounts for regions outside the basin. We might improve our interpretation of ABS data. The reality is that such improvements will arise from funding for studies of irrigation for regions outside the basin. The focus may or may not turn to South Australia’s South East Statistical Division. A reason non-basin regions may be of increasing interest is that drought and buybacks have induced some pessimism over the future of irrigation within the basin. Since fl oods rather than drought were the major issue in the basin by early 2011, and since our fi ndings are that buybacks will have only a small negative impact on farm output within the basin, such pessimism may be misplaced. Nevertheless, water is an issue in every farming region. A key issue in the South East region, for example, may concern the extent to which expansion of irrigation farming will deplete groundwater supply over time.

5.6 Review

This chapter has detailed the theoretical modifi cations required to transform the basic theory of TERM into a model that can deal with farming in irrigation regions in considerable detail. In order to capture observed changes in water use


for individual crops in response to changes in water availability, it is necessary to include a theory of farm factor mobility as elaborated in Sect. 5.3.2 . Without factor mobility (based in TERM-H2O on CET functions), even with full water trading between regions, it is not possible to replicate the observed movement in water between activities (see Sect. 7.2 ).

A cornerstone of farm factor movements in TERM-H2O is the fi xed water requirement per hectare of land for each irrigation output, as detailed in Sect. 5.3.2.4 . This implies that as water availability falls, the amount of irrigable land used for irrigation may decline. Part of the response to reduced water availability may be also via a movement of water into other irrigation activities. But a striking response to falling water availability has occurred in the case of dairy production. As water becomes scarce, dairy producers switch from irrigated to dryland technologies. In effect, the use of irrigated pasture falls while fodder inputs increase. Dairy produc-ers may have a preference for using irrigated pasture when possible, but during drought, fodder inputs become relatively less scarce than irrigation water.

A concern arises from allowing farm factor mobility in TERM-H2O. This is that perennial crops (orchards and vineyards) include substantial specifi c capital, refl ect-ing the costs of establishing a plantation, and the lag it takes before the plantation is in commercial production. Consequently, each irrigation activity designated as a perennial (including the livestock sectors) includes some specifi c capital. This makes it more diffi cult for these sectors to reduce water usage as water scarcity worsens – unless a dryland production technology is available, as in the case of livestock. 8

Model features of this sort are necessary when it comes to dealing with policy-relevant scenarios. In the study of the buyback policy detailed in Chap. 6 , we see that the rising price of water and falling price of land are keys to explaining differ-ences in regional outcomes in the scenario. At a sectoral level, in the buyback sce-nario, we might think of rice producers, with a high water-to-land value ratio (due to large water allocations), doing better than orchard or vineyard producers, with a lower water-to-land value ratio (due to a large amount of fi xed capital).

References

ABS (Australian Bureau of Statistics) (2004) Water account Australia, 2000-01. ABS, Canberra. http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/4610.02000-01?OpenDocument . Accessed 31 Mar 2009

ABS (Australian Bureau of Statistics) (2006a) Australian National Accounts: input-output tables, 2001-02. ABS, Canberra. http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/5209.0.55.0012001-02?OpenDocument . Accessed 31 Mar 2009

ABS (Australian Bureau of Statistics) (2006b) Water account Australia, 2004-05. ABS, Canberra. http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/4610.02004-05?OpenDocument . Accessed 31 Mar 2009

8 A more recent version of TERM-H2O allows specifi c livestock capital to move between irrigated and dryland activities.


http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/5209.0.55.0012001-02?OpenDocument

http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/5209.0.55.0012001-02?OpenDocument



ABS (Australian Bureau of Statistics) (2008a) Agricultural commodities: small area data, Australia. ABS, Canberra. http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/7125.02005-06%20%28Reissue%29?OpenDocument . Accessed 31 Mar 2009

ABS (Australian Bureau of Statistics) (2008b) Water use on Australian farms, 2005-06. ABS, Canberra. http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/4618.02005-06?OpenDocument . Accessed 31 Mar 2009

ABS (Australian Bureau of Statistics) (2008c) Agricultural commodities, Australia, 2005-06. ABS, Canberra. http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/7121.02005-06 Accessed 31 Mar 2009

ABS (Australian Bureau of Statistics) (2008d) Water and the Murray-Darling Basin – a statistical profi le, 2000-01 to 2005-06. http://www.abs.gov.au/AUSSTATS/[email protected]/-DetailsPage/4610.0.55.0072000-01%20to%202005-06?OpenDocument . Accessed 31 Mar 2009

ABS (Australian Bureau of Statistics) (2010) Experimental estimates of the gross value of irrigated agricultural production, 2000–01 to 2006–07. http://www.abs.gov.au/ausstats/[email protected]/mediareleasesbyReleasDate/88B62133AF6D7723CA25776F001B35C1?OpenDocument . Accessed 3 Feb 2011

Dixon P, Rimmer M (2002) Dynamic general equilibrium modelling for forecasting and policy: a practical guide and documentation of MONASH, Contributions to Economic Analysis 256. North-Holland, Amsterdam

Dixon P, Parmenter B, Sutton J, Vincent D (1982) ORANI: a multisectoral model of the Australian economy, Contributions to Economic Analysis 142. North-Holland, Amsterdam

Horridge M, Madden J, Wittwer G (2005) Using a highly disaggregated multi-regional single-country model to analyse the impacts of the 2002-03 drought on Australia. J Policy Model 27:285–308

http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/7125.02005-06%20%28Reissue%29?OpenDocument

http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/7125.02005-06%20%28Reissue%29?OpenDocument


http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/7121.02005-06

http://www.abs.gov.au/AUSSTATS/[email protected]/-DetailsPage/4610.0.55.0072000-01%20to%202005-06?OpenDocument

http://www.abs.gov.au/AUSSTATS/[email protected]/-DetailsPage/4610.0.55.0072000-01%20to%202005-06?OpenDocument

http://www.abs.gov.au/ausstats/[email protected]/mediareleasesbyReleasDate/88B62133AF6D7723CA25776F001B35C1?OpenDocument

http://www.abs.gov.au/ausstats/[email protected]/mediareleasesbyReleasDate/88B62133AF6D7723CA25776F001B35C1?OpenDocument


Abstract We use TERM-H2O in analysing the effects of the Australian Government buying back water from irrigators in the Southern Murray-Darling Basin (SMDB) and thereby increasing river fl ows. Results are explained using data from the model and simplifi ed theory. We refer to this as the ‘back-of-the-envelope’ approach. Back-of-the-envelope calculations and regressions allow us to explain key features of the results including differences in regional outcomes. Controversially, our results suggest that buyback would increase economic activity in SMDB. Although a scheme of environmentally useful size would sharply increase the price of irrigation water, there would be little effect on aggregate SMDB farm output. Instead, farm resources would be reallocated between activities. Because farmers are owners of water rights, they would benefi t from the price increase induced by buyback. Community anxiety in the basin over buybacks may have arisen because the buyback process started during a period of drought-induced stress.

Keywords Back-of-the-envelope checking • Macroeconomic impacts • Factor price impacts • Water buybacks • Asset value calculation

6.1 The Policy Context

On 25 January 2007, the then Prime Minister John Howard announced a 10-point plan to address problems arising from water allocation in rural Australia. This was a further development in policy dealing with the environmental health of the

Chapter 6 Buybacks to Restore the Southern Murray-Darling Basin*

Peter B. Dixon , Maureen T. Rimmer and Glyn Wittwer

P. B. Dixon • M. T. Rimmer • G. Wittwer (*) Centre of Policy Studies , Monash University , Wellington Rd, Clayton campus , Melbourne , VIC 3800 , Australia e-mail: [email protected] ; [email protected] ; [email protected]

* This chapter reproduces with permission substantial portions of Dixon et al. (2011).


Murray-Darling Basin. More than a decade earlier, the Council of Australian Governments (COAG) recognized the need for policy reform in the management of water in the basin (COAG 1994 ) . Although COAG reforms had proceeded slowly, several desirable changes were made, including the disentanglement of land and water ownership. Eight of the ten points in the 2007 plan are relevant to the Murray-Darling Basin:

1. A nationwide investment in Australia’s irrigation infrastructure to line and pipe major delivery channels

2. A nationwide programme to improve on-farm irrigation technology and metering 3. The sharing of water savings on a 50:50 basis between irrigators and the Australian

Government leading to greater water security and increased environmental fl ows 4. Addressing once and for all water over-allocation in the Murray-Darling Basin 5. A new set of governance arrangements for the Murray-Darling Basin 6. A sustainable cap on surface and groundwater use in the Murray-Darling Basin 7. Major engineering works at key sites in the Murray-Darling Basin such as the Barmah

Choke and Menindee Lakes 8. Expanding the role of the Bureau of Meteorology to provide the water data necessary

for good decision making by governments and industry

Australian Government ( 2007 ) , p. 1

The 2007 plan included provision of $3 billion for the Australian Government to spend on buybacks (point 4). The Murray-Darling Basin Authority (MDBA) formed in 2008 had the task of determining the volumes of water required to address point 6 of the plan, concerning sustainable diversion limits (SDLs). While provision for buybacks was always part of the plan, it was not clear how the Labor government, which came into offi ce later in 2007, would implement the volumes set by the Murray-Darling Basin Authority (MDBA). Until late in 2010, public servants in the Australian Government and various state governments were referring to ‘clawback’, which implied that there would be reductions in water volumes allocated to farmers without compensation. New Prime Minister Gillard announced in August 2010 that a Labor government would purchase all the water required by the MDBA to achieve sustainable fl ows in the Murray-Darling Basin (Kenny 2010 ) . This announcement removed clawback from the policy agenda.

6.2 The Effects of a Buyback Scheme

This chapter reports on TERM-H2O modeling 1 of the effects of an illustrative buy-back scheme implemented over the period 2009–16. We simulate a scheme which starts in 2009 with the Australian Government buying back from farmers in the Southern Murray-Darling Basin (SMDB) permanent rights to an annual allocation

1 We use a version of the model that identifi es 19 regions (13 in SMDB) and 35 industries (17 farm industries).

1016 Buybacks to Restore the Southern Murray-Darling Basin

of 187.5 gigalitres (GL) of irrigation water. Then in 2010, the Australian Government buys back a further 187.5 GL of permanent rights. The process is completed after 8 years (in 2016) by which time the Australian Government has purchased 1,500 GL of permanent water rights. 2

The increase in water availability to the river system and the corresponding reduction in water availability to farmers are the same as the Australian Government’s accumulated purchase of water rights in a given year only if allocations are equal to those of 2005–06. We assume that 2009, 2010 and 2011 are drought years with about 70%, 80% and 90% of normal rainfall and about 70%, 80% and 90% allocation of water to rights holders relative to 2005–06. Although the Australian Government buys 187.5 GL of permanent rights in 2009, this produces an extra fl ow for the river system of only about 131 GL (=0.7*187.5). In 2010 and 2011, the extra fl ows for the river system and correspondingly the buyback-related reductions in supply to irrigators, are 300 and 506 GL (=0.8*375 and 0.9*562.5). For the rest of the simu-lation period, up to 2018, we assume normal rainfall and allocations equal to those of 2005–06.

In our simulation, the Australian Government implements its scheme by buying permanent water rights in a fully functioning market. Because we assume that water can be freely traded in SMDB, our simulation results for the effects of the scheme on farm outputs, inputs and investment do not depend on what we assume about who sells their water rights. However, for the purpose of our computations, we assume that sales are at a uniform rate across rights holders. As explained in Sect. 6.4 , we calculate the sale price for permanent rights as the present value of future water allocations. In making this calculation, we estimate future prices via a model simulation.

6.2.1 How Does the Buyback Scheme Affect the Price of Water?

The paths of the price of irrigation water in the baseline run (without the buyback scheme) and the policy run (with the buyback scheme) are shown in Fig. 6.1 . Without the scheme, irrigation water becomes relatively cheap because the return of normal rainfall and 100% allocations in 2012 make water relatively abundant. The decline in the price of water continues beyond 2012. This refl ects an assumption built into our baseline that farmers achieve an annual rate of water-saving technical

2 This scheme is hypothetical but seemed realistic at the time we completed our modeling. PC ( 2010 , p. 261) refers to a 1,500-GL buyback as being the benchmark used by COAG. Subsequently, with the return of heavy rains in 2010–11, both the size of the buyback mentioned in public discussions (MDBA 2010 ) and estimates of the likely availability of irrigation water without the buyback have been revised upwards. The 1,500-GL scenario reported in this chapter is based on 2005–06, allocations. This represents a similar percentage of water removed from irrigation as 3,000 GL across the basin (north plus south) based on entitlements.


change of 1% a year, thereby reducing their demand for water. With the buyback scheme, the price of water still declines over the period 2009–12 with the assumed easing of the drought. The price rises steeply after 2012, refl ecting the build-up of the buyback scheme up to 2016.

To describe the results in Fig. 6.1 , we fi tted a regression system of the form

( )policy

base

( ) 1* 1 BBR( ) , 2009, ,2018,

( ) ( )

P tn n t t

P t t

æ ö= - = ¼ç ÷è ø

� �ε

(6.1)

and

( )base

base

( )1* , 2009, ,2018,

( ) 2008

P ta b t

t P

æ ö= + = ¼ç ÷è øε

(6.2)

where P

policy ( t ) and P

base ( t ) are the prices of water in Fig. 6.1

BBR( t ) is the proportionate reduction in the supply of irrigation water caused by buyback

e ( t ) is the elasticity of demand for irrigation water in year t a and b are parameters to be estimated Via ( 6.2 ), we treat the demand elasticity as a variable rather than a parameter.

Combining ( 6.1 ) and ( 6.2 ), we obtained the fi tted equation

( )policy base

base base

2

( ) ( )8.63 3.83* * 1 BBR( ) ,

( ) (2008)

2009, ,2018, 0.99.

P t P tn n t

P t P

t R

æ ö æ ö= - + -ç ÷ ç ÷è øè ø

= ¼ =

� �

(6.3)

policy (with buyback)

baseline (without buyback)

$

20080

140

120

100

80

60

40

20

2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Fig. 6.1 Price of irrigation water in SMDB ($ per megalitre)


In our baseline run, P base

( t )/ P base

(2008) varies from a high of 1.37, occurring in 2009, to a low of 0.23, occurring in 2018. This means that the implied elasticity of demand for water varies from −0.30 in 2009 when the price of water is high to −0.13 in 2018 when the price of water is low. 3 Thus, TERM-H2O implies that demand elasticities have low magnitude at low water prices and higher magnitude at high prices 4 : a given percentage increase in the price of water only has a signifi cant effect on demand when water prices are relatively high. 5

6.2.2 How Does the Buyback Scheme Affect the National Economy?

The TERM-H2O simulation shows that the national macro effects of the buyback scheme are small. As can be seen in Fig. 6.2 , at the end of the simulation period (2018) with the scheme fully operational, the reduction in GDP caused by allowing

-0.007

-0.006

-0.005

-0.004

-0.003

-0.002

-0.001

02008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

TERM-H2O

BoTE calculation

Fig. 6.2 Buyback-induced percentage effects on real GDP: TERM-H2O result and back-of-the-envelope calculation

3 −0.30 = 1/(−8.63 + 3.83*1.37) and −0.13 = 1/(−8.63 + 3.83*0.23). 4 This is consistent with the meta-analysis of water price elasticities by Scheierling et al. ( 2006 ) . 5 In assessing these implied elasticity values, it is necessary to consider the defi nition of irrigation water. On our defi nition, about 30% of irrigation water is obtained by irrigators directly from rain falling on their land. Our elasticities can be converted into elasticities of demand for diverted water by dividing by 0.7, giving a range from −0.18 to −0.42. This range is compatible with the value −0.3 quoted in a recent ABARE study (Hone et al. 2010 , p. 30) in which demand elasticities refer to diverted water.


0.0135

0.1154

5085 6585 Irrigation water (GL)

$m per GL

a b

c d

e$97m

Fig. 6.3 Demand for irrigation water in SMDB in 2018

6 Table 6.1 includes only 15 of the 17 farm industries mentioned in Chap. 5 . It excludes the two sugar industries that have negligible output in the SMDB.

an extra 1,500 GL of water to fl ow in the river system is about 0.0059%. This can be understood via a back-of-the-envelope (BoTE) calculation based on Fig. 6.3 .

The ce line in Fig. 6.3 shows SMDB’s demand curve for irrigation water implied by TERM-H2O for 2018. The points c and e are policy and baseline results (Fig. 6.1 ). On the basis of Fig. 6.3 , we expect a buyback-induced reduction in GDP in 2018 of about $97 million [= 0.5*(0.1154 + 0.0135)*1,500]. With baseline GDP in 2018 being $1,849 billion, this represents a percentage reduction in GDP of 0.0052, close to the TERM-H2O result of 0.0059. Similar BoTE calculations are shown in Fig. 6.2 for each simulation year. These calculations closely match the TERM-H2O results.

As is clear from this analysis, GDP in TERM-H2O gives no value to extra water in the river system: this water simply represents a loss of an input to production. Thus, the simulated reduction in GDP is purely a measure of the cost of the buyback scheme, against which benefi ts can be assessed.

TERM-H2O produces results on the percentage effects of buyback for the full range of national macro variables. However, in line with the GDP results, these are negligible and do not warrant attention here.

6.2.3 How Does the Buyback Scheme Affect Output by Farm Industry and Region in SMDB?

Table 6.1 shows percentage effects of the buyback scheme on farm industry outputs in regions of SMDB in 2018. 6 Similar tables could be provided for other years. We display the table for 2018 as representing the long-run effects of the scheme.

Tabl

e 6.

1 B

uyba

ck-i

nduc

ed p

erce

ntag

e de

viat

ions

in f

arm

out

puts

in S

MD

B r

egio

ns in

201

8

Reg

ion

(see

Fig

. 6.4

) In

dust

ry

2

3

4

5

6

8

9

10

11

12

13

14

17

WagCntMrm NSW

LMrmb NSW

AlbUpMrry NSW

CentMrry NSW

MrryDrlng NSW

MldWMalee Vic

EMallee Vic

BndNthLod Vic

SthLoddon Vic

ShepNGoul Vic

SSWGlbrn Vic

OvnsMurry Vic

MurrayLnds SA

SMDB Totals

1

Cer

eal-

dry

5.0

14.3

3.

7 33

.8

5.3

5.6

6.6

4.2

4.3

7.3

4.7

5.4

5.5

6.4

2

Cer

eal-

irri

g −

40.3

−

27.2

−

42.0

−

2.5

−35

.8

−37

.0

−36

.7

−40

.5

−41

.5

−37

.3

−40

.4

−39

.5

−43

.4

−20

.9

3

Ric

e −

39.9

−

26.1

N

A a

2.9

NA

−

30.4

−

30.3

N

A

NA

−

30.3

−

33.9

−

33.2

N

A

−20

.6

4

Dai

ryC

at-d

ry

15.1

21

.7

14.5

32

.0

14.2

13

.9

15.2

12

.8

13.0

17

.4

13.5

14

.3

15.5

17

.1

5

Dai

ryC

at-i

rrig

−

16.4

−

6.5

−17

.4

10.8

−

12.9

−

10.7

−

10.6

−

13.4

−

13.5

−

10.3

−

13.0

−

12.5

−

12.7

−

10.0

6

O

thL

ivst

o-dr

y 4.

1 8.

4 3.

1 35

.3

4.0

4.1

5.2

3.5

3.9

6.5

4.3

5.1

4.7

4.8

7

Oth

Liv

sto-

irri

g −

14.5

−

8.0

−15

.9

−18

.6

−11

.1

−14

.1

−13

.8

−18

.5

−16

.9

−14

.5

−16

.7

−15

.4

−16

.0

−16

.1

8

Cot

ton-

dry

NA

20

.3

NA

N

A

13.5

N

A

NA

N

A

NA

N

A

NA

N

A

NA

19

.5

9

Cot

ton-

irri

g N

A

−11

.9

NA

N

A

−13

.0

NA

N

A

NA

N

A

NA

N

A

NA

N

A

−12

.5

10

Gra

pes

−5.

7 1.

7 −

6.7

15.8

−

3.1

−4.

6 −

4.5

−7.

8 −

7.9

−8.

5 −

10.2

−

7.1

−5.

7 −

4.2

11

Veg

etab

les

2.1

18.4

0.

0 57

.6

6.6

3.5

4.6

0.7

1.0

5.4

1.6

2.4

1.3

5.0

12

Frui

t-dr

y 6.

0 15

.2

4.7

28.2

5.

9 6.

4 7.

6 5.

2 5.

2 8.

6 5.

8 6.

5 6.

2 8.

3 13

Fr

uit-

irri

g −

2.5

5.1

−3.

5 21

.0

−0.

1 −

1.6

−1.

2 −

3.2

−3.

2 −

1.0

−2.

8 −

2.5

−2.

7 −

0.5

14

Oth

Agr

i-dr

y 7.

3 10

.8

6.9

13.9

6.

0 7.

2 8.

1 7.

1 7.

1 8.

9 7.

3 7.

9 8.

8 8.

6 15

O

thA

gri-

irri

g −

7.2

2.6

−8.

4 23

.0

−4.

0 −

6.2

−5.

8 −

8.4

−8.

4 −

6.0

−7.

8 −

7.5

−9.

3 −

3.9

16

Reg

iona

l tot

als

−1.

6 −

2.4

−1.

5 −

2.1

−0.

6 −

1.1

−1.

4 −

1.6

−1.

3 −

1.9

−1.

7 −

1.8

−1.

1 −

1.3

17

Shar

e ir

rig

inds

in

farm

pri

mar

y fa

ctor

s %

25

69

23

97

74

48

58

32

42

64

38

48

66

54

18

% d

ev H

H

cons

umpt

ion

0.09

2.

11

−0.

07

2.80

0.

36

0.61

0.

58

−0.

02

−0.

06

0.48

0.

04

0.00

0.

39

0.34

a N

A s

igni

fi es

not a

pplic

able

– o

utpu

t in

the

regi

on in

the

base

per

iod

(200

8) is

neg

ligib

le


18

15

16

19

191

7

17

6 3

285

1413

4

1012

9

11

Fig. 6.4 Map of SMDB regions used in the buyback simulation. Notes: Regions – ( 1 ) Upper Darling-Lachlan, ( 2 ) Wagga-Central Murrumbidgee, ( 3 ) Lower Murrumbidgee, ( 4 ) Albury-Upper Murray, ( 5 ) Central Murray, ( 6 ) Murray-Darling, ( 7 ) Rest of VIC, ( 8 ) Mildura-West Mallee, ( 9 ) East Mallee, ( 10 ) Bendigo-North Loddon, ( 11 ) South Loddon, ( 12 ) Shepparton-North Goulburn, ( 13 ) South/Southwest Goulburn, ( 14 ) Ovens-Murray, ( 15 ) QLD, ( 16 ) Rest of SA, ( 17 ) Murray Lands SA, ( 18 ) Rest of Australia, ( 19 ) Rest of NSW. Southern Murray-Darling Basin (SMDB) consists of regions 2 – 6 , 8 – 14 and 17

The last entry in row 16 of Table 6.1 indicates that in the long run, the scheme reduces farm output in SMDB by 1.3%. Referring back to Fig. 6.3 , we recall that the scheme removes from farm production in 2018 a factor (irrigation water) that would make a contribution to output of $97 million. In the baseline, the return to primary factors (including water) in farm industries in SMDB in 2018 is $8,652 million. Thus, the removal of water explains a reduction in farm output of 1.1% (= 100*97/8,652). The rest of the reduction (1.3 less 1.1) implied by TERM-H2O is explained by a small reduction in the use by the farm sector of other primary factors, particularly specifi c capital in irrigation industries. The cut in the availability of irrigation water in SMDB causes a shrinkage in farm investment and consequently a long-run reduction in capital.

The industry totals in the last column of Table 6.1 show a clear pattern but with a puzzling exception. Irrigated industries generally show negative deviations in out-put while dryland industries show positive deviations. The exception is vegetables


(an entirely irrigated crop) which show a positive output deviation. In looking at the regional totals (row 16), we might anticipate seeing relatively large negative deviations for farm output in regions with high dependence on irrigated industries (large numbers in row 17) and relatively small negative deviations for regions with low dependence. This works for regions 3, 5, 9 and 12 (high dependence associated with relatively large output contraction) and for region 8 (low dependence associated with mild output contraction). However, regions 6 and 17 exhibit high dependence and mild reductions in farm output while regions 2, 4, 10, 11, 13 and 14 exhibit low depen-dence and relatively large reductions in farm output. A regression of the numbers in row 16 against those in row 17 gives an R 2 of 0.02.

Why does TERM-H2O tells us that a Commonwealth buyback scheme with a consequent increase in the price of irrigation water would stimulate SMDB’s output of an irrigation crop such as vegetables? At a superfi cial level, the answer is that when water prices are high, growing vegetables in SMDB is a better use of irrigable land and water than growing other crops such as rice. Why does TERM-H2O tells us that the scheme would have a relatively minor effect on farm output in an irriga-tion-intensive region such as Murray-Darling-NSW? This time, the superfi cial answer is that grapes- and fruit-irrig, major irrigation industries in the region, are not as badly affected by the scheme as other irrigation crops such as rice which are barely represented in the region.

To deepen our understanding of the regional and industry results generated by TERM-H2O, we develop an equation in which the scheme-induced percentage devia-tion in output of farm industry i in region d in year t [x(i,d,t)] is regressed against three measures of the scheme’s impact on (i,d)’s competiveness in year t . These measures cover: the SMDB-wide competitiveness of the commodity produced by industry i in markets external to SMDB, the competitiveness of industry (i,d)’s product against the corresponding product from other regions of SMDB, and the competitiveness of industry (i,d) against other producers of its product in region d. Our equation is:

( )( )

0 1 2

3

x(i,d,t) *Cctot(i,•,t) * Cc(i,d,t) Cctot(i,•,t)

* C(i,d,t) Cc(i,d,t) ,

2009, ,2018.for all industries i and regions d in SMDBand for t

= + + -

+ -

= ¼

α α αα

(6.4)

In this equation, C(i,d,t) is the impact effect of the scheme on costs per unit of output in industry i, region d and year t; Cc(i,d,t) is the scheme’s impact in year t on the cost in region d of the commodity that industry i produces. For example, if i’s product is fruit, then Cc(i,d,t) is an output-weighted average of the C(i,d,t)s for i equals fruit-irrig and fruit-dry.

Cctot(i,•,t) is the scheme’s impact in year t on the cost in SMDB of the commod-ity that industry i produces. For example, if i’s product is fruit, then Cctot(i,•,t) is an output-weighted average of the C(i,d,t)s for i equals fruit-irrig and fruit-dry and all regions in SMDB.

In Eq. 6.4 , we expect a 0 to be close to zero and a

1 , a

2 and a

3 to be negative. In

terms of a specifi c example, we expect the equation to imply that the output of the fruit-irrig industry in Murray-Darling-NSW is adversely affected by:


An overall increase in the cost of producing fruit in SMDB with a resulting • reduction in external demand An increase in the cost of producing fruit in Murray-Darling-NSW relative to the • average cost of producing fruit across all regions in SMDB An increase in the cost of producing fruit in the fruit-irrig industry in Murray-• Darling-NSW relative to the average cost (over the irrigation and dry fruit indus-tries) of producing fruit in this region

In formulating the cost-impact effect, C(i,d,t), we started by using:

basetry1

base

VWATER i,d,tC i,d,t * pwater(t),

COSTS i

( )( )

( ),d,t= (6.5)

where VWATER

base (i,d,t) is the baseline value of water used by industry i in region d in

year t 7 COSTS

base (i,d,t) is the total baseline value of inputs to industry i in region d in year t

pwater(t) is the percentage deviation in the price of water in year t caused by the buyback scheme.

With C measured according to ( 6.5 ), the fi tted version of ( 6.4 ) is:

2

x(i,d,t) 1.72 0.85*Cctot(i,•,t) 0.78*(Cc(i,d,t) Cctot(i,•,t))2.65*(C(i,d,t) Cc(i,d,t)),

2009,...,2018, 0.76

for all industries i and regions d in SMDB

and for t R

= - - -- -

= =

(6.6)

Although the equation fi ts quite well ( R 2 = 0.76) and the coeffi cients have the expected signs, on close inspection, we found it to be unsatisfactory. It strongly underestimates the TERM-H2O results for many irrigation industries, particularly in region 5 (Central-Murray-NSW). Consider, for example, rice in region 5, which shows a positive entry in Table 6.1 (2.9%). Examination of our data showed that rice production in this region is just as water intensive as in other regions. Consequently, ( 6.5 ) gives rice production in region 5 a similar cost impact to that of rice production in other SMDB regions. Thus, Eq. 6.4 fi tted with cost impacts measured according to ( 6.5 ) is unable to explain why rice in region 5 shows a positive output deviation in Table 6.1 while rice in other regions shows negative output deviations. Similarly, the regression equation was unable to explain why vegetables performed so strongly in region 5 (57.6% in Table 6.1 ) or why vegetable production shows positive output deviations in all SMDB regions.

These problems in explaining TERM-H2O results for irrigation industries led us to consider the role of the rental prices for irrigable land. We noticed that TERM-H2O implies that the buyback scheme would cause a collapse in these prices in region 5. While this would not make land-owning farmers happy, it would certainly explain why TERM-H2O shows relatively strong output responses to the buyback scheme for some industries in the region. For example, rice production, an intensive user of irrigable land, becomes relatively cheap in region 5. Thus, rice production in region 5 expands while that in other regions contracts.

7 Water costs were computed using baseline water prices (Fig. 6.3 ) even for water supplied under a free allocation or by rainfall.


What explains TERM-H2O’s simulated movements in the rental prices of irri-gable land? We expect increases in the price of irrigation water to cause relatively sharp reductions in rental prices of irrigable land in regions with two characteristics: (a) a high ratio for the value of irrigation water to irrigable land and (b) limited options for using irrigable land in dryland activities. We confi rmed these ideas by fi tting a regression equation, obtaining:

base

base base

base

2

VW _ TOT d,tpirrl _ tot(d,t) 0.28 0.98*

VW _ TOT d,t + VIR _ TOT (d,t)

( )

( )

( )*SH _ IR d,t * pwater(t)

2009, ,2018,

0.95

for all regions d in SMDB and for t

R

æ ö= - - ç ÷è ø

= ¼=

(6.7)

In ( 6.7 ), VW_TOT base

(d,t) and VIR_TOT base

(d,t) are baseline values for irrigation water and the total rental on irrigable land in region d in year t. These values are used to refl ect characteristic (a).

SHR_IR base

(d,t) is the share of the farm sector’s primary factor input in region d and year t that is devoted to irrigation activities. This share is used to refl ect charac-teristic (b): we take high values to imply limited opportunity for economically via-ble transfers of irrigable land out of irrigated activities.

pirrl_tot(d,t) is the percentage deviation in the average rental price of irrigable land in region d and year t caused by the buyback scheme.

With an R 2 of 0.95, Eq. 6.7 tightly tracks the TERM-H2O results for rental devia-tions on irrigable land. In particular, it explains the collapse in the average rental price in Central-Murray-NSW, a region with a high value ratio for water to irrigable land and very little dryland activity.

Having explained how TERM-H2O determines buyback-induced deviations in the rental prices of irrigable land, it is now legitimate to use these deviations to improve our explanation of the deviations in industry outputs. We start by re-esti-mating the cost impacts of the buyback scheme, taking account not only of devia-tions in the price of water but also in the rental prices of irrigable land:

basetry2

base

base

base

VWATER i,d,tC i,d,t * pwater(t)

COSTS i,d,t

VIRRL i,d,t* pirrl _ tot(d,t

CO

( )( )

( )

STS i,

(

d,

)

t( )

æ ö= ç ÷è ø

æ ö+ ç ÷è ø

(6.8)

where VIRRL base

(i,d,t) is the baseline rental value of irrigable land used by indus-try i in region d in year t.

With cost impacts estimated according to ( 6.8 ), the fi tted version of ( 6.4 ) is

( )( )

2

x(i,d,t) 1.89 0.88*Cctot(i,•,t) 2.13* Cc(i,d,t) Cctot(i,•,t)

3.11* C(i,d,t) Cc(i,d,t) ,

2009, ,2018, 0.80

for all industries i and regions d in SMDB and for

t R

= - - -

- -

= ¼ =

(6.9)


Under ( 6.8 ) rather than ( 6.5 ), Eq. 6.4 fi ts with a higher R 2 (0.80 rather than 0.76), and inspection of residuals indicates a much improved performance for irrigation industries. The regression now refl ects an important mechanism operating in TERM-H2O: in irrigation industries based on land with limited dryland uses, land rents rather than outputs bear much of the damage from higher water prices. In these industries, output reductions are muted and water shortfalls are made up by water purchases from regions in which irrigable land is more adaptable to dryland activities.

While the inclusion of deviations in the rental prices of irrigable land in our esti-mates of cost impacts improves the performance of Eq. 6.4 , it is still not entirely satisfactory. We found that version ( 6.9 ) does a poor job in explaining output devia-tions in dryland industries in region 5 (Central-Murray-NSW). On looking for an explanation, we noticed that TERM-H2O implies that the buyback scheme would cause a relatively large reduction in the rental price of dry land in region 5, enhanc-ing the competitiveness of the region’s dryland industries. Because TERM-H2O allows irrigable land to be used as dry land, we can expect the model to show a tight connection between deviations in rental rates on the two types of land. This idea is confi rmed in the following regression:

2

pdryl _ tot(d,t) 1.03 0.60* pirrl _ tot(d,t),

2009, ,2018, 0.97for all regions d in SMDB and for t R

= += ¼ = (6.10)

where pdryl_tot(d,t) is the percentage deviation in the average rental price of dry land in region d and year t caused by the buyback scheme.

Following the same argument that led to ( 6.8 ), we produce our fi nal estimate for the cost impacts of the buyback scheme:

base

base

base

base

base

base

( )

( )

(

VWATER i,d,tC(i,d,t) * pwater(t)

COSTS (i,d,t)

VIRRL i,d,t* pirrl _ tot(d,t)

COSTS i,d,t

VDRYL i,d,t* pdryl _ tot(d,t)

COSTS i,

)

( )

d,t( )

æ ö= ç ÷è ø

æ ö+ ç ÷è ø

æ ö+ ç ÷è ø

(6.11)

where VDRYL base

(i,d,t) is the baseline rental value of dry land used by industry i in region d in year t.

Using ( 6.11 ) in Eq. 6.4 gives:

( )( )

2

x(i,d,t) 0.67 0.91*Cctot(i,•,t) 2.05* Cc(i,d,t) Cctot(i,•,t)

2.98* C(i,d,t) Cc(i,d,t) ,

2009, ,2018, 0.89

for all industries i and regions d in SMDB and for

t R

= - - -- -

= ¼ = (6.12)


Equation 6.10 provides a tight description ( R 2 = 0.89) of the TERM-H2O results for outputs. The coeffi cients have the expected signs, and we found no unsatisfac-tory residual patterns. 8

Using ( 6.12 ), we can explain all of the puzzling results in Table 6.1 . Via the fi rst variable on the right-hand side, industry 11 (vegetables) does well throughout SMDB because its costs fall [Cctot(11,•, t) is negative]. For vegetable production, the value ratio for water to irrigable land is low, so that the falls in the rental prices of irrigable land more than compensate for the increases in the price of irrigation water. Via the second variable on the right-hand side of ( 6.12 ), vegetables do par-ticularly well in regions 3 and 5 (Lower-Murrumbidgee-NSW and Central-Murray-NSW). In these regions, declines in the rental prices of irrigable land [explained in ( 6.7 )] generate sharp reductions in production costs relative to those in other SMDB regions [ Cc(11,d,t) Cctot(11,•,t)- is strongly negative for d = 3 and 5]. Via the third variable on the right-hand side of ( 6.12 ), with one exception, dryland industries do well relative to corresponding irrigated industries. A dryland industry facing com-petition from an irrigated industry generally gains a competitive advantage when the price of water rises [ C(i,d,t) Cc(i,d,t)- is less than zero where i is a dryland industry]. The exception is industry 14 (Other-agriculture-dry) in region 5. In this region, industry 15 (Other-agriculture-irrig) gains a competitive advantage over industry 14 [ C(14,5,t) Cc(14,5,t)- is greater than zero]. The cost characteristics of industries 14 and 15 in region 5 that lead to this result are: (a) industry 15 uses a relatively large amount of irrigable land but a relatively small amount of water and (b) industry 14 uses relatively little dry land.

6.2.4 How Does the Buyback Scheme Affect Consumption in SMDB Regions?

TERM-H2O implies that the buyback scheme would have a positive effect on household consumption in most SMDB regions throughout the simulation period. Results for 2018 can be seen in the last row of Table 6.1 . In aggregate, the consump-tion gain for SMDB in 2018 is 0.34%.

This result can be understood via Fig. 6.3 . As explained earlier, the fi gure implies that the sale of water to the Commonwealth removes from SMDB a factor of pro-duction, 1,500 GL of water, which can be used to generate $97 million of income

8 In terms of conventional econometrics, ( 6.12 ) is impressive. Potentially, it is fi tted with 1,950 observations (= 15 industries by 13 regions by 10 years). In fact, we used only 1,753 observations because we excluded cells with zero production. The t-statistics on our three variables are large (over 30). However, we have not reported them here because we do not think they have a valid interpretation. Our equation is not derived from a sample of outcomes from a stochastic process. It is a summary of a set of results produced by a non-stochastic piece of arithmetic, the solution of a CGE model.


(area ceba). If instead of being used in production this water is sold to the Commonwealth, it generates $173 million for SMDB (area cdba). 9 Thus, the exis-tence of the buyback scheme has an impact effect on SMDB income of $76 million (= 173–197). This is about 0.2% of regional consumption. Local multipliers act on this impact effect to produce the total effect on consumption (0.34%).

The extent to which consumption is stimulated in any region of SMDB depends on the region’s net exports of water (sales to other regions and, to the government, less imports from other regions). Regions with large net exports relative to their total consumption show the largest percentage gains in consumption from an increase in the price of water. With the buyback scheme, all regions become net exporters. However, Central-Murray-NSW and Lower-Murrumbidgee-NSW, both with quite small economies, have easily the largest buyback sales to the govern-ment. Consequently, they show much higher percentage increases in consumption than other regions.

An equation that provides a tight description of the TERM-H2O consumption deviations for 2018 is:

2

cons(d) 0.18 2.94* WXSH(d)* PWAT,

, 0.95for all regions d in SMDB R

= - + D= (6.13)

where cons(d) is the percentage deviation in consumption in region d in 2018 caused by the buyback scheme (the numbers in row 18 of Table 6.1 ).

WXSH(d) is 100 times the ratio of the quantity of net water exports from region d in 2018 (averaged across baseline and policy runs) to the value of regional household consumption.

D PWAT is the buyback-induced change in the price of water in 2018 (=0.1154–0.0135, see Fig. 6.3 ).

A key assumption underlying our consumption results is that buyback income is spent by farmers in the same way that they spend other income. However, it is pos-sible that buyback would induce some farmers to realize their windfall gain and leave SMDB. This would not necessarily invalidate our assumption that the number of owner operators is unaffected by buyback. However, it would mean that the pro-portion of the windfall gains spent inside SMDB would be lower than that for other SMDB farm income. In a sensitivity simulation, we assumed that only half of the windfall gains are spent in accordance with other farm income while the remainder is spent entirely outside SMDB. This simulation showed a consumption gain for SMDB in 2018 of 0.09%, down from 0.34%. Because demand for SMDB farm products depends very little on consumer demand in SMDB, our results for farm outputs were barely changed.

9 We assume that farmers in SMDB consume each year 5% of the amount that they receive from the government for their permanent water rights. This turns out in the long run to be similar to a situ-ation in which farmers sell their water each year to the government at the average price applying in that year and, for consumption purposes, treat their water revenue as current income.


6.3 Economic Conclusions and Modeling Insights

Both Coalition and Labor governments at the national level and through COAG have implemented policies to restore the environmental health of the Murray-Darling Basin. However, buybacks to restore environmental fl ows face opposition from people who fear adverse consequences for rural economies. For example, Nationals leader Warren Truss was quoted in The Australian (11 June 2009) as saying:

Buying up farms and water entitlements in NSW and Victoria to revitalize the ailing Murray-Darling river system will lead to a ‘national tragedy’, in which rural towns are slowly strangled to death.

The analysis in this chapter does not support this pessimistic view. It suggests that economic activity (represented by consumption) in most SMDB regions would be enhanced by a buyback scheme.

Our model indicates that water prices would rise substantially if 1,500 GL of allocated water were diverted to environmental uses. However, rather than causing a sharp reduction in farm activity in SMDB, we fi nd that the increase in the price of water would cause a reallocation of farm resources between activities.

In all SMDB regions, some irrigable land would revert to dryland uses, and dry-land farming would expand relative to irrigation farming. In some regions, the main reallocation of resources would be between irrigation activities. With more expen-sive water, irrigation-intensive regions would move from crops which use a large amount of irrigation water per hectare of irrigable land to crops which use lower amounts. For example, Lower-Murrumbidgee-NSW would reduce its outputs of rice and cotton and increase its outputs of grapes, vegetables, and fruit-irrig, while Central-Murray-NSW would reduce its output of other livestock-irrig and increase its outputs of vegetables and fruit-irrig.

Groups such as the NSW Irrigators’ Council (see NSWIC 2010 ) fi nd it hard to believe simulation results which show that large increases in the price of water could be accommodated with only a minor reduction (1.3%) in overall SMDB farm output. They point out that because of heavy reliance on export markets, farmers in SMDB are poorly placed to pass on cost increases to their customers. As shown by our general equilibrium calculations, increases in the price of water would be largely offset by reductions in the rental price of irrigable land. With these offsetting changes in factor prices, the international competitiveness of SMDB farming would be barely affected.

In assessing a policy involving large changes in factor prices, equity must be considered. In the case of buyback, we do not think that there is a severe equity problem. This is because the owners of water rights, the factor which would experi-ence a sharp increase in price, are largely the same people as the owners of irrigable land, the factor that would experience a sharp decrease in price.

While we would expect a 1,500-GL buyback scheme to have only a moderate effect on SMDB farm output, we would expect the effect to be negative. Con-sequently, it was initially a surprise that our simulations gave a positive effect for household consumption in SMDB and in most of its regions. The explanation is


that the bu yback scheme is an additional demand for a factor (water) that is owned by SMDB residents. This must make them better off: they will only sell their water rights if the market price exceeds the value to them of the water in production activities.

With regard to the national economy, our results show that an environmentally valuable buyback scheme could be implemented with no detectable effect on mac-roeconomic performance. The simulated long-run reduction in GDP caused by a 1,500-GL scheme is 0.0059%. This represents about 17-h economic growth. If in the absence of the scheme GDP was going to reach a certain level at midnight on 31 December 2018, then with the scheme, we would have to wait until 5 p.m. on 1 January 2019, to reach the same level. 10

6.3.1 The Response to Buybacks in Basin Communities

Late in 2010, the Murray-Darling Basin Authority met with hostile audiences when it toured the basin attempting to explain the details of its basin plan. This is under-standable if we think that, in the short term, there was substantial interaction between drought and community acceptance of buybacks. This is not because buybacks in any way worsened the plight of farmers during drought, but rather that initial buy-backs were associated with a time of extreme stress. Indeed, drought may have accel-erated the pace at which the buyback process proceeded. This is because the buyback process provides farmers with another fi nancial option and is a substitute for bor-rowing from a bank. 11 Drought caused severe fi nancial hardship. Buyback could not stop all cases of drought-induced fi nancial insolvency, but as an option for fi nanc-ing, it may have helped some farmers cope better with drought. Chapter 7 analyses the impacts of a prolonged drought on the basin and has become policy-relevant through the community confusion between buyback and drought impacts. 12

In public meetings, the message that buybacks were voluntary and entailed full compensation to farmers was lost. Community responses appeared to be consistent with the expectation of uncompensated ‘clawback’, a process ruled out several months before the MDBA roadshow. Perhaps, the voluntary nature of buyback was

10 We assume normal GDP growth of 3% a year. This means that GDP grows by 0.0059% every 17 h. 11 Prolonged drought worsened the environmental crisis in the Coorong and lower lakes. Senator Xenophon agreed to support a stimulus package devised by the then Rudd government in exchange for a short-term commitment of $500 million for buybacks (Keane 2009 ) . 12 One of our motivations for using a dynamic CGE model to estimate the impacts of buyback is that baseline conditions infl uence policy impacts. No factor varies as much in scarcity as water, refl ected in both irrigation water availability and dryland productivity. Therefore, the interaction between drought and the buyback process may be important in analysis of the economic impacts of buyback.


muddied when Prime Minister Gillard announced that the government’s aim was to complete the buyback process by 2014 (Kenny 2010 ) . An early fi nishing date for the plan relative to the timeline envisaged in the 10-point Howard plan implied a hurried process, and one which would reduce the potential role for water-saving technological change over time to ease the adjustment process. But perhaps, the greatest problem was that, without careful communication, buybacks may carry the connotation that farmers have been responsible for environmental problems. 13 COAG reforms through the separation of land and water ownership should have altered government and community perspectives. By treating water according to its scarcity, which varies widely from year to year, farmers are part of the solution in basin management.

Despite the commitment of state governments to COAG reforms, there has been state meddling in buybacks. A campaign led by the Victorian Farmers Federation persuaded the then Victorian government to impose a cap on perma-nent trades of water outside catchment regions within the state. The impact of this interference with the business plans of farmers soon became obvious. There was a stampede to trade water with the commencement of each season so as to trade within the cap. This led to an oversubscribed lottery every year for those wishing to sell water in the Goulburn region, emphasising how little support the cap had in practice (Frontier Economics 2009 ) . Despite this, the Victorian Farmers Federation ( 2008 ) misleadingly gave the impression that they had the support of all Victorian farmers for the cap. Irrigators wishing to downscale permanent plantings due to an oversupply of grapes or depressed prices for citrus crops have been frustrated by the Victorian cap. Buyback should have provided another sales option for these farmers.

Before the heartening impacts of good spring rains across the basin were replaced by summer fl oods, Sunrice announced that it would be reopening the Deniliquin rice mill after it had been foreclosed for 3 years (Sunrice 2010 ) . This is not consistent with the assertion of some lobbyists that buyback is like a perma-nent drought and hence worse than usual droughts. Buybacks continued through 2010: that a mill processing the most water-intensive crop in the basin reopened implies that buyback impacts are minimal. Drought was responsible for job losses in the basin. Drought closed the Deniliquin rice mill, and the end of drought led to its reopening. Given the context in which the buyback process started, during a severe drought, the drought modeling of Chap. 7 has policy relevance. Ongoing buybacks have impacts that are minor and will not slow the recovery of the basin from drought.

13 Government initiatives that established irrigation schemes did not involve market signals. Other objectives were at play, including soldier settlement schemes after both of the world wars. There is an almost universal tendency among irrigation schemes towards over-allocation in the absence of market signals.


6.4 Appendix : Calculating the Price of a Permanent Right to a Unit of Irrigation Water in the Southern Murray-Darling Basin

We calculate the price [PPerR(t)] that a farmer would need to receive in year t (t = 2009, …, 2016) to induce him/her to give up the permanent right to an annual allocation of one unit of irrigation water according to:

( ) y ty t

E[P(y)]* E[S(y)]PPerR t t 2009, ,2016

(1 d)

¥

-=

= = ¼+å (6.14)

where

E indicates expectation P(y) is the price of water in year y d is the discount rate (assumed to be 0.08 refl ecting 3% infl ation and a 5% real rate of interest) S(y) is the share of water rights in year y that is in fact allocated

As mentioned in Sect. 6.2 , the S(y)s in 2009, 2010 and 2011 were assumed to be 0.7, 0.8 and 0.9, refl ecting drought conditions that have made delivery of full water allocations impossible. For 2012–18, we set S(y) at one.

We assume that the expected values for P(y) and S(y) are given as follows:

E[P y ] PS(y), y 2009,( ) ,2018= = ¼ (6.15)

E[S(y)] S(y), y 2009, ,2018= = ¼ (6.16)

y 2018E[P(y)] PS(2018)*1.03 *SF(y) y 2018-= > (6.17)

{ }E[S(y)] S(t) y 2018, t 2009, ,2018 and

y t 10* n for n a positive integer

= > Î ¼

= + (6.18)

and

( )

( )( )( )( )

1 if E S y 1

1.4 if E S y 0.9SF y

1.84 if E S y 0.8

2.4 if E S y 0.7

ì é ù =ë ûïé ù =ï ë û= íé ù =ï ë û

ï é ù =ë ûî (6.19)

Via ( 6.15 ), we set expectations for water prices in 2009–18 according to the simulated values [PS(y)] obtained in our policy simulation, that is, with the buyback scheme in place. Via ( 6.16 ), we set the expected allocation shares in 2009–2018 according to the values adopted in our simulation. Via ( 6.17 ), we allow for 3% infl a-tion in the determination of expected water prices for years beyond 2018. We also


introduce a scarcity factor [SF(y)] to refl ect periodic droughts. As shown in ( 6.19 ), in years in which the expected allocation share is less than one, the scarcity factor magnifi es the expected price of water. The magnifi cations (1.4, 1.84 and 2.4) were calculated via simulations showing the effects on prices of reduced allocations. Via ( 6.18 ), we assume that the pattern of droughts (and hence allocation shares) in the decades beyond 2018 repeats the pattern assumed for the decade from 2009 to 2018.

Prices in 2009, dollars per ML calculated from ( 6.14 ) to ( 6.19 ), are shown in Table 6.2 . They imply an average price between 2009 and 2016 of $2,081. The aver-age price per ML for sales of permanent water rights in SMDB up until 31 January 2010 was $1,654 on sales of 502 GL (the average annual equivalent of 796 GL of entitlements; see PC ( 2010 , Table 1.2)). A somewhat higher price could be expected if the Commonwealth implemented a buyback scheme totalling 1,500 GL as assumed in this paper. The Howard plan included a provision of $3 billion for buy-backs (Australian Government 2007 ) . Given the prices in Table 6.2 , the total cost of the 1,500-GL program is $3.1 billion.

The expected frequency of future droughts is likely to be higher when the basin remains in drought and then fall after the drought ends. Available data indicate that the asset price of water fell in 2010. 14 That the price only fell more than a year after the drought had broken indicates that market players remained cautious for a time before changing their expectations. The calculated costs to the Australian Government of the scenario analysed in this chapter would have fallen had the assumed frequency of future droughts fallen.

References

Australian Government (2007) A national plan for water security. http://www.nalwt.gov.au/fi les/national_plan_for_water_security.pdf . Accessed 4 Feb 2011

COAG (1994) Water reform framework. http://www.environment.gov.au/water/publications/action/pubs/policyframework.pdf . Accessed 18 Feb 2011

Dixon P, Rimmer M, Wittwer G (2011) Saving the Southern Murray-Darling Basin: the economic effects of a buyback of irrigation water. Econ Rec 87:153–168

Table 6.2 Prices of permanent water rights ($ per ML, 2009 prices)

Year 2009 2010 2011 2012 2013 2014 2015 2016 Price 1,961 1,993 2,025 2,060 2,105 2,144 2,171 2,190

14 This trend is based on offer prices provided by the Australian Government throughout 2010: changes in expectations lagged the breaking of the drought by a year or so: http://www.environ-ment.gov.au/water/policy-programs/entitlement-purchasing/average-prices.html .

http://www.nalwt.gov.au/files/national_plan_for_water_security.pdf

http://www.nalwt.gov.au/files/national_plan_for_water_security.pdf



http://www.environment.gov.au/water/policy-programs/entitlement-purchasing/average-prices.html

http://www.environment.gov.au/water/policy-programs/entitlement-purchasing/average-prices.html


Frontier Economics (2009) Volumetric restrictions on water entitlement trade. A report prepared for the ACCC. http://www.accc.gov.au/content/item.phtml?itemId=891946&nodeId=1af536ef67d202bc227c59d94dcf1295&fn=Frontier%20Economics%20-%20Volumetric%20restric-tions%20on%20entitlement%20trade.pdf . Accessed 7 Feb 2011

Hone S, Foster A, Hafi A, Goesch T, Sanders O, Mackinnon D, Dyack B (2010) Assessing the future impact of the Australian Government environmental water purchase program. http://www.abare.gov.au/publications_html/landwater/landwater_10/waterbuyback.pdf . Accessed 18 Feb 2011

Keane B (2009) Xenophon’s deal turns water into gold. http://www.crikey.com.au/2009/02/16/xenophons-deal-turns-water-into-gold/ . Accessed 7 Feb 2011

Kenny M (2010) Gillard to commit to River Murray water buy-back. The Advertiser, August 10 MDBA (2010) Guide to the proposed Basin plan. http://download.mdba.gov.au/Guide_to_the_

Basin_Plan_Volume_1_web.pdf . Accessed 7 Feb 2011 NSW Irrigators’ Council (2010) Submission to productivity commission: market mechanisms for

recovering water in the Murray-Darling Basin. http://www.pc.gov.au/__data/assets/pdf_fi le/0011/94790/subdr072.pdf . Accessed 30 Aug 2010

Productivity Commission (2010) Market mechanisms for recovering water in the Murray-Darling Basin. Australian Government, Melbourne

Scheierling S, Loomis J, Young R (2006) Irrigation water demand: a meta analysis of price elas-ticities. Water Resour Res. doi: 10.1029/2005WR004009

Sunrice (2010) SunRice announces Deniliquin Mill will reopen in 2011. http://www.sunrice.com.au/uploads/MR_-_Deniliquin_Mill_to_Re-open.pdf . Accessed 7 Feb 2011

Victorian Farmers Federation (2008) Victorian farmers united against early cap review. http://www.vff.org.au/newsite/media_centre/media_print.php?id=570 . Accessed 7 Feb 2011

http://www.accc.gov.au/content/item.phtml?itemId=891946&nodeId=1af536ef67d202bc227c59d94dcf1295&fn=Frontier%20Economics%20-%20Volumetric%20restrictions%20on%20entitlement%20trade.pdf



http://www.abare.gov.au/publications_html/landwater/landwater_10/waterbuyback.pdf


http://www.crikey.com.au/2009/02/16/xenophons-deal-turns-water-into-gold/

http://www.crikey.com.au/2009/02/16/xenophons-deal-turns-water-into-gold/

http://download.mdba.gov.au/Guide_to_the_Basin_Plan_Volume_1_web.pdf

http://download.mdba.gov.au/Guide_to_the_Basin_Plan_Volume_1_web.pdf

http://www.pc.gov.au/__data/assets/pdf_file/0011/94790/subdr072.pdf

http://www.pc.gov.au/__data/assets/pdf_file/0011/94790/subdr072.pdf

http://www.sunrice.com.au/uploads/MR_-_Deniliquin_Mill_to_Re-open.pdf


http://www.vff.org.au/newsite/media_centre/media_print.php?id=570

http://www.vff.org.au/newsite/media_centre/media_print.php?id=570


Abstract The Australian government’s water buyback program started in earnest during a prolonged drought. TERM-H2O modeling indicates that in the short term, drought-induced job losses amount to around 6,000 jobs. Despite a recovery to aver-age seasons, depressed investment during drought lowers levels of farm capital in the long run. In turn, long-run employment in the region will remain around 1,500 jobs below forecast. In both the short and long terms, job losses arising from drought appear to be manyfold those arising from water buybacks.

Keywords Disinvestment • Excess capacity • Large inward supply shifts • Modeled v. observed impacts • Productivity losses

7.1 Introduction

The decade starting in 2000 was hotter than average and much drier than average in south-eastern Australia. In terms of the extent of drought, the period April–December 2002 was the second worst 9-month rainfall defi cit event on record in Australia, with 58.6% of the nation recording rainfall below the 10th decile (Bureau of Meteorology 2003 ) . In isolation, this was a signifi cant event, suffi cient to reduce livestock herd numbers, for example, across the nation. Discussions concerning drought proofi ng arose in the spring of 2002 as it became obvious that virtually all the continent, with the exception of several pockets in the sparsely populated

Chapter 7 The Economic Consequences of a Prolonged Drought in the Southern Murray-Darling Basin*

Glyn Wittwer and Marnie Griffi th

G. Wittwer (*) • M. Griffi th Centre of Policy Studies , Monash University , Wellington Rd, Clayton campus , Clayton , VIC 3800 , Australia e-mail: [email protected] ; marnie.griffi [email protected]

* This chapter reproduces with permission substantial portions of Wittwer and Griffi th (2011).

120 G. Wittwer and M. Griffi th

interior, was in drought. The Wentworth Group of Concerned Scientists ( 2002 ) responded with their blueprint for the future. Among the group’s themes was that we need to learn to live with drought. In the Murray-Darling basin, the group pro-posed reforms to water allocations and increased environmental fl ows. Some of these proposals became part of the 2007 Howard plan (see Sect. 6.1 ).

From 2003–04 to 2005–06, there was a partial recovery to near-average rainfall in the southern Murray-Darling basin (SMDB). Then the alpine regions of Victoria and New South Wales, which are the source of the Murray River, suffered record rainfall defi cits in the period from 2006–07 to 2008–09. 1 This resulted in recurrent reductions in water allocations throughout the SMDB. The Goulburn-Murray water authority’s allocations illustrate the severity of the fi rst decade of the new millen-nium: it formerly aimed to provide 100% allocations in 97 years out of 100 for the Goulburn system (though the authority removed this aim from its website early in 2010) but has failed to do so in fi ve out of eight irrigation seasons starting with 2002–03.

Extreme heat events were also recorded late in the decade. Adelaide, for exam-ple, experienced 15 consecutive days on which the maximum temperature exceeded 35°C in March 2008 (Bureau of Meteorology 2008 ) . Then there were two extraor-dinary heat events in January and February of 2009 separated by little more than week. Some stations in south-eastern Australia recorded their highest ever tempera-ture late in January, to have it exceeded again in February. Melbourne had three consecutive days exceeding 43°C in January and then recorded 46.4°C on 7 February 2009, on a day when bush fi res razed settlements on the fringes of Melbourne and beyond (Bureau of Meteorology 2009 ) .

Such conditions resulted in many stressful years for farmers in the Murray-Darling basin, as they dealt with repeated droughts, unprecedented shortfalls in water allocations, and the threat of bushfi res. These conditions also remind us of the setting in which policymakers proceeded with plans to supplement urban water sup-plies in each of the mainland state capitals (see Chap. 8 ). Investments that seem superfi cially prudent during prolonged periods of rainfall defi cits may appear to be extravagant when rain returns.

7.1.1 Focus of this Study

This chapter analyses the regional economic impacts of a prolonged period of recur-rent droughts using TERM-H2O. We concentrate on the regions of the southern Murray-Darling basin (SMDB). One issue concerns the dynamics of drought and drought recovery. In particular, we focus on the implications of drought for regional

1 A map showing rainfall deciles for the 3 years ending December 2008 is downloadable from http://www.bom.gov.au/jsp/awap/rain/archive.jsp?colour=colour&map=decile&year=2008&month=12&period=36month&area=nat .

http://www.bom.gov.au/jsp/awap/rain/archive.jsp?colour=colour&map=decile&year=2008&month=12&period=36month&area=nat

http://www.bom.gov.au/jsp/awap/rain/archive.jsp?colour=colour&map=decile&year=2008&month=12&period=36month&area=nat

1217 The Economic Consequences of a Prolonged Drought in the Southern…

investment and capital. Another issue is the broader regional economic and employment implications of drought. TERM-H2O models the interaction between irrigation and dryland agriculture in times of drought, allowing reallocation of resources across these two activities. Finally, modeling of the impacts of drought provides a benchmark for analysing the impacts of the Australian government’s water ‘buyback’ policy.

Some analysts and lobbyists have asserted that planned reductions in water used by irrigators in the Murray-Darling basin are similar to the effects of drought (Rizza 2010 ) . Regional impacts generated by various models including TERM-H2O (Dixon et al. 2010 ) and an ABARE-BRS model ( 2010 ) have been dismissed as understating the probable employment impacts of reducing allocations, most notably by now former Murray-Darling Basin Authority board members (Akerman 2010 ) . It would appear that water buybacks, which started during drought, were blamed for job losses which actually arose from drought. Therefore, there is some value in modeling the impacts of drought and estimated impacts on basin employment.

Drought is hard to model, as it entails substantial inward supply shifts for farm sectors. Large-change simulations are a challenge for modelers. Linear program-ming models are likely to reach unrealistic corner solutions with relatively modest supply shifts. CGE models that include CES functional forms will perform better, but most still struggle in large-change cases. Consequently, studies on CGE model-ing of drought are rare: the only previous studies of which we are aware are Sherony et al. ( 1991 ) and Horridge et al. ( 2005 ) , using a version of TERM without water accounts; and Pauw et al. ( 2011 ) . To depict the impacts of a drought as severe as that in southern Australia from 2006–07 to 2008–09 is an extreme test of a multi-regional CGE model. This paper outlines various theoretical modifi cations under-taken to improve the modeling of drought in a computable general equilibrium (CGE) framework and then applies the model to the period from 2005–06 to 2017–18. In particular, we apply a theory of excess capacity to downstream processing sectors.

7.2 Enhancing the Representation of Irrigation in TERM

The fi rst application of the original TERM was to the Australian drought of 2002–03 (Horridge et al. 2005 ) . The original model underestimated the observed change in the composition of farm output. The model did not include water accounts, did not distinguish between dryland and irrigation technologies, and therefore did not cap-ture factor mobility between dryland and irrigation activities. Despite its limita-tions, the model estimated statewide macro impacts reasonably well (Horridge et al. 2005 , Table 4).

Incremental enhancements to TERM started with the inclusion of water accounts (Wittwer 2003 ) . The database of a typical CGE model is based on an input-output structure designated in values. Irrigation water can vary greatly in price between users and years. It is necessary to include volumetric accounts so as to capture differences in water usage per dollar of output between different agricultural outputs.


Yet early applications of this version of TERM did not closely track observed changes in water usage between farm activities in response to changes in water availability. For example, using a version of TERM with water accounts, Young et al. ( 2006 ) modeled relatively modest declines in rice output in response to worsening water scarcity. This did not tally with available evidence. Water usage in rice production is highly responsive to changes in water scarcity: total water usage in the Murray-Darling basin dropped by 29% from 2001–02 to 2002–03, yet usage for rice pro-duction in the region dropped by 70% (Table 7.1 ). Following the drought of 2002–03, there was only one year prior to the end of the decade, namely 2005–06, in which water usage in rice production reached half of what it was in the years prior to 2002–03.

Chapter 5 details the theory of TERM-H2O which increases the responsiveness among irrigators to changes in water availability. Farm factors such as irrigable land, farm capital, and owner-operator inputs may move between irrigation and dryland technologies or between different irrigation sectors or different dryland sec-tors. Specifi c capital is immobile between sectors, refl ecting the relative infl exibility of perennial cropping. Water trading is permitted in TERM-H2O between regions in the southern basin. The main impacts of these theoretical modifi cations are to widen differences in the responsiveness of different activities to changes in water avail-ability while increasing farm factor mobility. This was a fi rst step in undertaking large-change simulations.

The next step in modeling irrigation sectors and regions in TERM-H2O was to move from a representation at the statistical division level to the statistical subdivi-sion (SSD) level. In the context of irrigation, the fi ner level of representation aligns more closely with catchment regions. This causes further modeling diffi culties. The statistical division level tends to include regions dominated in economic structure by large towns. This makes these regions more service intensive and less agriculture intensive than is the case for rural regions at the SSD level. 2 At the statistical division

Table 7.1 Water consumption (GL) by crop in the Murray-Darling basin, 2001–02 to 2005–06

2001–02 2002–03 2003–04 2004–05 2005–06

Livestock pasture 2,971 2,343 2,549 2,371 2,571 Rice 1,978 615 814 619 1,252 Cereals (excl. rice) 1,015 1,230 876 844 782 Cotton 2,581 1,428 1,186 1,743 1,574 Grapes and fruit 868 916 871 909 928 Vegetables 152 143 194 152 152 Other agriculture 504 475 596 564 460

Total agriculture 10,069 7,150 7,087 7,204 7,720

Source: ABS ( 2008 ) , Table 3.20

2 Farm income in the Murray statistical division in the 2006 TERM database accounted for around 12% of total regional income. One town, Albury, accounts for 40% of the population of Murray (ABS 2009 ) . Within the Murray statistical division, the farm share of GDP excluding Albury SSD (i.e., the Central Murray and Murray-Darling statistical sub-divisions) exceeds 20%.


level, farm output price rises are moderated by the impact on production costs of downstream processing sectors. But at the SSD level, not all regions contain substan-tial downstream processing sectors. Therefore, the higher concentration of farm activ-ity may result in farm output prices making a larger contribution to terms-of-trade impacts in the smaller regions without being offset signifi cantly by increased costs to downstream users in the same region. Consequently, the model without further modifi cations may predict unrealistically large terms-of-trade gains in rural regions.

The consumption function in TERM links nominal consumption to disposable income. Terms-of-trade gains affect the price of regional exports (inter-regional plus international) which are included in GDP but not consumption. Regional imports are included in household consumption but not GDP. Therefore, an increase in the price of regional exports relative to regional imports (a terms-of-trade gain) raises the ratio of regional real consumption to real GDP. There is a danger that we may model perverse real consumption gains in small regions in times of drought. Rectifying this requires a further theoretical modifi cation.

7.2.1 Why Not Make Demands for Farm Products More Elastic?

Pen-and-paper models often use the small country assumption in which demands are elastic. This simplifi es the impacts of inward supply shifts by guaranteeing that revenues fall as output decreases. But it can also lead to quite unrealistic results. For example, if Australia’s farm supply curve moves inwards due to drought and demands are elastic, would not imports substantially or entirely replace domestic supplies? In practice, there is a degree of substitution towards food imports during drought, but for most commodities, there is no evidence of a complete switch to imports. Armington ( 1969, 1970 ) helped modelers move towards more realistic results (i.e., away from fl ip-fl op solutions) by introducing the assumption of imper-fect import substitutability.

More generally, food products follow Engel’s law with income elasticities below one. Usual functional forms for household demands in CGE models (Stone-Geary or constant difference of elasticities forms, suitable for broad aggregations of com-modities where specifi c substitutability is not an issue) result in household demands for food around −0.5 or even smaller. TERM-H2O has export demand elasticities for most commodities other than wool of −4. 3 The higher the share of domestic consumption of Australia’s agricultural production, the more likely are outputs to have relatively low total share-weighted demand elasticities.

A CGE database includes details of sales for each farm commodity in a given year to downstream processors, households, and exports, plus the value of sales of

3 Dixon and Rimmer ( 2002 , pp. 222–225) derive a formula for export demand elasticities based on import substitution equations. This formula makes such elasticities in a national model consistent with the Armington parameters in a global model such as GTAP.


competing imports to each user. The database weights and various model parameters determine the total elasticity of demand for each farm output. Although much of Australia’s agriculture is export-oriented, drought-induced inward supply shifts reduce export supplies and thereby lower the total elasticity of demand for farm commodities by increasing the domestic share of total sales. Moreover, drought also increases demand for grains and hay as livestock feed, which pushes up local prices. Freight costs limit the extent to which farmers can purchase feed from distant sources. Imposing more elastic total demands on the model through higher import substitution parameters, higher export demand elasticities, and higher expenditure elasticities (the latter in violation of Engel’s law) may lead to fl ip-fl op solutions and move us further away from realism.

7.2.2 The Need to Model Excess Capacity in Downstream Sectors

To fi nd a way of depicting an extreme drought in a CGE model, we consider the impact of drought on downstream sectors. The ability of the downstream manufac-turers to cope with lower supplies of inputs depends on a number of factors.

For example, drought from 2006–07 to 2008–09 put dairy processors based in northern Victoria and southern New South Wales under fi nancial pressure that led to cost cutting via such measures as retrenchments. There was no substantial rationali-sation of capacity in the wake of the drought. A number of factors contributed to this. First, milk is produced Australia-wide, and processors have the option, though expen-sive, of transporting milk from non-drought-affected regions. For example, seasonal conditions were relatively favourable in northern New South Wales in 2007 and 2008, resulting in milk being transported south. As a means of reducing industry-wide transport costs, milk swaps between companies (where milk contracted to a given company is supplied instead to the nearest processor and swapped for milk elsewhere) have become commonplace. In addition, the changing feed-base away from irrigated pastures has lead to a fl atter pattern of milk production through the year, favourable to the production of the high-valued cheese relative to milk powder. This fl exibility in output mix helped maintain processor margins in the region.

Other industries do not have as many options. Whereas milk production out of the SMDB has fallen in the order of one third since its peak in 2001–02, rice output fell by 98% in 2007–08 relative to 2005–06, with no potential to prop up capacity utilisation by transporting in raw product from elsewhere. The Deniliquin rice mill, previously the biggest rice mill in the southern hemisphere, closed in 2008. In November 2010, Sunrice ( 2010 ) announced that that the mill would reopen in the coming months. With the return of average or above-average rainfall and a restora-tion of irrigation water allocations, water has once again become cheap enough to enable signifi cant levels of rice production.

A standard CGE model does not capture a reduction in capacity utilisation in downstream processing sectors in response to drought instead of solving for large inward farm supply shifts with consequent implausibly large farm output prices.


Far from modeling a drought-induced regional recession, there is a danger that spu-rious terms-of-trade gains will dominate the scenario. This is not to say that farm output prices do not increase in response to drought. Rather, such price hikes tend to be small relative to output declines. Drought usually is a time of rural hardship, not of regional windfall gains.

In initial attempts to analyse the impact of the global fi nancial crisis on the US economy, Dixon and Rimmer ( 2010 ) could not ascribe large inward macro demand shifts to their model without a consequent large real depreciation. There were no observed large real exchange rate adjustments to the US economy during the crisis. In response, the authors devised a mechanism to mimic excess capacity, motivated by the Keynesian theory of multiple equilibria, in which price adjustments alone will not dig an economy out of recession. The implementation of excess capacity solved the problem. By analogy, since the excess capacity mechanism choked off an unrealistic real depreciation in Dixon and Rimmer ( 2010 ) , we felt that it could also subdue modeled farm output price hikes within TERM-H2O in an extreme drought simulation. That is, allowing excess capacity in downstream processing sectors would reduce their demand for farm inputs as the scarcity of inputs worsened due to drought .

Dixon and Rimmer ( 2010 ) depicted excess capacity via a theory of sticky capital adjustment. The usual theory (i.e., constant returns) is that industries operate at full capital so that used capital (KU

jr,t for industry j in region r and time period t ) is equal

to existing capital (KE jr,t

). With a sticky rental adjustment assumption, we can think of the rental rate as a profi t markup on variable costs. This markup will adjust down-wards slowly in response to excess capacity.

, , 1 ,

,, , 1 ,

1 1 1-

-

æ ö æ ö æ ö- = - + - +ç ÷ ç ÷ ç ÷

è ø è ø è øjr t jr t jr t

jr tjr t jr t jr t

R R KUS

Rf Rf KEa (7.1)

, ,( )jr t jr tR f KU= (7.2)

In ( 7.1 ), R jr,t

and Rf jr,t

are the rental rates for industry j in region r and year t in the respective policy and forecast runs. 4 S

jr,t is a slack variable which implements ( 7.1 )

and a a positive parameter. Equation 7.2 is the capital demand equation in which f is a decreasing function of KU

jr,t . During drought, we invoke the sticky rental adjust-

ment mechanism for downstream processing industry j (i.e., S jr,t

= 0). This means that used capital KU

jr,t falls relative to existing capital KE

jr,t . Instead of responding

to reduced farm output by paying much higher input prices, processors reduce capi-tal utilisation. This is equivalent to an inward movement in processing supply curves and an accompanying reduction in demand for farm inputs. While this will have little impact on processing sector output prices, it will reduce the demand for and moderate scarcity-induced price hikes of farm inputs and consequently moderate the fall in the rate-of-return on capital in the processing sector. In turn, smaller farm

4 In dynamic modeling, we run a baseline forecast and a policy run. In the case of an adverse event such as drought, the ‘policy’ run is more accurately labelled the ‘perturbation’ run.


output price hikes will moderate terms-of-trade effects in small regions during drought. When better seasons return, the industry resumes full-capacity utilization. In the full-capacity state, S

jr,t ³ 0 and KU

jr,t = KE

jr,t . The change of state between

full capacity (with market-clearing rate-of-return adjustments) and excess capacity (a quantity adjustment) requires the use of a complementarity condition, implemented in the model using GEMPACK software, as described by Harrison et al. ( 2004 ) .

Why did we not model excess capacity via the inclusion of fi xed costs? A num-ber of CGE applications have done so, including Harris ( 1984 ) , Abayasiri-Silva and Horridge ( 1998 ) , and other studies downloadable from the GTAP website: Kharitonov and Walmsley ( 2004 ) , Kuik and Gerlagh ( 2005 ) , and Hertel and Swaminathan ( 1996 ) . Applications that include fi xed costs in the model formulation typically apply to relatively small change cases. The inclusion of fi xed costs would have further complicated TERM-H2O without guaranteeing a solution in large-change simulations. Instead, we followed Dixon and Rimmer ( 2010 ) , who accepted excess capacity as a real world phenomenon. Their method concentrates on obtain-ing a plausible model solution in a large-change case without attempting to explain excess capacity.

7.3 Drought in South-Eastern Australia from 2006–07 to 2008–09

Bureau of Meteorology data indicate that the entire SMDB basin had either decile 1 rainfall or the lowest on record for the 3-year period between July 2006 and June 2009 (see Footnote 1). Recurrent droughts affected both dryland and irrigated pro-duction. Dryland production was most adversely affected in 2006–07 and 2007–08, with a partial recovery in some regions in 2008–09. For irrigators, the impacts of catchment shortfalls on water allocations continued until the fl ood events of 2010–11. Table 7.2 shows the modeled percentage shortfalls in water availability by region for 2007–08.

We use TERM-H2O with a theory of sticky capital adjustment in some down-stream processing sectors to simulate drought impacts. The exogenous policy shocks are the estimated direct impacts on both dryland productivity and irrigation water allocations for 2006–07 to 2008–09, with an assumed recovery in dryland produc-tivity in 2009–10 and eventual full recovery in water allocations by 2011–12 (the modeling was completed before the fl oods of 2010–11).

7.3.1 Comparing Naïve Calculations and Modeled Impacts for 2007–08

We start with an analysis of our results for 2007–08. Lack of rainfall in 2007–08 meant that dryland productivity in the SMDB was below average. Irrigation alloca-tions were at a low point after two successive years of drought. We can calculate a


Tabl

e 7.

2 Im

pact

s of

dro

ught

by

regi

on, 2

007–

08 r

elat

ive

to n

o-dr

ough

t bas

elin

e (%

)

Reg

ion

(see

Fig

. 7.1

) 1

2 3

4 5

7 8

9 10

11

12

13

15

SM

DB

Wat

er a

lloca

tions

and

pro

duct

ivity

leve

ls (

100

= av

erag

e)

(1)

Dry

land

pro

duct

ivity

a 42

42

42

42

42

36

36

69

69

69

69

69

36

51

(2

) W

ater

b 51

51

14

14

40

42

42

46

40

46

46

45

60

44

Con

trib

utio

ns to

GD

P in

200

5–06

bas

e (%

) (3

) D

ry la

nd

8.3

8.4

6.4

2.3

8.0

13.6

14

.4

3.9

1.4

6.9

7.4

3.8

8.0

6.8

(4)

Irri

gatio

n 1.

9 15

.3

1.2

19.6

12

.1

8.0

14.5

1.

5 0.

7 9.

2 3.

2 2.

8 14

.1

6.1

(5)

Tota

l 10

.2

23.7

7.

6 21

.9

20.1

21

.6

28.9

5.

4 2.

1 16

.1

10.6

6.

6 22

.1

12.9

Naï

ve e

stim

ates

of

cont

ribu

tions

to G

DP

(%)

(6)

Dry

land

−

4.8

−4.

9 −

3.7

−1.

3 −

4.6

−8.

7 −

9.2

−1.

2 −

0.4

−2.

1 −

2.3

−1.

2 −

5.1

−3.

3 (7

) Ir

riga

tion

−0.

9 −

7.5

−1.

0 −

16.9

−

7.3

−4.

6 −

8.4

−0.

8 −

0.4

−5.

0 −

1.7

−1.

5 −

5.6

−3.

4 (8

) To

tal

−5.

7 −

12.4

−

4.7

−18

.2

−11

.9

−13

.3

−17

.6

−2.

0 −

0.9

−7.

1 −

4.0

−2.

7 −

10.8

−

6.7

Mod

eled

con

trib

utio

ns b

y br

oad

sect

or

(9)

Dry

land

−

4.4

−2.

6 −

3.1

0.1

−3.

0 −

9.0

−9.

2 −

0.7

−0.

3 −

0.9

−1.

4 −

0.8

−5.

6 −

2.7

(10)

Irr

igat

ion

−1.

0 −

10.2

−

0.3

−13

.2

−2.

0 −

1.1

−0.

5 −

0.5

−0.

1 −

2.1

−1.

0 −

0.8

−0.

5 −

1.9

(11)

Foo

d −

0.3

−0.

6 −

0.1

−0.

2 0.

0 −

0.1

−0.

2 −

0.4

−0.

1 −

0.7

−0.

1 −

0.2

−0.

6 −

0.3

(12)

Res

t −

1.1

−0.

5 −

0.7

−1.

6 −

0.6

−1.

7 −

1.3

−0.

2 −

0.3

−0.

6 −

0.4

−0.

9 −

0.9

−0.

8 (1

3) N

et w

ater

0.

4 3.

8 −

0.2

−4.

9 −

1.5

−0.

7 −

1.6

0.1

0.0

0.6

0.3

−0.

2 −

0.7

0.0

(14)

GD

P −

6.4

−10

.0

−4.

4 −

19.7

−

7.1

−12

.7

−12

.8

−1.

6 −

0.7

−3.

6 −

2.6

−2.

8 −

8.3

−5.

7 (1

5) N

et w

ater

sol

d (G

L)

83

456

−38

−

194

−33

−

29

−86

5

−4

−10

4 2

−20

−

39

0

a Aut

hors

’ es

timat

es b

ased

on

rain

fall

defi c

ienc

ies

b Dat

a pr

ovid

ed b

y M

urra

y-D

arlin

g B

asin

Aut

hori

ty


naïve or fi rst-guess estimate of the contribution of a farm subset k of all industries j to a percentage change in GDP in region r (gdp

r ) as:

gdp (PRIM . ) / PRIM= å år kr kr jrk j

q (7.3)

PRIM is the level of value-added output of each sector and q is the percentage change in output. As a starting point for our naïve calculation, we assume that for irrigation sectors i , q

i = xwat

i , where the latter is the percentage difference in water

allocations from normal. Additionally, our naïve calculation of lost output in dry-land sectors j equals the technological deterioration due to drought (aprim

j ) so that

q j = aprim

j . Our initial estimate of the impact of drought, in which a refers to all

industries in region r is:

[ ( ) ( )] / PRIgdp PRIM . PRIM . Må + å= å ari j

r ir ira

jr jrq q (7.4)

In Table 7.2 , row (1) shows dryland productivity and row (2) an index of water availability relative to a normal year. Rows (3)–(5) provide estimates of the contri-butions of dryland plus irrigation farming to GDP in each region. Based on 7.4, rows (6)–(8) contain our fi rst-guess contributions of irrigation and dryland sectors to changes in real GDP in the regions of SMDB. The modeled contributions to changes in regional GDP by broad sector and irrigation water are shown in rows (9)–(14). Row (15) shows the volume of net water sold by region.

If there were no movements of farm factors including water between sectors, the fi rst-guess impacts on farm sectors shown in rows (6) and (7) would equal the TERM-H2O impacts shown in rows (9) and (10) of Table 7.2 . This would be equiv-alent to CET parameters for farm sectors being set to zero, with zero substitutability between water and other primary factors. Comparing rows (6) and (9), we see that dryland fi rst-guess losses predict modeled broad sectoral losses quite closely in some but not all regions. Variations arise from some resource movements. In Lower Murrumbidgee, farm factors move from irrigated towards dryland production as irrigation water is exported to other regions. 5 Note that although the contribution of water to regional GDP increases during drought, as its demand is inelastic and therefore its value rises as its availability falls, the contribution to real GDP of water trading is zero. That is, we see the benefi ts of water trading by comparing the fi nal column entries for rows (10) and (7) in Table 7.2 : due substantially to water trading (and to a lesser extent, factor substitution away from water), the contribution of irrigation sectors to real GDP in SMDB is −1.9% instead of −3.4% as given by the fi rst-guess calculation.

As mentioned, by comparing rows (9) and (10) with rows (6) and (7) in Table 7.2 , we see the impact of imposing non-zero farm factor CET and CES parameters (and water trading) on the results. What difference would altering the parameters make

5 The solution procedure is Euler 60 steps (and Euler 256 steps in the fi rst 4 years of drought and recovery), used to eliminate solution errors from the linearized model in this large-change simulation (Dixon et al. 1982 , chapter 5).


to results? Our experience with TERM-H2O is that the most important modifi cation necessary to capture observed movements of water as water scarcity changes or relative farm output prices change is the inclusion of CET farm factor movements. In addition, moving from full trading in the SMDB to no inter-regional trading tends to have a bigger impact than parametric variation (within reasonable limits), especially if there are pronounced differences in water allocation shortfalls and productivity losses between regions. Rather than present results of parametric varia-tion here, we instead compare modeled results with actual ABS data later in this section.

We might expect modeled GDP losses in each region to be somewhat larger than our naïve calculation of losses. This is through negative impacts on downstream sectors, and the impact of reduced household consumption on service sectors in each region. Modeled GDP losses are larger than our fi rst-guess calculation of losses for some but not all regions shown in Table 7.2 . Water trading between sectors and regions, combined with mobility of farm factors, alleviates some of the losses. For example, Lower Murrumbidgee is a substantial exporter of water to other regions in the drought years of the scenario. The movement of factors including water partly offsets productivity losses and water allocation shortfalls so that the modeled GDP loss is smaller than the fi rst-guess calculation of the GDP loss in this region. Changes in output by sector in part refl ect differences in water’s share of total costs, but are also infl uenced by different input-substitution possibilities. For example, the dairy and other livestock sectors can substitute between land and cereal inputs.

Finally, in Table 7.2 we see that the impacts on downstream sectors, although negative, do not imply large regional multipliers. Regional aggregate consumption falls relative to forecast due to drought. In the short-term, housing which accounts for a substantial share of consumption is fi xed in supply. Therefore, all adjustments are going to be on housing rentals. Prolonged adverse conditions in MDB regions would lead to long-run quantity adjustments in housing and enlarged negative multipliers overall, but in the short-term, price adjustments reduce the size of mul-tipliers driven by the spending effect.

7.3.2 Comparing Modeled and Observed Impacts

Table 7.3 compares modeled outcomes for farm products in the SMDB with avail-able data on percentage changes in 2007–08 relative to 2005–06. Columns (1)–(3) show the modeled deviations from forecast due to drought (versus a hypothetical no-drought baseline for 2007–08), and columns (4)–(6) estimated actual basin-wide changes from 2005–06 to 2007–08. Hence, the comparisons are not between like and like, but are the best we can do.

Cereal production did not shrink as much in the observed period as we modeled. This refl ects soaring cereal prices in the observed period: the actual price hike was twice the modeled drought-induced price hike. World prices of cereals in 2007–08 were driven up by increased use of biofuels and other international developments


beyond price hikes arising from drought within Australia. The output outcome for grapes turned out better than modeled, again with observed prices rising more than the modeled deviation. Dairy cattle’s output and use of water dropped more than we modeled. There was also a larger-than-modeled movement from irrigated to dryland production (implied by the fall in water use relative to output). Since dairy output prices were high in 2007–08, dairy producers were willing to move to dryland and pay for cereal feed (grains and hay). Although the observed value of dairy cattle output rose, with the price increase more than offsetting the output decrease percent (Table 7.3 , columns (4) and (5)), the value-added almost certainly dropped signifi -cantly, refl ecting high feed costs.

There was a greater movement of water out of rice production than we modeled. Other than rice (for which the commodity price hike was smaller than modeled), dairy and other agriculture did worse than modeled. The other agriculture sector includes nursery products: Australia’s mainland capitals with the exception of Darwin all faced water restrictions in this period which drove down demand for this sector from household gardeners. Observed vegetables output did better than the modeled outcome with actual prices rising more than modeled.

One sector in which the model did not explain changes in outputs at all well was other livestock. In a drought scenario, farmers face the choice between keeping livestock alive through fodder purchases, selling livestock or in the worse circum-stances, culling herds. Selling livestock, a form of destocking, will increase the supply for processing, which will increase short-term output at the expense of long-term output. Destocking drives down the price of livestock output. TERM-H2O has no theory to capture destocking so that the observed short-term price and output impacts were in the opposite direction to modeled impacts. A motivation for culling in response to drought is that destocking drives down the price of livestock to the

Table 7.3 Comparing modeled SMDB outcomes to observed changes

Modeled outcome deviation from 2007–08 base (%)

Observed 2007–08 relative to 2005–06 a (%)

Output b Price Water used c Output d Price Water used c

(1) (2) (3) (4) (5) (6)

Cereal −55.3 43.6 −78.8 −45.8 92.1 −9.9 Rice −84.9 86.2 −90.7 −98.2 46.3 −97.8 Dairy cattle −13.6 29.5 −40.9 −26.5 52.0 −64.4 Oth. livestock −23.1 41.4 −44.6 −1.2 −9.2 −70.6 Grapes −17.9 18.0 −49.0 2.7 44.6 −15.7 Fruit −7.7 13.5 −23.1 9.3 7.3 −13.7 Vegetables 3.5 6.8 −1.4 21.8 14.9 −18.4 Oth. agri. 17.3 7.9 12.6 NA NA −27.0

Source: ABS catalogue no. 7125.0, Anderson et al. ( 2010 ) , and ABARES ( 2010 ) a Entire Murray-Darling basin b Value-added basis c Water used in irrigation production d Value of output, not value-added


point at which the costs of transportation and handling exceed the price paid for livestock.

Next, we examine water prices. We would expect water prices to have increased between 2005–06 and 2007–08 by a larger amount than modeled, due to the observed surge in commodity prices for some major irrigation products. This is so: the mod-eled increase was $285 per megalitre relative to forecast in 2007–08, compared with an observed increase in the Goulburn region relative to 2005–06 of around $500 per megalitre. A weakening of commodity prices in 2008–09 resulted in the price of water falling to $275 per megalitre above 2005–06 levels, closer to the modeled outcome. 6

The Australian Water Market Report for 2007–08 shows an observed pattern of net downstream trade (National Water Commission 2009 ) . Small amounts were transferred out from the upper Murray reaches in both NSW and Victoria (i.e., the Ovens-Murray and Albury-Upper Murray regions, Fig. 7.1 ), and larger amounts from the Goulburn and lower NSW Murray reaches. The largest net seller was the Murrumbidgee valley, refl ecting the infl uence of rice. Rice is grown in better years; however, a moderate worsening of water scarcity is suffi cient to make it more profi t-able for growers to sell their water allocation for a year than to continue growing rice. The buyers of water were the Victorian Mallee regions and most notably

1714

15

18

6

165 2

7 14

1312

39 11

8

10

Fig. 7.1 Map of SMDB regions in TERM-H2O. Regions: 1 Wagga-Central Murrumbidgee, 2 Lower Murrumbidgee, 3 Albury-Upper Murray, 4 Central Murray, 5 Murray-Darling, 6 rest of Vic, 7 Mildura-West Mallee, 8 East Mallee, 9 Bendigo-North Loddon, 10 South Loddon, 11 Shepparton-North Goulburn, 12 South/South West Goulburn, 13 Ovens-Murray, 14 QLD, 15 rest of SA, 16 Murraylands SA, 17 rest of Australia, 18 rest of NSW

6 Watermove weekly data, downloaded from www.watermove.com.au , authors’ calculations.

http://www.watermove.com.au


South Australia. In terms of how the modeling replicated this pattern, the main differences are that the model projected higher than observed net water exports from the Murrumbidgee regions, and lower than observed net imports to South Australia. The latter was due partly to purchases by the South Australian government. In 2008–09, the observed pattern of water trading moved closer to that modeled by TERM-H2O. The Ovens-Murray region became a net importer of water as modeled (Table 7.2 , row (14)).

7.3.3 Dynamic Analysis of Drought Followed by a Prolonged Recovery

Our simulation consists of widespread drought conditions from 2006–07 to 2008–09, with a dryland recovery in 2009–10 but some delay before the restoration of full water allocations for irrigation sectors. As shown in Table 7.3 , real GDP in SMDB fell 5.7% below forecast in 2007–08 due to drought. The simulated out-come for 2009–10 remains below forecast due to irrigation water not being fully restored.

In subsequent years, as a consequence of sharp falls in investment during the drought years, aggregate capital stocks persist below forecast (−0.12% in 2017–18). Similarly, employment does not recover fully in the simulation period. SMDB employment fell to 1.3% below forecast in 2007–08, equivalent to 6,000 jobs. Even in 2017–18, long after the recovery, employment persists at 0.36% (around 1,500 jobs) below forecast (Fig. 7.2 ). 7

7 National employment is exogenous in the policy simulation. Section 3.2.3 discusses the choice of regional labour market theory in TERM-H2O

-6.0

-5.0

-4.0

-3.0

-2.0

-1.0

0.0

1.0

2005

-06

2017

- 18

2014

- 15

2011

- 1220

08 -

09

RealconsumptionRealinvestmentReal GDP

Employment

Capital stocks

Fig. 7.2 Macroeconomic outcomes for SMDB (% deviation from forecast)


Figure 7.3 shows the impact of drought on downstream processing sectors in SMDB. In the initial year of drought (2006–07), capital stocks do not change in response to drought. They are predetermined by the link between current period capital stocks, lagged capital stocks net of depreciation, and lagged investment. However, by introducing a theory of excess capacity to the model, we allow a tem-porary gap to occur between available and used capital for downstream processing sectors. This gap gradually closes between 2006–07 and 2008–09 as falling invest-ment erodes the capital base. With a substantial recovery in 2009–10, the gap is eliminated and the usual theory of constant returns to scale is resumed within the model. The prolonged drought and irrigation water allocation shortfalls have a neg-ative impact on farm investment so that farm capital persists at almost 2% below forecast in 2017–18.

7.3.4 Explaining an Unexpected Regional Outcome

One strange result is that Lower Murrumbidgee, in terms of aggregate consump-tion, appears to do worse relative to forecast in the fi rst year of recovery from drought than during the drought (Table 7.4 , comparing 2008–09 and 2009–10).

-6.0

-5.0

-4.0

-3.0

-2.0

-1.0

0.0

2005

-06

2008

-09

2011

-12

2014

-15

2017

-18

Available Used Farm capital

Fig. 7.3 Downstream processing and farm capital, SMDB (% deviation from forecast) (Used and available capital in SMDB for the aggregate of meat products, dairy products, wine and other beverages, and fl our and processed cereals)

Table 7.4 GDP defl ator, income and household spending in Lower Murrumbidgee

Contributions to defl ator 2006–07 2007–08 2008–09 2009–10

Consumption −1.4 −1.0 −0.7 0.3 Other absorption (I + G) −0.4 −0.1 0.0 0.2 Exports 8.7 8.9 8.5 −0.2 Imports −2.9 −3.0 −3.3 −0.1 Net water exports 2.2 3.9 2.7 −0.5 GDP defl ator 6.2 8.6 7.2 −0.4 Aggregate consumption −3.2 −0.6 −1.0 −3.2 Real GDP −10.4 −10.0 −8.1 0.6


To explain this result, we need to examine prices as they affect producers and con-sumers in the region. We have the following regional GDP identity:

( ) ( ) ( )waterGDP C I G X M X M= + + + - + - (7.5)

In addition to the usual components of GDP (consumption C , investment I , government spending G , and inter-regional plus international trades), there is another source of trades – water. We calculate real GDP as the share-weighted sum of the percentage changes in the volumes of ( 7.5 ). We can calculate the contribu-tions of the components to the GDP defl ator similarly, as shown in Table 7.4 . The two main contributors to the change in the GDP defl ator relative to forecast are exports and net water exports. In 2008–09, these two components contribute to a rise of 11.2% in the GDP defl ator (= 8.5% + 2.7%, Table 7.4 ). Exports consist of farm outputs produced in Lower Murrumbidgee. The CPI falls in the region due to drought, refl ected mainly in a fall in housing rentals. The substantial gap between the GDP defl ator (+7.2% in 2008–09) and CPI (with a share-weighted contribution of −0.7% to the GDP defl ator in 2008–09) increases the ratio of real household consumption to real GDP, as the consumption function links nominal consumption and nominal disposable income (which is closely related to nominal GDP). Therefore, aggregate consumption declines by only 1.0% in 2008–09 despite real GDP falling by 8.1%.

When we move from the last of the drought years (2008–09) to a recovery year (2009–10), there is a sharp fall in the price of water, which makes a negative contri-bution to the GDP defl ator in Lower Murrumbidgee (since it continues to be a net exporter of water). 8 At the same time, farm output prices fall back to close to baseline forecast levels. The contribution of commodity and water exports to the GDP defl ator drops from 11.2% in the last of the drought years to −0.7% in 2009–10. This contri-bution is larger than the jump in real GDP from 8.1% below forecast in 2008–09 to 0.6% above forecast in 2009–10. Moreover, CPI rises relative to forecast giving a contribution of 0.3% to the GDP defl ator (Fig. 7.4 ) (Table 7.4 ).

This strongly agricultural economy which obtains substantial income from water trading during drought does better than most drought-affected regions during drought. But in the recovery years, employment and capital stocks do not return to forecast levels. There is a small structural change in Lower Murrumbidgee’s econ-omy: net water exports remain above forecast. This refl ects compositional change across SMDB as a consequence of the drought. Farm output remains below forecast in 2017–18. Employment is 0.57% or 120 jobs below forecast in 2017–18, with services output also below forecast (Fig . 7.5 ).

Water moves into irrigated non-dairy livestock production in SMDB in the long term relative to forecast. The irrigated sector is small relative to dryland non-dairy

8 A regional terms-of-trade variable, with separate commodity and water contributions, has been added to later versions of TERM-H2O. During drought, the water component makes a substantial positive contribution to Lower Murrumbidgee’s terms of trade.


livestock in SMDB, unlike for dairy which is mainly irrigated. Overall, farm output remains below forecast in SMDB, refl ecting the fall in farm capital relative to forecast (Figs. 7.4 and 7.6 ).

7.4 Drought Without Reduced Water Allocations

In the drought of 2002–03, there were not substantial reductions in water allocations in most catchment regions. A notable exception was the Goulburn region. One insight that has arisen out of TERM-H2O modeling is that even if allocations remain

-14.0

-12.0

-10.0

-8.0

-6.0

-4.0

-2.0

0.0

2.0

4.0

2005

-06

2008

-09

2011

-12

2014

-15

2017

-18

Agriculture

Downstream

Services

Water exportsEmployment

Fig. 7.4 Employment and industry contributions to GDP, Lower Murrumbidgee (% change relative to forecast)

-5.0

-4.5

-4.0

-3.5

-3.0

-2.5

-2.0

-1.5

-1.0

-0.5

0.0

2005

-06

2008

-09

2011

-12

2014

-15

2017

-18

Agriculture Downstream Services, other

Fig. 7.5 Industry contributions to GDP, all SMDB (% change relative to forecast)


at average levels, the value of the marginal product of water rises during drought. One reason for this is that irrigation water requirements increase as rainfall reduces. A 100-mm shortfall in rainfall adds one megalitre per hectare to irrigation water requirements, thereby pushing up the price of water. Another reason is that drought-induced declines in dryland productivity will move factors of production into other industries, including irrigation activities. This raises the ratio of non-water factors to water, thereby increasing the marginal product of water.

We might think of 2002–03 as being a year of dryland drought with near-normal water allocations although allocations were lower than in 2001–02 (Table 7.1 ). 2002–03 entailed similar aggregate water allocations across the basin as 2005–06 (i.e., 7,150 GL in 2002–03 and 7,720 GL in 2005–06 compared with 10,069 GL in 2001–02), but the composition of use was quite different. The main difference was that 2002–03 was a year of widespread drought and 2005–06 was not. Water used in irrigated cereals in 2002–03 was higher than in 2005–06, but water use in rice production was much lower in 2002–03 than 2005–06. From this, we can infer that the price of water was much higher in 2002–03 than 2005–06: rice producers are highly responsive to water price changes, fi nding it more profi table to sell water than use it in production as its price rises. Indeed, the water price in the Goulburn region in 2002–03 averaged $364 per megalitre, compared with only $57 per mega-litre in 2005–06. As discussed in Sect. 7.3.2 , commodity prices also infl uence water usage. Had cereal (non-rice) prices been high or rising in 2005–06, water may have moved from rice to other cereals with a rising water price.

What are the policy implications of this? If we had included future droughts in the baseline of the buyback scenario elaborated in Chap. 6 , in each year of drought, there would have been an upward spike in the water price in both the baseline and policy runs. In modeling undertaken for the MDBA of water buybacks, Wittwer

0

50

100

150

200

250

300

2010

2013

2016

2019

2022

2025

Baseline Policy

$/ML

Drought years

Fig. 7.6 Price of water in buyback scenario: droughts in 2015 and 2020 (Source: Wittwer 2010 , Fig. 1)


( 2010 ) included baseline droughts in 2015 and 2020, resulting in such spikes (Fig. 7.7 ). 9 Similarly, buybacks resulted in larger deviations in irrigation output in drought years from forecast than non-drought years (Fig. 7.6 ). Section 6.4 includes expected future droughts in the calculation of the asset value of water. As the expected frequency of droughts increases, the asset value of water increases. The purely economic costs of buybacks will also increase with the frequency of drought. The increase in asset value and increase in costs may occur even if water allocations do not fall. This may be possible in decades when droughts are neither extreme nor consecutive.

7.5 Regression Analysis of Water Prices

The years from 2001–02 to 2009–10 provide a wide range of water allocation vol-umes, farm output prices and rainfall conditions in the southern Murray-Darling basin. In ( 7.6 ), we report a regression of trade-weighted annual water prices ( P

wat,t ,

compiled using Watermove data) against allocation volumes for the southern basin ( V

alloc,t based on ABS ( 2011 ) data), a drought index ( D

t ), based on observed rainfall

defi cits for the 9-month period March–November from Bureau of Meteorology data (i.e., the index for 2007–08 is based on the rainfall defi cit for March–November 2007) and a price index of farm outputs ( P

farm,t ), based on ABARES ( 2010 ) data.

According to TERM-H2O results, the price of irrigation water is highly sensitive to drought conditions and moderately sensitive to the volume of irrigation water allocated each year. The model also predicts that a strengthening of farm output

-1000

-800

-600

-400

-200

0

200

400

600

2010

2013

2160

2019

2022

2025

Dry-land Irrigated

Fig. 7.7 MDB farm output: buyback scenario ($m output relative to forecast) (Source: Wittwer 2010 , Fig. 3)

9 Wittwer ( 2010 ) modeled the impacts of buybacks purchased gradually between 2011 and 2022. The baseline differed from that of Chap. 6, in terms of water availability and consequent prices.


prices will lead to a hike in the rental of farm factors, including the water price. We are able to check whether these predictions align with observed data.

We do so by estimating a regression of observed prices against explanatory vari-ables, using data shown in Table 7.5 for the southern basin. Column (1) shows the price of irrigation water ( P

wat, t ), 10 column (2) the allocation of irrigation water ( V

alloc, t )

in the southern basin and column (3) a drought index D t , based on observed rainfall

defi cits for the 9-month period March–November (i.e., the index for 2007–08 is based on the rainfall defi cit for March–November 2007). Column (4) shows a price index of farm outputs ( P

farm, t ), based on ABARES indexes, modifi ed to refl ect pro-

duction weights in the basin. We use columns (2)–(4) to explain variations in the price of water :

2wat,t alloc,t t farm,t

( 2.97) ( 4.41) (7.04) (2.35)t statlog P 1.629 0.129* /1000 0.568* D 0.009* P R 0.98V

- --= - + + = (7.6)

In ( 7.6 ), each of the coeffi cients on the explanatory variables has the expected sign. As water allocations increase, the price of water falls. The presence of drought imposes dramatic upward pressure on the water price. The coeffi cient on farm out-put prices is positive as expected.

The alignment so far of TERM-H2O results with actual data is encouraging. As part of further model calibration, our next step will be to run TERM-H2O ascribing dryland productivity shocks and water availability shocks year-by-year in the south-ern basin, using the data in columns (2)–(4) of Table 7.5 as the basis of these shocks. The water prices and changes in the composition of farm output in the southern basin generated by this exercise will enable us to fi ne tune TERM-H2O, thereby moving from a qualitative to quantitative checking of the model’s performance.

Table 7.5 Data used in irrigation water price regression

$/ML GL Drought index P (output)

P wat, t

(1) (2) (3) (4)

2001–02 35.00 7,477 0 102.7 2002–03 364.02 4,856 1.0 101.5 2003–04 66.63 5,551 0 97.2 2004–05 60.03 5,622 0 96.0 2005–06 57.25 6,585 0 100.0 2006–07 440.59 3,639 0.75 115.4 2007–08 562.16 2,682 0.4 129.8 2008–09 338.57 2,703 0.5 114.9 2009–10 153.52 4,237 0 111.4

Sources : (1) Watermove; (2) NWC data scaled to ABS and authors’ estimates; (3) Bureau of Meteorology; (4) ABARES Commodity Statistics 2010

10 This is based on data from the Goulburn basin only. Anecdotal evidence indicates a close corre-spondence between prices across regions in the southern basin where inter-regional trading is possible.


7.6 Conclusions

The signifi cant contribution of this chapter is to model very large inward supply shocks to estimate the impact of drought in the SMDB on regional economies. Drought is an inevitable part of farming, but few studies report on the economy-wide modeling of drought impacts. Modelers face obstacles in reaching realistic solutions in drought simulations. In TERM-H2O, it is necessary to introduce excess capacity to downstream processing sectors in order to keep farm output price hikes within realistic bounds in response to drought-induced shrinkages in farm supplies.

The main fi nding of this study is that in the short term, drought across the SMDB reduces employment relative to forecast by around 6,000 jobs. Even after a return to average seasons, the impacts of drought remain. Depressed farm investment during drought results in farm capital persisting below baseline levels, even many years after drought has ended. Consequently, employment in SMDB does not return to baseline levels but remains at 1,500 jobs below forecast in 2017–18.

It is possible to check some results against actual outcomes. TERM-H2O results relative to forecast for 2007–08 give a reasonable account of observed changes from 2005–06 to 2007–08. Most differences between modeled and observed outcomes arise from global conditions that were not included in the drought scenario. For example, world grain prices rose strongly in 2007–08. For grains, TERM-H2O pre-dicted a larger decline in output and smaller hike in prices than observed. Consequently, TERM-H2O underestimated the dollar per megalitre rise in the price of irrigation water for 2007–08, yet tracked the irrigation water price reasonably in 2008–09 when grain prices fell.

At the sub-state level, data on employment numbers are hard to obtain between censuses. The ABS conducts the Labour Force Survey regularly but data are state-wide estimates. Beyond anecdotes and other employment estimates that are certain to be patchy rather than comprehensive, a better picture of the impact of drought on SMDB may have to wait until small region employment by industry numbers appear after the 2011 census.

TERM-H2O modeling indicates that employment in SMDB may persist at 1,500 jobs below forecast in 2017–18 in the wake of a drought that occurred a decade earlier. The same model indicated only modest job losses in SMDB arising from increased environmental fl ows implemented through buyback (Chap. 6 ). No other model used to estimate the regional employment consequences of buyback or simi-lar policy proposals has been tested in a drought scenario.

It appears highly probable that if the current policy of purchasing water from farmers for environmental fl ows continues in the Murray-Darling basin, the claims of job losses arising from the policy amounting to many thousands will not stand up to scrutiny. Even using a simple calculation based on farm shares of regional GDP, the regional economic impacts of drought are manyfold worse than the probable impacts of water buybacks. It is possible that a high Australian dollar will impose greater diffi culties on SMDB farmers in the present decade than either drought or water buyback policy.


From the perspective of both the environment and irrigators, with the benefi t of hindsight, it would have been preferable to introduce environmental fl ows before the fi rst decade of the new millennium, with its repeated droughts. This proved not to be possible under the slow and diffi cult process of policy evolution. Indeed, the Wentworth Group of Concerned Scientists ( 2002 ) revealed the need for further pol-icy reforms going into the decade. Had there been a lower volume of highly secure irrigation allocations leading into the past decade, there would have been fewer farmers caught with insuffi cient water particularly for perennials, as investments in the latter would have decreased. Even so, in this counter-factual scenario, lobbyists would have fought to have environmental water returned to farmers during the drought. It appears inevitable that balancing environmental and economic needs will lead to further discussions concerning temporary water trades between envi-ronmental managers and irrigators.

References

ABARE–BRS (2010) Environmentally sustainable diversion limits in the Murray–Darling Basin: socioeconomic analysis. http://www.daff.gov.au/__data/assets/pdf_fi le/0011/1812971/regional-impact-mdba-2010.pdf . Accessed 2 Mar 2011

ABARES (Australian Bureau of Agricultural and Resource Economics and Sciences) (2010) Australian Commodity Statistics 2010. ABARE, Canberra

Abayasiri-Silva K, Horridge M (1998) Economies of scale and imperfect competition in an applied general equilibrium model of Australia. In: Arrow K, Ng Y-K, Yang X (eds) Increasing returns and economic analysis. Macmillan Press Ltd., Basingstoke, pp 307–334

ABS (Australian Bureau of Statistics) (2008) Water and the Murray Darling Basin: a statistical profi le, 2000–01 to 2005–06. Catalogue 4610.0.55.007

ABS (Australian Bureau of Statistics) (2009) Value of agricultural commodities produced, Australia, 2007–08. Catalogue 7503.0. Australian Bureau of Statistics, Canberra

ABS (Australian Bureau of Statistics) (2011) Water use on Australian farms. http://www.abs.gov.au/ausstats/[email protected]/mf/4618.0/ . Accessed 26 Aug 2011

Akerman P (2010) Farmers pour scorn on Murray-Darling water blueprint. The Australian, October 13. http://www.theaustralian.com.au/national-affairs/climate/farmers-pour-scorn-on-murray-darling-water-blueprint/story-e6frg6xf-1225937895134 . Accessed 26 Aug 2011

Anderson K, Nelgen S, Valenzuela E, Wittwer G (2010) Economic contributions and characteris-tics of grapes and wine in Australia’s wine regions. Wine Economics Research Centre Working Paper No. 0110, University of Adelaide

Armington P (1969) The geographic pattern of trade and the effects of price changes. IMF Staff Pap XVI:176–199

Armington P (1970) Adjustment of trade balances: some experiments with a model of trade among many countries. IMF Staff Pap XVII:488–523

Bureau of Meteorology (2003) Drought statement. http://www.bom.gov.au/climate/drought/archive/20021101.shtml . Accessed 2 Mar 2011

Bureau of Meteorology (2008) Special climate statement 15: an exceptional and prolonged heat-wave in southern Australia. http://www.bom.gov.au/climate/current/statements/scs15b.pdf . Accessed 2 Mar 2011

Bureau of Meteorology (2009) Special climate statement 17: the exceptional January–February 2009 heatwave in south-eastern Australia. http://www.bom.gov.au/climate/current/statements/scs17d.pdf . Accessed 2 Mar 2011

http://www.daff.gov.au/__data/assets/pdf_file/0011/1812971/regional-impact-mdba-2010.pdf

http://www.daff.gov.au/__data/assets/pdf_file/0011/1812971/regional-impact-mdba-2010.pdf

http://www.abs.gov.au/ausstats/[email protected]/mf/4618.0/

http://www.abs.gov.au/ausstats/[email protected]/mf/4618.0/

http://www.theaustralian.com.au/national-affairs/climate/farmers-pour-scorn-on-murray-darling-water-blueprint/story-e6frg6xf-1225937895134

http://www.theaustralian.com.au/national-affairs/climate/farmers-pour-scorn-on-murray-darling-water-blueprint/story-e6frg6xf-1225937895134

http://www.bom.gov.au/climate/drought/archive/20021101.shtml

http://www.bom.gov.au/climate/drought/archive/20021101.shtml

http://www.bom.gov.au/climate/current/statements/scs15b.pdf

http://www.bom.gov.au/climate/current/statements/scs17d.pdf

http://www.bom.gov.au/climate/current/statements/scs17d.pdf


Dixon P, Rimmer M (2002) Dynamic general equilibrium modelling for forecasting and policy: a practical guide and documentation of MONASH. North-Holland, Amsterdam

Dixon P, Rimmer M (2010) You can’t have a CGE recession without excess capacity. Econ Model 28:602–613

Dixon P, Rimmer M, Wittwer G (2010) Modelling the Australian government’s buyback scheme with a dynamic multi-regional CGE model. http://www.monash.edu.au/policy/ftp/workpapr/g-186.pdf . Accessed 15 Feb 2011

Dixon P, Parmenter B, Sutton J, Vincent D (1982) ORANI: a multisectoral model of the Australian Economy. Contributions to economic analysis 142, North-Holland, Amsterdam

Harris R (1984) AGE analysis of small open economies with scale economies and imperfect com-petition. Am Econ Rev 74:1016–1032

Harrison J, Horridge M, Pearson K, Wittwer G (2004) A practical method for explicitly modelling quotas and other complementarities. Comput Econ 23:325–341

Hertel T, Swaminathan P (1996) Introducing monopolistic competition into the GTAP Model. GTAP technical paper no. 06


Kharitonov V, Walmsley T (2004) Impact of Russia’s WTO accession on the structure of the Russian economy. Presented at the 7th annual conference on global economic analysis, Washington, DC

Kuik O, Gerlagh R (2005) PSR emissions trading: an alternative for conventional emissions trad-ing that would prevent carbon leakage and protect exposed sectors of industry? Presented at the 8th Annual Conference on Global Economic Analysis, Lübeck, Germany

National Water Commission (2009 and earlier issues) Australian water markets report 2008–2009. National Water Commission, Canberra

Pauw K, Thurlow J, Bachu M, van Seventer DE (2011) The economic costs of extreme weather events: a hydro-meteorological CGE analysis for Malawi. Environ Dev Econ 16:177–198

Rizza A (2010) The potential effects of changes to water allocation policy on fi nancing the agricultural sector and businesses in the Murray Darling Basin. Report prepared for the Murray-Darling Basin Authority, October

Sherony K, Knowles G, Boyd R (1991) The economic impact of crop losses: a computable general equilibrium approach. West J Agric Econ 16:144–155

Sunrice (2010) SunRice announces Deniliquin Mill will reopen in 2011. http://www.sunrice.com.au/uploads/MR_-_Deniliquin_Mill_to_Re-open.pdf . Accessed 20 Nov 2010

Wentworth Group of Concerned Scientists (2002) Blueprint for a living continent. http://www.wwf.org.au/publications/blueprint_for_a_living_contintent/ . Accessed 1 Mar 2011

Wittwer G (2003) An outline of TERM and modifi cations to include water usage in the Murray-Darling Basin. Downloadable from www.monash.edu.au/policy/archivep.htm TPGW0050


Wittwer G, Griffi th M (2011) Modelling drought and recovery in the southern Murray-Darling basin. Aust J Agric Resour Econ 49:342–359

Young M, Proctor W, Qureshi E, Wittwer G (2006) Without water: the economics of supplying water to 5 million more Australians. http://www.myoung.net.au/water/publictions/WithoutWater.pdf . Accessed 1 Mar 2011





http://www.wwf.org.au/publications/blueprint_for_a_living_contintent/





http://www.myoung.net.au/water/publictions/WithoutWater.pdf

http://www.myoung.net.au/water/publictions/WithoutWater.pdf


Abstract South-east Queensland suffered a record drought for several years after 2005 while accounting for almost one quarter of Australia’s entire population growth. This resulted in an urban water supply crisis. The state government’s response was to plan new dams, construct pipelines to create a water grid, and build a massive recycling plant and a desalination plant. Policy makers are in an invidious position, in so far as they will be accused of not preparing for the future should they underinvest in water infrastructure. Were a drought to continue indefi nitely, it would be possible to justify much of the new infrastructure. Yet with a return of average rains, analysts may regard some of the new infrastructure as an excessively expensive means of maintaining a secure water supply.

Keywords Desalination • Flood mitigation • Price rationing • Supply augmentation • Urban water crisis

8.1 Introduction

Although Australia is a land-abundant nation, its population is concentrated in a handful of urban coastal regions. More than three fi fths of the population resides in fi ve mainland capital cities: Sydney, Melbourne, Brisbane, Perth and Adelaide. The most rapidly growing population is in the south-east corner of Queensland, covering Brisbane, Gold Coast, Sunshine Coast and West Moreton statistical divisions. This corner of Queensland has accounted for over 21% of national population growth since 2001 (Table 8.1 and Fig. 8.1 ).

G. Wittwer (*) Centre of Policy Studies , Monash University , Wellington Rd, Clayton campus , Melbourne , VIC 3800 , Australia e-mail: [email protected]

Chapter 8 Urban Water Supply: A Case Study of South-East Queensland

Glyn Wittwer

144 G. Wittwer

Rapid population growth tends to strain urban infrastructure, whether it concerns private and public transport services, hospitals, schools, community and aged care, energy utilities, shopping centres or urban water supply. In this chapter, we use the provision of urban water supply in south-east Queensland as a case study.

Each of the mainland capital cities and many regional cities experienced reduc-tions in water availability in the fi rst decade of the 2000s. Melbourne, with the most rapidly growing population outside south-east Queensland and Perth, received below average rainfall in every year between 1997 and 2009. This kept Melbourne’s reservoirs at historically low levels from 2006 on (Melbourne Water 2011 ) . The then Victorian state government responded to ongoing water shortages by proceed-ing with a $5.7-billion desalination plant (net present value, 2010 dollars). This plant, located west of Wonthaggi in South Gippsland, will produce up to 150 GL of potable water per annum (Victorian Government 2009 ) . The Victorian government has also constructed a pipeline linking Eildon reservoir, part of the Murray-Darling basin catchment, with Melbourne’s water supply, which is on the other side of the Great Dividing Range. The pipeline is part of a larger works program to link regional centres in Victoria to a larger water supply. Given the population growth fi gures shown in Table 8.1 and prevalence of a number of years of drought in both south-east Queensland and Victoria in the decade following the turn of the millennium, it is not surprising that these two states have allocated substantial amounts of public funding to water supply projects.

In Perth, average rainfall since the mid-1970s has declined by around 30%, with a much larger percentage decrease in catchment volumes for the city’s water supply. A bore sunk at Pinjar in 1996 yields almost 18 GL per annum, while another sunk at Neerabup in 1999 yields 26 GL per annum. Additional groundwater sources were exploited in succeeding years amounting to 86 GL per annum. Samson Brook dam opened in 2003 with an expected yield of 13 GL per annum, and Wokalup Creek dam in the same year, yielding 10 GL per annum (Water Corporation 2011a ) .

Table 8.1 Population growth: south-east Queensland, other mainland capitals and rest of Australia

2001 (1,000 s)

2008 (1,000 s) Growth (%)

Land area (1,000 s km 2 )

Share of national population growth

Brisbane 1663.1 2043.2 22.9 5.9 13.0 Gold Coast 387.1 527.8 36.3 1.3 4.8 Sunshine Coast 247.2 330.9 33.9 3.1 2.9 West Moreton 77.2 97.4 26.2 11.9 0.7 Total SEQ 2374.6 2999.3 26.3 22.3 21.4 Sydney 4128.3 4575.5 10.8 12.1 15.3 Melbourne 3471.6 4077.0 17.4 7.7 20.8 Adelaide/Outer

Adelaide 1222.0 1342.7 9.9 13.5 4.1

Perth 1393.0 1696.1 21.8 5.4 10.4 Rest of Australia 6823.7 7638.2 11.9 7619.1 27.9 Australia 19413.2 22328.8 15.0 7702.4 100.0

Source : ABS ( 2011 )

1458 Urban Water Supply: A Case Study of South-East Queensland

Fig. 8.1 Map of south-east Queensland. Key: 1 Brisbane; 2 Gold Coast; 3 Sunshine Coast; 4 West Moreton; 5 Rest of Qld; 6 Rest of Australia

A $387-million desalination plant was opened at Kwinana in November 2006. This plant has a capacity of 45 GL per annum (Water Corporation 2011b ) .

Although Adelaide has the slowest population growth of the mainland capitals, it too suffered an urban water crisis following a 3-year drought in the heart of the Murray-Darling basin catchment. Adelaide is alone among Australia’s capital cities in relying on the basin to meet a substantial share of its water requirements. In the past, South Australia (of which Adelaide and Outer Adelaide account for over 80% of the population) drew around 30% of its water needs from the Murray River in wet

146 G. Wittwer

years and up to 85% during drought (Atlas South Australia 2004 ) . With such a reliance on the Murray, it becomes apparent why Adelaide and the rest of South Australia had major water shortages in the wake of a precipitous fall in catchment volumes in the Murray-Darling basin between 2006 and 2009. This has resulted in major water supply augmentation projects becoming policy priorities, led by a $1.8-billion desalination plant at Port Stanvac. This will produce up to 100 GL of water per annum (Swan 2009 ) . The plant started producing water at the end of July 2011 (McGregor 2011 ) .

Sydney’s main water augmentation project is the desalination plant at Kurnel. Costing $1.9 billion, it has a capacity of 90 GL per annum. The then state premier’s media release of 28 January 2010 announcing the opening of the plant noted that Sydney’s population consumes the same volume of water as in the early 1970s, when the population was 1.5 million smaller (NSW Government 2010 ) . This raises questions concerning the volume of necessary new water supplies even in rapidly growing cities. That is, given water effi ciency gains over time and changes in house-hold water consumption patterns, water supply volumes need not necessarily grow as rapidly as the population.

8.1.1 South-East Queensland Water Management

We use CGE analysis to evaluate a single project in south-east Queensland. This project, the Traveston dam, has now entered the realm of the hypothetical, having been rejected by the Commonwealth government in November 2009. Nevertheless, the process of modeling raises some important recurring issues concerning infra-structure development. In particular, at what point do the marginal costs of a project exceed the marginal benefi ts? From an engineering perspective, for an urban region to get through a drought without water restrictions may be a measure of success. An economist may think the social costs of this high degree of risk aversion excessive and that various forms of water rationing in response to scarcity are more likely to be socially optimal.

Young et al. ( 2006 ) used an early version of TERM that included water accounts for both urban users and agricultural users to model the impact of various supply augmentation and rural-urban water trading scenarios. The main theme of the study was to project the economic impacts of supplying water to a national population growing by fi ve million, with the assumption that most growth would be concen-trated on the eastern seaboard. The headline fi nding of the study was that in a do-nothing scenario, in which urban water supplies were not augmented, urban water prices in some urban regions would increase in real terms by more than fi vefold. The do-nothing scenario did not refl ect policy at the time, as various water supply augmentation programs were either under construction or in a planning stage. The assumption underlying the modeling was that water prices would bear most of the adjustment to rising demand in the face of fi xed supply. In practice, we might expect substitution away from water in industrial and household usage, while users may


seek to augment supplies. For example, household and industrial users could increase their use of rainwater tanks.

The Young et al. ( 2006 ) study showed that a hypothetical do-nothing approach would have been costly. This was despite an underestimate of the actual rate of population growth in the study. Moreover, drought conditions in the next few years in south-eastern Australia ensured that the assumptions concerning water availabil-ity within the Young et al. ( 2006 ) study looked optimistic relative to actual avail-ability in the short term. Indeed, during the period from 2006 to 2009, the imposition on households, gardeners, garden nurseries and other industries of non-price water rationing may have equated to costs approaching $10 per kilolitre. Now, the ques-tion concerning future water needs is switching from an evaluation of the costs of doing nothing to an evaluation of the costs of overinvestment. To put the dangers of overinvestment into context, the annualised cost of the Wonthaggi desalination plant could add more than $1 per kilolitre to the fi xed costs of water supply to Melbourne. 1 Desalination plants are a costly form of water supply security to build and, in some rainfall scenarios, could appear to be white elephants for decades at a time. Nevertheless, when weighed against the non-price rationing costs during periods of extreme water scarcity, desalination may be a politically justifi able form of water security. But the high costs of desalination also need to be weighed against alterna-tive measures including storm water catchment and recycling.

An economist’s perspective on desalination is that a relatively small plant may be a justifi able form of water supply insurance. The larger the plant, the more likely it is that the marginal costs will outweigh the marginal risk-adjusted benefi ts. In par-ticular, the plants in Victoria and South Australia are being constructed at a larger scale than initially planned. That a major prolonged drought ended before the com-pletion of construction has reduced the marginal benefi ts of this form of supply augmentation in these two states.

8.1.2 Queensland Strategies

The Queensland government ( 2006 ) has considered a number of options to increase urban water supply in the south-east corner, including dam construction. One of the largest contributions to increased water supply was to have come from the Traveston dam, which had been planned to yield at least 70 GL per annum, based on an initial capacity of 180 GL. However, the Commonwealth blocked the dam in November 2009, forcing the Queensland government to shelve the project.

1 The annual fi xed cost calculated as 5% of the net present value cost is $285 million (2010 dollars), but the annual payment by the state government to AquaSure will total $654 million in 2012–13, even if no water is delivered (The Age 2011 ) . Melbourne’s annual water usage is around 300–400 GL (inferred from ABS catalogue no. 4610.0). The Baillieu state government elected in 2010 was unable to back out of the desalination contract (Jenkins 2011 ) . Given South Australia’s small population, its desalination plant is also a costly form of water security.

148 G. Wittwer

Other planned supply augmentation measures as shown in Table 8.2 include construction of Wyaralong dam, now approved by the Commonwealth government; an extension to Hinze dam; a desalination plant; water recycling; and pipeline construction. 2 The Queensland government is also exploring the possibility of supplementing recycled water with storm water catchment.

Each of the options proposed to deal with supply augmentation has its own envi-ronmental and economic issues. In the case of desalination, although the costs of the reverse osmosis process have come down markedly, the process remains energy intensive. In addition, fi nding a suitable site on the intensively developed coastline of south-east Queensland was not straightforward. There are concerns over hydro-dynamics and brine disposal on the coastline. Social issues include noise and visual amenity. The costs of the desalination plant on the Gold Coast (and at Port Stanvac in South Australia) are several-fold higher than those at Kwinana (Perth) for a simi-lar capacity. According to John Kobelke, the then Water Resources minister in Western Australia, any additional desalination plant in the state would have cost much more than Kwinana due to additional environmental compliance costs (Spagnolo 2006 ) .

Queensland experienced widespread high rainfall and fl ooding in the second half of 2010. The Queensland government announced in December 2010 that they would mothball the Gold Coast desalination plant, with anticipated savings of $10 million per year from moving to standby mode (Brisbane Times 2010 ). Ironically, with the major fl oods of January 2011, the desalination plant resumed production to redress a fl ood-induced shortage of potable water (Lappeman 2011 ).

A signifi cant proportion of the costs in upgrading south-east Queensland’s water supplies entailed pipeline construction so as to connect a formerly fragmented series

Table 8.2 Major water supply projects in south-east Queensland

Project Cost ($m) Average annual yield (GL)

Western Corridor Water Recycling Project 2,500 80 Gold Coast desalination 1,200 45 Wyaralong dam 350 21 Hinze dam extension 395 6 Bromelton off-stream storage 40 5 Traveston (cancelled) 1,600 70 Pipelines: Southern 900 Toowoomba 187 Northern Eastern

Sources: http://www.ancr.com.au/Hinze_Dam.pdf and http://www.dip.qld.gov.au/seqwatergrid

2 The Queensland government report ( 2006 , p. 18) notes that a combination of these measures may reduce additional supply needs by 40–80%.

http://www.ancr.com.au/Hinze_Dam.pdf

http://www.dip.qld.gov.au/seqwatergrid


of dams. The appeal of pipelines to connect the different supplies arises from the wide variability in rainfall patterns across the region. At a given time, one reservoir may be at a low level while another is near capacity. Extending water supply to a single grid may reduce critical water scarcity in parts of south-east Queensland while at the same time adding recycled water to the grid.

8.1.3 Demand-Side Management, Hypothetical and Actual

Current demand management strategies being used by water suppliers in Australia refl ect a chasm between urban water managers (refl ecting government policy) and economists. Without exception, water suppliers have introduced quantitative restric-tions rather than increases in marginal prices in response to worsening scarcity. In the case of south-east Queensland, the intent of stage 2 restrictions (in place during 2006) was to reduce water consumption by 15%. Given that water has a low cost share in non-irrigation industries and a low budget share in households, and that it is not readily substitutable, it has a relatively low own-price elasticity of demand in urban usage. If the elasticity is around −0.2, the smallest elasticity cited by Edwards ( 2006 ) , then raising the marginal price of water by 125% would achieve the 15% reduction in usage. However, water utilities impose a substantial fi xed service charge on users, with only part of each water bill being for water usage. Existing estimates of the price responsiveness of users may be masked by the fi xed charge.

Silby ( 2006 ) explains that effi cient pricing by a natural monopoly such as a water utility in usual circumstances entails imposing both a fi xed and a marginal charge on water users to cover average costs. But as water scarcity worsens, effi cient pricing would entail a reduction in the fi xed charge and an increase in the marginal charge. This would increase the price responsiveness of users as scarcity worsens. It would also allow users to choose whether they are willing to bear the rising costs of maintaining a garden or sports fi eld or reduce expenditure on water through reduced usage.

The average cost of water supply rises as water availability decreases due to drought. Raising the price of water to users in response to worsening scarcity would increase the cost-effectiveness of water-saving initiatives. As mains water marginal prices increase, the opportunities for cost-effective supply augmentation increase. Again, as the price of mains water rises, more alternative supply sources and water-saving measures become competitive. On the other hand, some consumers have shown a willingness to pay an extremely high price for water, notably by fi tting of rainwater tanks to established homes or by sinking a bore. Bores have potential negative externalities, although many suburban houses are likely to deplete an aqui-fer at a slower rate than a few irrigation farms.

With price rationing instead of water restrictions, some consumers, at least in the absence of subsidies on water tanks, may have chosen to buy water at a high price instead of paying a large fi xed cost for tank installation or a bore. Others would have found that with price rationing, their installation yielded a higher return. Price rationing provides a more direct mechanism for allocation than alternatives, including watering restrictions and subsidies for rainwater tanks.

150 G. Wittwer

Economists have proposed pricing reform in urban water management for a number of decades (e.g., Moncur and Pollock 1988 ) . A practical issue concerns when and how often water managers should alter the marginal price of water. Dam levels tend to have both a seasonal cycle and – in the case of rising scarcity due to drought and population growth – a longer term trend. Water managers should use target storage levels as an early indicator to raise marginal prices and lower fi xed charges. 3 Arguably, current water restrictions have been imposed on users only when scarcity has been evident for a signifi cant time beforehand. Prices are easier to change than levels of water restrictions which are often accompanied by ever-more complicated rules about when and who can use water outdoors. Even though some of the urban water crises could have been managed better with the introduc-tion of changing marginal prices a decade or so ago, such price signals could still play a vital role in future water management.

The only pricing policy that appears to prevail among state governments at pres-ent is to raise the fi xed and marginal costs of water to fund water infrastructure projects. It appears that from a government perspective, centralised solutions are superior to measures that might encourage individual fi rms or households to catch rainwater or recycle water.

As is so elsewhere in Australia, restrictions on the time and type of outdoor watering have been imposed on south-east Queenslanders to reduce their water con-sumption. The Queensland government ( 2006 ) report noted that ‘a consumption rate of 230 L per day (per capita) is considered to be a very diffi cult target to achieve’ by 2020. Yet, a year later, a Queensland Water Commission website recorded that average daily usage in the region was 130 L, even less than the target of level fi ve restrictions of 140 L per day. 4 While such a low usage may not be sustainable in the long term, given that it includes onerous restrictions on outdoor water usage (no car washing and restricted watering in gardens), the response to these restrictions by the community may include initiatives that are permanent. Such water savings already achieved may be relevant as the Queensland government considers further supply augmentation measures. The government’s 2006 report estimated that additional supply requirements by 2050 would be 300 GL per annum, based on a per capita consumption rate of 230 L per day.

The Queensland government has devised building codes that promote greater water effi ciency in new buildings. Under legislation introduced in 2006, the govern-ment expects new houses to use 36% less water than houses without the water effi -ciency measures (Queensland Government 2006 ) . Since south-east Queensland is growing rapidly, new housing is appearing faster than elsewhere in Australia, so that the effect of such legislation will be greater than elsewhere. The government is also subsidising water-effi cient devices. The main avenue that has not yet been explored is that of raising water prices to reduce the opportunity costs faced by households and industries in improving water effi ciency.

3 Melbourne’s reservoirs take time to fi ll. Those near Brisbane do not. Consequently, in the case of Brisbane, variable marginal pricing would be harder to implement. 4 Downloaded from http://www.qwc.qld.gov.au/Level+5+restrictions . No longer available.

http://www.qwc.qld.gov.au/Level+5+restrictions


8.2 Modeling a Water Infrastructure Project

This study uses TERM-DYN to analyse urban water in south-east Queensland. In TERM-DYN, interregional wage relativities may change relative to forecast: if regional labour market conditions improve or deteriorate relative to forecast, adjust-ment occurs in the short term mainly via changes in employment. Regional wages adjust sluggishly, with gradual adjustment in regional labour market supply (i.e., through migration between regions). Real wages will fall or rise to close the gap between employment and slowly adjusting labour supply. Within this theory, adjust-ment in the longer term occurs via a combination of altered regional labour supply and real wages that deviate relative to those in other regions (see Sect. 3.2.3 ). 5

TERM-DYN contains a standard theory of primary factor allocation as in the ORANI model (Dixon et al. 1982 ), without the theoretical modifi cations of TERM-H2O outlined in Chap. 5 . Another difference between TERM-DYN and TERM-H2O is that there are no satellite water accounts in TERM-DYN. The input-output data-base of a CGE model is represented in values. Without satellite water accounts, we must assume that the volume of water by user relates to the base value of the water sector in the input-output table. That is, each user pays the same price for water. 6

The reader may ponder why we have chosen to use TERM-DYN rather than TERM-H2O. First, this chapter concerns urban water management. Although TERM-H2O covers regions and sectors outside the Murray-Darling basin, its focus is on competition for farm factors between irrigated and dryland sectors. This feature has little relevance to urban water. Perhaps we could have added satellite water accounts to TERM-DYN, but without widely varying water prices between users, this provides little advantage. In the case of south-east Queensland, different urban users (notably households) who pay a similar price dominate water consumption, so assuming that each user pays the same price for water is defensible. If low-priced users dominated regional water consumption, we would need to link satellite accounts of water quantities to the model, as in Chaps. 3 and 4 . Similarly, to capture differ-ences between the willingness to pay of different urban users, we would need to combine water accounts with a fi ner level of sectoral disaggregation than is available in the published input-output table. That is, we might represent nurseries separate from other wholesale and retail trade activities, and sporting clubs who prefer green ovals separate from other recreational activities. It turns out that modeling results hinge on the marginal benefi ts of additional water net of the costs of supply augmen-tation. The unique attributes of TERM-H2O would add little to the analysis.

5 This contrasts with TERM-H2O, in which all regional labour market adjustment is via migration rather than a combination of interregional wage differentials and migration. This refl ects the policy focus of TERM-H2O, which is on farm factor adjustment. 6 The average price of water in the baseline in year 2006 is $1 per kilolitre, including both fi xed and marginal costs, within south-east Queensland.

152 G. Wittwer

We assume that the output of the water and drains sector in the model is equivalent to the volume of urban water supply. Within a dynamic CGE baseline, water can be treated as an exogenous resource, the scarcity of which worsens with economic growth. Alternatively, the forecast baseline could include a return to a higher rain-fall that to some extent offsets rising demands. Moreover, the baseline typically includes technological change, that is, changes in inputs per unit of output over time. Each industry within the model has a given water input per unit of output, and the representative household’s consumption bundle includes water. On the demand side, households respond to changes in the price of water relative to other commodi-ties and also respond to changes in disposable income. The water savings we assume over time by each user in the model have some impact on changing water scarcity. We have assumed in the baseline that neither industry nor household savings are as rapid as output and population growth so that the relative scarcity of water worsens over time.

A number of issues arise from defi ning the forecast baseline and estimating the welfare benefi ts of the dam relative to the baseline. Such benefi ts depend critically on underlying assumptions concerning future water availability due to rainfall pat-terns, general economic growth and future water savings by industrial and house-hold users. In turn, the rate at which water savings occur may depend on demand management strategies. Moreover, a particular project that increases the water supply within a region will reduce the marginal economic benefi ts of further water supply projects.

In the policy scenario, the change in consumption of additional water by all users in the model will depend on activity growth within the baseline and may not be proportional to current usage. For example, households might dominate additional usage of augmented water supply. That is, we expect forecast household growth to be higher than forecast manufacturing industry growth in a rapidly growing region. This change in the structure of urban water usage may be diffi cult to capture in models that are not dynamic.

8.2.1 Accounting for Project Costs and Benefi ts

The welfare gains from a project decrease as the gap lengthens between costs being incurred and benefi ts being realised. In this hypothetical study, very large invest-ments from 2009 to 2011 provide signifi cant stimuli to south-east Queensland, with positive impacts on aggregate employment and other macro variables including household consumption in the region. However, projects of this magnitude need to be paid for. Within the model, net foreign (both international and interregional) liabilities in south-east Queensland rise as a consequence of the large investments from 2009 to 2011, so that aggregate consumption in the region falls as a share of regional income (or GRP = gross regional product) thereafter, refl ecting an after-math of additional interest payments on additional borrowings incurred during the investment phase. If the direct and indirect benefi ts of the operational phase of the


dam do not offset construction costs, the project will result in a net welfare loss. This is likely to be the case under certain future rainfall scenarios.

8.2.2 Costs and Benefi ts Not Considered in Dynamic TERM Modeling

One cost we have not attempted to model in this project is that of quantitative water restrictions on each user relative to price rationing. In the context of irrigation users, CGE modeling that includes satellite water accounts has demonstrated that trading in water as scarcity worsens provides welfare benefi ts relative to no trading. In the context of urban users, relatively water-intensive activities (nurseries, home gardens and sports fi elds) have been hindered by water restrictions. Under price rationing, we would expect some diversion of water usage towards such usage based on will-ingness to pay. To capture these effects in a CGE model would have required a fi ner level of sectoral disaggregation than is currently available for these predominantly urban users.

In the case of Traveston dam, one possible benefi t is in the form of fl ood mitiga-tion in river valleys (see Sect. 8.3.1 ). Benefi ts could be estimated from the frequency and severity of fl ood events and data on insurance claims. The frequency of expected fl ood events raises the costs of dam construction in south-east Queensland relative to other parts of Australia. Similarly, the expected benefi ts from fl ood mitigation are higher. However, the Mary River valley is sparsely populated when compared with the Brisbane River valley, implying that the fl ood mitigation benefi ts of Traveston would have been comparatively small had the project gone ahead. Another benefi t that arises from construction of a dam concerns water supply risk. Increased dam capacity within a region increases the ability of water managers to supply water during a prolonged period of rainfall defi ciency. 7 Such risk benefi ts are not included in the analysis of this chapter.

Future water storage requirements in south-east Queensland will depend on rain-fall patterns and regional economic growth. If rainfall returns to levels observed prior to 1996, annual yields will increase and the need for prolonged storage capac-ity will diminish. The extent to which water-saving practices remain intact follow-ing the easing of restrictions and new water-saving technologies will also affect future water needs. It is possible that future water needs for south-east Queensland will be revised downwards from 300 GL per annum by 2050. If so, the marginal economic benefi ts of any particular water infrastructure project will decline. Unless a very high risk premium is attached to increased water supply security, a number of the water projects announced by the Queensland government in response to the south-east corner’s water crisis may not be economically justifi able.

7 Pipelines to establish a water supply grid may not increase the overall supply of water but provide benefi ts by reducing supply risk.

154 G. Wittwer

8.2.3 Modeling the Economic Impacts of the Traveston Dam Project

We report the results of a TERM-DYN aggregation that includes three regions: (1) south-east Queensland (Brisbane, Sunshine Coast, Gold Coast and West Moreton statistical divisions combined), (2) rest of Queensland and (3) rest of Australia. In addition, the sectors are relatively aggregated, refl ecting in part the absence of substantial irrigation agriculture in the region. An exception is West Moreton, in which water supplies are not well connected to those of the urban areas. West Moreton’s irrigation sectors account for a shrinking proportion of south-east Queensland’s water usage. 8

Construction of Traveston dam on Mary River, in this now hypothetical scenario, commences in 2009 and proceeds over 3 years. The estimated cost of the project is $1.6 billion. The project will raise the present water yields of existing catchments in south-east Queensland by at least 70 GL. On the supply side of the forecast base-line, we assume that there is no action to augment south-east Queensland’s water supply. On the demand side, we assume that year-by-year reductions in water requirements by all users are much slower than forecast economic growth. Therefore, the price of the fi xed water resource rises relative to other commodities over time.

The construction phase will strengthen the labour market in south-east Queensland, which is already rapidly growing in the baseline forecast. The impact of each dollar of investment in the water sector on the labour market is not as large as each dollar spent on other forms of investment. Housing construction, for exam-ple, is more labour intensive and less material intensive than investment in the water sector within the TERM database. Despite this, employment still rises in south-east Queensland by 0.14% or 1,700 jobs relative to forecast in 2009 (Fig. 8.2 ). In the theory of dynamic TERM, real wages adjust sluggishly to changing labour market conditions. Therefore, average real wages in the region rise by only 0.06% relative to forecast in 2009. In succeeding years, real wages continue to rise. Over the remaining 2 years of the construction phase, rising wages choke off employment slightly. By 2011, employment is 0.09% or 1,100 jobs above control while real wages are 0.14% above forecast.

In years 2010 and 2011, construction of Traveston dam adds to the capital stock of south-east Queensland relative to forecast. But since the dam is not yet opera-tional, real GRP does not rise in 2010 and 2011 as capital rises. Instead, real GRP falls as regional employment moves back towards control in these years due to ris-ing regional real wages. Once the dam is operational in 2012 (i.e., the new capital is no longer idle), real GRP starts rising relative to forecast although employment falls back towards forecast (Fig. 8.3 ).

8 The 19 sectors in the aggregation used in this study are agriculture, forestry and fi shing; mining; food, beverages and tobacco; other manufactures; metals; utilities (excluding water); water supply; construction; trade; hotels and cafes; transport; communication; property and business services; ownership of dwellings; government and defence; education; health and community services; cul-tural and recreational services; and other services.


0

0.05

0.1

0.15

0.2

0.25

0.3

2008 2011 2014 2017 2020 2023 2026 2029

Employment

Labour supply

Real wage

% deviation from forecast

Fig. 8.2 Impact of the Traveston dam project on south-east Queensland’s labour market

-0.05

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

2008 2011 2014 2017 2020 2023 2026 2029

Employment

Real GDP

Capital stock


Fig. 8.3 Impact of project on south-east Queensland’s GRP, capital and labour

In this scenario, we model the output of the water supply industry in south-east Queensland as exogenous. In years leading up to and including 2011, water supply depends on dam capacity as existed prior to the turn of millennium. As demands for a fi xed supply of water grow, the price of water rises. We assume that by 2008, drought constrains water usage. With prolonged drought and a rapidly growing population, water scarcity in south-east Queensland worsened. The rising price of water in TERM-DYN may not refl ect actual price hikes imposed by water authorities. Given the usual demand strategies of water managers, the modeled price hikes are a

156 G. Wittwer

proxy for the economic costs of water imposed on communities through various non-price rationing measures. When the dam becomes operational in 2012 in the hypothetical scenario, there is an outward shift in water supply that alleviates the previous price hikes.

In 2009, aggregate investment rises to 1.2% above forecast and aggregate con-sumption to 0.32% above forecast in south-east Queensland (Fig. 8.4 ). Real current government expenditure is unchanged by assumption. The share-weighted sum of the contribution of consumption and investment to the deviation in real GRP in 2009 is 0.65%. Since real GRP is only 0.13% higher than the baseline in 2009 (Fig. 8.3 ), this implies that, consistent with a real appreciation in the local economy, the net interregional and international trade defi cit of south-east Queensland has enlarged. Figure 8.5 expresses exports from south-east Queensland to the rest of the world and imports from the rest of the world as a share of real GRP. From 2009 to 2011, exports (through shrinkage) and imports (through an increase) relative to forecast make negative contributions to real GRP. When construction ceases and the dam becomes operational in 2012, south-east Queensland’s export and import con-tributions jump to near forecast. Thereafter, both export and import volumes are slightly above forecast, making almost offsetting contributions to real GRP. Both real investment and household consumption persist above forecast, in line with the increase in the region’s real GRP (Figs. 8.3 and 8.4 ).

The labour market response to the Traveston dam project occurs through a com-bination of higher real wages in south-east Queensland and higher employment in the region. The labour market adjustment at the national level in the long run is almost entirely through deviations in wages rather than employment. This implies that if south-east Queensland’s employment increases relative to forecast in the long term, it draws labour from other regions. When interregional migration is involved,

0

0.2

0.4

0.6

0.8

1

1.2

1.4

2008 2011 2014 2017 2020 2023 2026 2029

Aggregate investment

Aggregateconsumption


Fig. 8.4 Impact on south-east Queensland’s aggregate consumption and investment


it is most convenient to measure welfare at the national rather than regional level. We calculate welfare as the net present value of the deviation in national aggregate consumption from forecast, which amounts to $3.4 billion at a real discount rate of 5%. Next, we consider the conditionality of this welfare calculation.

The welfare impact depends on reducing the price of water through supply aug-mentation, net of the costs of augmentation. The change in relative scarcity of water in the baseline is critical in estimating the welfare outcome. If rainfall returns to historical average levels by the time Traveston dam is operational, higher water sup-ply in the baseline reduces the marginal benefi t of Traveston dam to near zero. In such a circumstance, the upwards price pressures alleviated by Traveston would be smaller than during a prolonged period of relative scarce water. That is, an increase in average annual water yields of 35% in the baseline from 2011 reduces the dis-counted net present value of welfare to around $300 million. 9 In such circumstances, the project might still be justifi ed through a valuation of water supply risk and fl ood mitigation benefi ts net of environmental costs. Similarly, slower economic growth in the forecast period would reduce the welfare benefi ts of the project.

Larger net annual water savings by users would reduce the welfare benefi t of the project. In practice, water savings, notably those arising from more stringent building regulations, are not costless. The reduced levels in per capita water usage in south-east Queensland mentioned earlier have come at considerable inconvenience to users and may be reversible to some extent. Investment costs associated with rea-lised water savings would need to be accounted for. Increased water savings and their associated costs could form the basis of an alternative policy scenario.

-0.4

-0.35

-0.3

-0.25

-0.2

-0.15

-0.1

-0.05

0

0.05

Inter-regional + int'l imports

Inter-regional + int'l exports


20292008 2011 2014 2017 2020 2023 2026

Fig. 8.5 Contribution of trade to overall deviation in south-east Queensland’s real GDP

9 A 35% increase in run-off would result from a smaller percentage increase in rainfall, as higher rainfall is usually accompanied by lower evaporation, indicating a non-linear relationship between rainfall and run-off.

158 G. Wittwer

8.3 Discussion

In our case study modeling an increase in south-east Queensland’s water supply, we have chosen a project which has been blocked by the Commonwealth. In addition, we have modeled the project in isolation by not considering other urban water projects already going ahead in the region. We have already discussed a number of aspects of the scenario that make the welfare calculation highly conditional, notably baseline water supplies and forecast water savings by urban users. Moreover, since the smaller Wyaralong dam was approved by the Commonwealth a year before Traveston was blocked, it would have been more appropriate to model Traveston with Wyaralong already in the baseline. That we have not included Wyaralong simply refl ects that at the time modeling of Traveston was undertaken initially, we did not know that Wyaralong would be approved and Traveston rejected. The magnitudes of modeled impacts may vary between the two options, but they face similar policy issues.

The most expensive single water supply project in south-east Queensland has been the Western Corridor Water Recycling Project, costing $2.5 billion. Yet, although the recycling plant is capable of delivering 80 GL per annum of potable water to south-east Queensland’s water supply, only about half the recycled water supplied by the plant is used. This use is confi ned to three power stations, namely Swanbank, Tarong and Tarong North. The plan is to supplement Wivenhoe dam with recycled water when the combined level of three dams (the largest of which is Wivenhoe) falls below 40% of capacity. The political resistance to the use of recy-cled water remains after rejection of such water (supplied by Wivenhoe dam) in a local referendum in Toowoomba in 2006.

Western Australia’s Water Corporation ( 2011a ) is aiming to increase community acceptance of recycled water for drinking through its website, providing links to com-panies recycling in other parts of the world. Companies referenced in the United States include two in each of California and Georgia, plus one each in Arizona, Colorado, Virginia and Texas. Other referenced water recycling companies are located in England, Namibia and Singapore. Virtually all of the referenced companies recycle water for drinking purposes. It appears that the Water Corporation is aiming to improve the sustainability of groundwater resources developed since the mid-1990s by upgrad-ing and extending water recycling plants to replenish reserves. The corporation’s information sheet on the Kwinana Water Reclamation Plant indicates an incremental expansion of recycling capacity, hinting that eventually, water from the plant will be used for drinking in addition to the present industrial purposes. Perhaps the Toowoomba referendum served a useful purpose in showing that water authorities must cultivate rather than assume community acceptance of water recycling.

From a modeling perspective, there are a number of lessons that emerge from the Queensland case study. First, the most appropriate way to model an infrastructure project is to consider its marginal costs and benefi ts in isolation. This does not mean we should ignore other projects but rather that we include those that are already approved in the baseline. In the case of south-east Queensland, we did not know in the early stages which projects would and which projects would not be approved.


Second, if a project is enhanced by other infrastructure, such as connecting pipelines, at least we should recognise the conditionality of the project on support-ing infrastructure. We did not model the cost of pipelines in the policy run, which would have connected Traveston to the entire south-east grid. We have treated south-east Queensland as a single region in TERM-DYN, when without the con-necting pipelines, Traveston may only have provided water to a small part of the overall region. Clearly, the pipelines will enhance the marginal benefi ts of any sin-gle water supply augmentation project. 10 But dams, desalination and recycling, to some extent, are competing water supply sources, while the pipelines are comple-mentary to ensure region-wide coverage. Indeed, the pipelines, by completing a grid over the entire region, should reduce the number of water supply augmentation proj-ects required to meet a given level of future water needs. Given the extremities of rainfall conditions faced in south-east Queensland, a valuation of changes in water supply risk and valuation of expected fl ood mitigation benefi ts would be useful.

There will always be debate over major infrastructure projects. Some analysts may regard a project as necessary while another regards it as wasteful or redundant. When the Queensland government was proceeding with a number of major water projects in the south-east corner, one joke doing the rounds was that the government was praying that it would not rain. This encapsulates the invidious position that gov-ernments are in when dealing with the unknown. The state government’s response to the driest decade recorded during white settlement in south-east Queensland was to secure additional water supplies on the assumption that continuing dry years were highly probable. Had the government not responded, there may have been damning criticism of their failure to cater for a growing population during drought. From an economic perspective, it appears reasonable to conclude that the Queensland govern-ment may have gone a little too far with its supply response, as is evident in the pres-ent underuse of potable recycled water. On the other hand, we may regard water supplies such as recycling and desalination as part of a water insurance policy.

A Queensland Water Commission ( 2010 ) report indicates how much the urban policy debate has moved in south-east Queensland in a few years. The commission’s strategy on water management urges residents to keep per capita water consumption at or below 200 L per day, a far cry from the government’s concern 4 years earlier that 230 L per capita daily was an ambitious target. The commission’s report antici-pates growing community acceptance of recycled water for drinking. Land is being set aside for treatment plants and pipelines to leave open the possibility of expanding recycling capacity. Security of water supply during future droughts will be enhanced by the use of recycled water in Wivenhoe dam and operation of desalination plants at full capacity. The commission’s 2010 report notes that additional desalination plants may not be required until after 2021. This differs from Premier Bligh’s initial response, after the Traveston dam was blocked by the Commonwealth in 2009, to proceed in the near future with additional desalination plants (Moore 2009 ) .

10 Queensland Water Commission ( 2010 ) estimates that the pipelines raise the LOS (level of ser-vice) system yield by 14%. With growing population and demands, the marginal benefi ts of the pipelines will grow over time.

160 G. Wittwer

Concerning drought, demand management still has a role to play. Yet during the worst of the drought, there were no marginal price adjustments to urban water. At considerable expense, ‘drought proofi ng’ through recycling and desalination is now within reach.

8.3.1 The Aftermath of the January 2011 Floods in Queensland

South-east Queensland has been subjected to the extremes of drought and fl oods in the space of 5 years, testing the polar opposites of water supply security and fl ood damage. In the response to the devastating fl oods that occurred in Queensland in January 2011, the leader of the Federal Opposition Tony Abbott announced the for-mation of an opposition task group to examine the role of new or bigger dams. Policy analysts need to weigh the economic and environmental costs and benefi ts of using dams for either fl ood mitigation or water supply against alternative strategies. If water supply augmentation is now suffi cient in south-east Queensland at least in the medium term, this confi nes the role of new or bigger dams to fl ood mitigation. The Australian Water Association chief executive Tom Mollenkopf noted that fl ood miti-gation and water storage are separate engineering tasks (Lester and Boland-Rudder 2011 ) . Indeed, Wivenhoe dam includes two compartments: a 1,150 GL compartment for drinking water storage overlaid by a 1,450 GL fl ood mitigation compartment.

There is debate as to whether the dam management released suffi cient fl ows from Wivenhoe prior to heavy rainfall upstream to minimise fl ooding downstream. A Queensland government publication indicated that all water in the fl ood storage compartment of Wivenhoe (1,450 GL) should be released within 7 days (SEQ Water Grid 2011 ) . All agree that without Wivenhoe, fl ood damage in Brisbane in January 2011 would have been much worse. The issue is whether releases from the usual water storage compartment of the dam prior to January 2011 might have reduced further the fl ooding that followed (Drugan 2011 ). 11 Yet in another year, such pre-emptive action (based on a Bureau of Meteorology forecast for a wet season) might jeopardise the security of south-east Queensland’s water supply. However, in the driest scenario subsequent to such action, water managers could add recycled water to Wivenhoe to maintain supply security (Drugan 2011 ) . In any case, low lying parts of Brisbane were subjected to tidal fl oods on an almost daily basis from several weeks before the major fl ood in mid-January 2011, indicating that pre-emptive releases may have caused at least some fl ood damage (ABC News 2011 ) . Given all considerations, by February 2011, there appeared to have been a change in dam management practices associated with a La Nina event. The Queensland govern-ment announced that 25% of the storage compartment of Wivenhoe would be

11 Following torrential rain and fl ooding in the Lockyer Valley, Wivenhoe’s level rose from 103% to 189% in 5 days on 12 January 2011: http://www.seqwater.com.au/public/catch-store-treat/dams/wivenhoe-dam (accessed 21 February 2011).

http://www.seqwater.com.au/public/catch-store-treat/dams/wivenhoe-dam

http://www.seqwater.com.au/public/catch-store-treat/dams/wivenhoe-dam


released over a weekend to reduce the probability of further fl ooding (Bentley 2011 ) . The subsequent Queensland Floods Commission of Inquiry ( 2011 ) recommended that in the event of future forecasts of wet seasons similar to 2010–11, the full supply level of Wivenhoe dam should be reduced to 75%.

In the case of the now hypothetical Traveston dam, even if the Commonwealth had approved the project, dam construction would not have been suffi ciently advanced to play any role in fl ood mitigation by January 2011. In any event, the number of businesses and properties that may have benefi ted from fl ood mitigation along the Mary River would have been small compared to the case of the Brisbane River.

The rainfall variation in south-east Queensland is extraordinary. For example, the now closed weather station of Crohamhurst in the Glass House Mountains north-west of Brisbane exceeded its average annual rainfall (1,795 mm) in the fi rst 4 days of February 1893 (Bureau of Meteorology 2011 ) . 12 Given the potential for extreme rainfall events, it may be foolish for policy makers to expect engineers to succeed in fully ‘fl ood proofi ng’ the region. An important lesson from the January 2011 fl ood is that building codes in the Brisbane River valley should refl ect the expectation of occasional major fl ood events. Existing codes prevented worse damage from being infl icted by the fl ood. Nevertheless, there must be a point at which the marginal costs of further fl ood-risk reductions exceed the marginal benefi ts. In the event of a major fl ood, preservation of human life through evacuations ought to take prece-dence over preservation of property.

References

ABC News (2011) Brisbane’s CBD spared as king tide peak passes. http://www.abc.net.au/news/stories/2011/01/21/3118209.htm . Accessed 2 Feb 2011

ABS (Australian Bureau of Statistics) (2011) Regional population growth, Australia. Catalogue no. 3218.0. Canberra

Atlas South Australia (2004) Water supply. http://www.atlas.sa.gov.au/go/resources/atlas-of-south-australia-1986/environment-resources/water-supply . Accessed 2 Feb 2011

Bentley A (2011) Wivenhoe release to fend off fl ooding. Brisbane Times, February 14. http://www.brisbanetimes.com.au/environment/weather/wet-season-retreats-ahead-of-wivenhoe-release-20110214-1asdn.html . Accessed 14 Feb 2011

Brisbane Times (2010) Tugun desalination plant to be mothballed, December 5. http://www.bris-banetimes.com.au/queensland/tugun-desalination-plant-to-be-mothballed-20101205-18l30.html . Accessed 2 Feb 2011

Bureau of Meteorology (2011) Climate date online. http://www.bom.gov.au/jsp/ncc/cdio/weather-Data/av?p_nccObsCode=136&p_display_type=dailyDataFile&p_startYear=1893&p_c=−327439264&p_stn_num=040062 . Accessed 2 Feb 2011

Dixon P, Parmenter B, Sutton J, Vincent D (1982) ORANI: A multisectoral model of the Australian economy. Contributions to Economic Analysis 142, Amsterdam, North-Holland

Drugan A (2011) Far too much water left in the dam. The Australian, January 19. http://www.theaustralian.com.au/news/opinion/far-too-much-water-left-in-the-dam/story-e6frg6zo-1225990589929. Accessed 20 Jan 2011

12 Brisbane’s major fl oods were in 1893, 1974 and 2011.



http://www.atlas.sa.gov.au/go/resources/atlas-of-south-australia-1986/environment-resources/water-supply

http://www.atlas.sa.gov.au/go/resources/atlas-of-south-australia-1986/environment-resources/water-supply

http://www.brisbanetimes.com.au/environment/weather/wet-season-retreats-ahead-of-wivenhoe-release-20110214-1asdn.html.



http://www.brisbanetimes.com.au/queensland/tugun-desalination-plant-to-be-mothballed-20101205-18l30.html



http://www.bom.gov.au/jsp/ncc/cdio/weatherData/av?p_nccObsCode=136&p_display_type=dailyDataFile&p_startYear=1893&p_c=<2212>327439264&p_stn_num=040062



http://www.theaustralian.com.au/news/opinion/far-too-much-water-left-in-the-dam/story-e6frg6zo-%0a1225990589929.



162 G. Wittwer

Edwards G (2006) Whose values count? Demand management for Melbourne’s water. The Econ Rec 82:S54–S63

Jenkins M (2011) Govt can’t change desal contract: Baillieu. http://news.smh.com.au/breaking-news-national/govt-cant-change-desal-contract-baillieu-20110228-1bbjk.html . Accessed 2 June 2011

Lappeman S (2011) Tugun desal plant ramps up production. http://www.goldcoast.com.au/article/2011/01/12/282651_gold-coast-fl ood-watch.html . Accessed 2 Feb 2011

Lester T, Boland-Rudder H (2011) Water experts slam Abbott’s dam plan. http://www.theage.com.au/environment/water-issues/water-experts-slam-abbotts-dam-plan-20110107-19iwj.html . Accessed 2 Feb 2011

McGregor K (2011) Water begins to fl ow from Adelaide’s new desalination plant. http://www.adelaidenow.com.au/water-begins-to-flow-from-adelaides-new-desalination-plant/story-e6frea6u-1226105463464 . Accessed 1 Aug 2011

Melbourne Water (2011) Water report. http://www.melbournewater.com.au/content/water_storages/water_report/water_report.asp . Accessed 2 Feb 2011

Moncur J, Pollock R (1988) Scarcity rents for water: a valuation and pricing model. Land Econ 65:62–72

Moore T (2009) Premier hits back at Traveston rejection. http://www.brisbanetimes.com.au/queensland/premier-hits-back-at-traveston-rejection-20091111-i9d3.html . Accessed 2 Feb 2011

NSW Government (2010) Sydney’s wind powered desalination plant now online. http://www.sydneywater.com.au/WhoWeAre/MediaCentre/documents/ministerial/Keneally_switch%20on_280110.pdf . Accessed 2 Feb 2011

Queensland Floods Commission of Inquiry (2011) Interim report, August. http://www.fl ood-commission.qld.gov.au/__data/assets/pdf_fi le/0006/8781/QFCI-Interim-Report-August-2011.pdf . Accessed 2 Aug 2011

Queensland Government (2006) Water for south-east Queensland: a long term solution. http://pandora.nla.gov.au/pan/61484/20060810/www.nrm.qld.gov.au/water/water_infrastructure/pdf/long_term_solution.pdf . Accessed 2 Feb 2011

Queensland Water Commission (2010) South-east Queensland water strategy. http://www.qwc.qld.gov.au/security/pdf/seqws-full.pdf . Accessed 2 Feb 2011

SEQ Water Grid (2011) Wivenhoe and Somerset Dams, providing water supply and fl ood control for south-east Queensland. http://www.cabinet.qld.gov.au/MMS/MediaAttachments/2010/pdf/32253%20SEQWG%20Wivenhoe%20Fact%20Sheet%20A4%202pp%20F.pdf . Accessed 19 Jan 2011

Silby H (2006) Effi cient urban water pricing. The Aust Econ Rev 39:227–237 Spagnolo J (2006) New desalination plant likely. http://www.perthnow.com.au/news/western-

australia/new-desalination-plant-likely/story-e6frg13u-1111112546154 . Accessed 2 Feb 2011 Swan K (2009) Acciona back to expand Adelaide desalination. http://www.abc.net.au/news/

stories/2009/06/28/2610554.htm?site=news . Accessed 2 Feb 2011 The Age (2011) Baillieu ‘powerless’ over $24 billion desal bill, 28 February. http://www.theage.

com.au/victoria/baillieu-powerless-over-24-billion-desal-bill-20110228-1bbc5.html . Accessed 3 Mar 2011

Victorian Government (2009) Victorian desalination plant. http://www.ourwater.vic.gov.au/resources/library/resources/factsheets/desalination/Desalination-Fact-Facts_Aug09.pdf . Accessed 2 Feb 2011

Water Corporation (2011a) Thinking 50 years ahead. http://www.watercorporation.com.au/W/water_sources_new.cfm?uid=9809-4354-6198-5978 . Accessed 2 Feb 2011

Water Corporation (2011b) Perth seawater desalination plant. http://www.watercorporation.com.au/D/desalination.cfm . Accessed 2 Feb 2011

Young M, Proctor W, Qureshi E, Wittwer G (2006) Without water, the economics of supplying water to 5 million more Australians. CSIRO and Monash University, Melbourne

http://news.smh.com.au/breaking-news-national/govt-cant-change-desal-contract-baillieu-20110228-1bbjk.html

http://news.smh.com.au/breaking-news-national/govt-cant-change-desal-contract-baillieu-20110228-1bbjk.html

http://www.goldcoast.com.au/article/2011/01/12/282651_gold-coast-flood-watch.html

http://www.goldcoast.com.au/article/2011/01/12/282651_gold-coast-flood-watch.html

http://www.theage.com.au/environment/water-issues/water-experts-slam-abbotts-dam-plan-20110107-19iwj.html

http://www.theage.com.au/environment/water-issues/water-experts-slam-abbotts-dam-plan-20110107-19iwj.html

http://www.adelaidenow.com.au/water-begins-to-flow-from-adelaides-new-desalination-plant/story-e6frea6u-1226105463464



http://www.melbournewater.com.au/content/water_storages/water_report/water_report.asp

http://www.melbournewater.com.au/content/water_storages/water_report/water_report.asp

http://www.brisbanetimes.com.au/queensland/premier-hits-back-at-traveston-rejection-20091111-i9d3.html

http://www.brisbanetimes.com.au/queensland/premier-hits-back-at-traveston-rejection-20091111-i9d3.html

http://www.sydneywater.com.au/WhoWeAre/MediaCentre/documents/ministerial/Keneally_switch%20on_280110.pdf



http://www.floodcommission.qld.gov.au/__data/assets/pdf_file/0006/8781/QFCI-Interim-Report-August-2011.pdf



http://pandora.nla.gov.au/pan/61484/20060810/www.nrm.qld.gov.au/water/water_infrastructure/pdf/long_term_solution.pdf



http://www.qwc.qld.gov.au/security/pdf/seqws-full.pdf

http://www.qwc.qld.gov.au/security/pdf/seqws-full.pdf

http://www.cabinet.qld.gov.au/MMS/MediaAttachments/2010/pdf/32253%20SEQWG%20Wivenhoe%20Fact%20Sheet%20A4%202pp%20F.pdf

http://www.cabinet.qld.gov.au/MMS/MediaAttachments/2010/pdf/32253%20SEQWG%20Wivenhoe%20Fact%20Sheet%20A4%202pp%20F.pdf

http://www.perthnow.com.au/news/western-australia/new-desalination-plant-likely/story-e6frg13u-1111112546154

http://www.perthnow.com.au/news/western-australia/new-desalination-plant-likely/story-e6frg13u-1111112546154

http://www.abc.net.au/news/stories/2009/06/28/2610554.htm?site=news

http://www.abc.net.au/news/stories/2009/06/28/2610554.htm?site=news

http://www.theage.com.au/victoria/baillieu-powerless-over-24-billion-desal-bill-20110228-1bbc5.html

http://www.theage.com.au/victoria/baillieu-powerless-over-24-billion-desal-bill-20110228-1bbc5.html

http://www.ourwater.vic.gov.au/resources/library/resources/factsheets/desalination/Desalination-Fact-Facts_Aug09.pdf

http://www.ourwater.vic.gov.au/resources/library/resources/factsheets/desalination/Desalination-Fact-Facts_Aug09.pdf

http://www.watercorporation.com.au/W/water_sources_new.cfm?uid=9809-4354-6198-5978

http://www.watercorporation.com.au/W/water_sources_new.cfm?uid=9809-4354-6198-5978

http://www.watercorporation.com.au/D/desalination.cfm

http://www.watercorporation.com.au/D/desalination.cfm

163

Abstract Dynamic CGE modeling has been useful in analyzing water policy issues within Australia. This chapter explores the possibility of applying a version of TERM-H2O to other countries.

Keywords China • CGE modeling • International water accounts • Water infrastructure • Water supply variability

9.1 Refl ections on the Australian Experience

The Australian experience has shown that there is a substantial role for CGE modeling to cover water policy issues. Chapter 8 details an urban water scenario using an earlier version of dynamic TERM. Prior to that, a comparative static version of TERM predicted substantial urban water price pressures in the absence of policies to deal with rapidly growing urban populations (Young et al. 2006 ) . In economic mod-eling of irrigation water, TERM-H2O made an important contribution to policy anal-ysis of the Australian government’s water buybacks, as detailed in Chaps. 6 and 7 .

The water buyback issue became a hot topic in Australia late in 2010. There appeared to be a long-standing public acceptance of the need to undertake restora-tion of the health of the Murray-Darling Basin. The Murray-Darling Basin Authority (MDBA) initially commissioned a government department to undertake partial equilibrium modeling of water buybacks (Hone et al. 2010 ) . Their model had a good pedigree, in so far as it allowed farm factors to be mobile between irrigated and dryland farming. This is essential in the Murray-Darling Basin (MDB) where

G. Wittwer (*) Centre of Policy Studies , Monash University , Wellington Rd, Clayton campus , Clayton , VIC 3800 , Australia e-mail: [email protected]

Chapter 9 Applying TERM-H2O to Other Countries

Glyn Wittwer

G. Wittwer (ed.), Economic Modeling of Water: The Australian CGE Experience, Global Issues in Water Policy 3, DOI 10.1007/978-94-007-2876-9_9, © Springer Science+Business Media Dordrecht 2012

164 G. Wittwer

there are 50 hectares of dryland farming for every hectare of irrigation. Dryland production accounts for over 60% of the basin’s farm output. However, a number of policy details required CGE analysis.

9.1.1 CGE Modeling of Water

Why was a CGE model (and preferably a dynamic model) necessary? Modeling showed that buybacks would have only a minor impact on basin employment. Lobbyists, basin communities, and some members of the MDBA board refused to believe that the employment impacts of buyback could be so small – this despite the plan to compensate farmers fully for buyback water, implying that they should have been no worse off. The reason there was disbelief in modeling results is that buy-backs commenced during drought. That is, buybacks were blamed for job losses that arose from drought. As shown in Chap. 7 , drought may have resulted in 6,000 jobs being lost across the basin in the short run while leaving the future employment level in the basin 1,500 jobs below forecast even after a full recovery. The timing of implementation of buybacks made drought modeling policy relevant.

A number of modeling details arose from using a dynamic CGE model. The initial modeling commissioned by the MDBA (Hone et al. 2010 ) did not include payments to farmers. The difference this made to modeling outcomes was marginal. The modeled scenario resulted in 800 job losses without compensation or 500 job losses with compensation (Wittwer 2010 ) . By including buyback payments in TERM-H2O modeling, we made clients and other modelers aware that this was a necessary feature of buyback scenarios. Moreover, the process of thinking about the calculation of both the asset price of water and annual spending of buyback proceeds led to the methodology described in Sect. 6.4 . An important insight from this was the link between the expected frequency of future droughts and the asset value of water. In addition, a CGE model made us think of Australian government’s budget constraint. Buybacks not only compete with other measures dealing with water, such as irrigation infrastructure upgrades, but also other budgetary needs.

One reason dynamics are important is that buybacks should proceed over time – maybe more than a decade (unless the environmental consequences of delay are irreversible). This is because the process is voluntary. Farmers need time to make adjustments, including uptake of water-saving technologies. A dynamic model can account for the impact of water-saving and primary factor technological change over time. In addition, since the frequency of future drought affects both the asset value of water and the purely economic welfare impacts of buyback, the inclusion of some droughts in a baseline is a useful enhancement provided through dynamic modeling. TERM-H2O has become the preferred model for analysis of water policy scenarios in the MDB.

The model is highly disaggregated in the regional dimension so that many regions earn a high proportion of total income from farming. Had we attempted to represent irrigation sectors in a national top-down model, we may have arrived at similar

1659 Applying TERM-H2O to Other Countries

national conclusions as with TERM-H2O but without the insights that we obtained on regional outcomes. That is, we could have defi ned the regions in a top-down module of the national module in which regional prices in all sectors followed national prices. This would have taken away differences that arose between regions that became the subject of careful analysis in Chap. 6 . Indeed, in some small regions, irrigation water’s share of regional income is suffi ciently large that water trading can have signifi cant impacts on the spending power of the region.

Given the above, a CGE model is useful in modeling water; it is diffi cult to understand why doubt remains as to its role. Perhaps, the line of reasoning is that in developed nations, agriculture may account for only 1% or 2% of GDP. In turn, irrigation agriculture may account for only a fraction of total agricultural output, while water’s value share of irrigation agriculture is small. What is missing from this line of reasoning is additional detail that increases the importance of water. First, agriculture may account for a small share of GDP but a much larger share of a national workforce. In China, for example, agriculture’s share of GDP is no more than 6%, yet it accounts for around one third of national employment (Mai and Peng 2009 ) . Globally, agriculture accounts for less than 6% of global GDP, yet it employs around 37% of the global workforce (CIA 2010 ) . Second, even though agriculture accounts for a small percentage of GDP in developed economies, it still accounts for much of the total water used by industries and households in many countries, including Australia. Agriculture uses about 70% of the water used in human activities (UN Water 2009 ) . Agricultural users may compete with urban users. Water is an economy-wide issue.

9.2 Why CGE Modeling of Water may be Useful in Other Countries

Much of the remainder of this chapter deals with the possibility of applying TERM-H2O to countries other than Australia. We have established the need for economy-wide analysis of water in Australia. If anything, the growing competition between competing users of water is greater in many other countries than in Australia. Australia’s water availability is high by global standards by dint of the nation’s low population density (Food and Agriculture Organization 2010 ) . Water scarcity in Australia results from the urban population being concentrated in the south-eastern coastal regions of the nation, and a substantial concentration of agricultural activity in the Murray-Darling Basin. Water resources in northern tropical Australia remain relatively undeveloped.

International confl icts over water are a recurrent theme in history. Some military exchanges in the Middle East have arisen from disputes over water diversion. During World War II, both sides regarded dams as military targets. Flooding induced by the sudden release of reservoir water has been used as a military tool in many confl icts. Tensions simmer among nations that straddle the Nile concerning their access to the lifeline of north-east Africa. During the Gulf War of 1991, Iraq destroyed much of

166 G. Wittwer

Kuwait’s desalination capacity. Ecuador and Peru battled over control of the headwaters of Cenepa River in 1995. Singapore remains nervous of its reliance on water imported from Malaysia and has extended desalination and recycling capacity in response. These are but a few of the numerous examples of international water confl icts. 1

Maybe the number of countries that share river systems and water resources without confl ict is even more notable. Among such nations, establishing water rights and international trading of water could lead to allocative gains. In the sub-national context, reforms in the Murray-Darling Basin have evolved gradually, against the background of competing objectives of various state governments and water author-ities. Reaching the stage of mature markets between nations will be more complex than within a nation. Moreover, the cart cannot come before the horse. That is, mature water markets require properly defi ned and separable land and water rights. With considerable effort, it would be possible to devise a model covering several countries that share a river system. Apart from a multi-regional database, possibly some sub-national detail and multi-country water accounts, this again would require some understanding of water allocation arrangements and key water policy issues within these nations.

Major water projects have transformed cities. Rapid growth in Los Angeles’ population in the twentieth century was made possible by a diversion of water from the Owens Valley. In the space of a few decades from late in the nineteenth century, when white settlers overran the holdings of Native Americans (who had developed irrigation techniques of their own), the valley became an irrigation settlement of 7,000 people and then gradually died as much of the valley land was purchased by Los Angeles City to secure appurtenant water rights (Sauder 1994 ) .

Libya undertook a massive engineering project to pipe fossil water from beneath Saharan sands in the south-east corner of the country to the cities in the north. The man-made river project commenced in 1983 and was completed around a quarter of a century later, tapping over 2,000 GL per annum from the desert aquifer (Watkins 2006 ) . The project may transform lifestyles within Libya by making running water available to much of the population. However, piping water to cities is not equiva-lent to making water available to households. During the 2011 confl ict, Worth ( 2011 ) reported that the city of Benghazi (renowned as a centre of resistance against the Gaddafi regime) had no working sewage system.

It remains to be seen whether Libya’s dependence of food imports will decrease over time. While leaders within the country may regard increased food self-suffi ciency as a worthwhile national goal, it is of less interest to economists who believe that nations should trade in line with comparative advantage. More relevant is the impact that the man-made river project will have on the percentage of Libyans with access to safe drinking water.

1 Dr. Peter Gleick has prepared a chronology of water confl icts downloadable at http://www.world-water.org/confl ictchronology.pdf

http://www.worldwater.org/conflictchronology.pdf

http://www.worldwater.org/conflictchronology.pdf


9.2.1 Rural or Urban Water?

The modeling requirements for urban water scenarios differ from those for rural water. Usually, there is a huge gap between the price paid per ML by irrigators and urban users. In circumstances in which there is no water connection between an urban region and any irrigation region, there may be no need for satellite water accounts in the urban region if competing users pay a similar price for water. This is so in the case study of Chap. 8 .

One task for the modeler is to deal with changes in urban water supply. Section 8.2.3 discusses the conditionality of marginal impacts of water supply augmentation on rainfall patterns. If drought prevails for a long time, the marginal impacts of a new water resource are high. In times of higher rainfall, the marginal benefi ts fall. Although it is not straightforward to translate an increase in water supply into a supply shift within a CGE model, we know that utilization of new water resources will decrease as rainfall increases. We also know the expected project costs. By varying the supply shifts in the industry that includes water supply, we see how the net benefi ts of a project will change with a change in expected rainfall.

Such an approach has some relevance. After the modeling of the now-scrapped Traveston dam, the author was commissioned to model the economic impacts of building the Tillegra dam in the Hunter Valley of New South Wales. 2 This dam was also scrapped (ABC News 2010 ) . The main simulation modeled a substantial wel-fare gain, based on the assumption of continuing rainfall at below the historical level. The report included a section on sensitivity analysis which concluded that under a different set of rainfall assumptions, the net benefi ts of the dam would diminish. Economists are no more immune from changing expectations than engi-neers or indeed fi nancial markets. The report was written just as the drought was breaking in northern New South Wales, and its pessimism concerning future rainfall conditions in the main scenario refl ected expectations at the time.

9.2.2 Irrigation Water

The linking of water accounts to national accounts is a relatively new concept. John Anthony Allen coined the term ‘virtual water’ in 1993 and was awarded the Stockholm Water Prize in 2008 for his contribution to the fi eld (Reuters 2008 ) . The concept fi ts in with trade theory concerning production according to comparative advantage. That is, a water-scarce nation or region should not aim for food self-suffi ciency but rather aim to supply suffi cient water for basic human needs. 3 Nations

2 The report is downloadable from http://majorprojects.planning.nsw.gov.au/fi les/40222/G%20Socio%20Economic%20Assessment.pdf 3 If water is still too scarce, water trading (as in the case of Singapore’s imports from Malaysia) has a role.

http://majorprojects.planning.nsw.gov.au/files/40222/G%20Socio%20Economic%20Assessment.pdf

http://majorprojects.planning.nsw.gov.au/files/40222/G%20Socio%20Economic%20Assessment.pdf

168 G. Wittwer

with a comparative advantage in water abundance, including Australia given its higher per capita endowment of water, are likely to be net exporters of food, helping to meet the requirements of countries with relatively scarce water.

The concept of virtual water has been extended to CGE models including TERM-H2O. This allows us to sharpen our insights into appropriate use of water. A good example in Australia concerns rice production. A recurrent myth is that it is absurd for a ‘water-scarce’ country such as Australia to grow rice. For some commentators, the construction of a pipeline that linked Melbourne’s urban water supply to the Murray-Darling Basin might have been a way of redressing an inappropriate use of water in irrigation. Such a pipeline might have been a cheap way of moving water from a ‘low-value’ use such as rice production to a ‘high-value’ urban uses. In theory, there might be substantial gains from water trading.

What is wrong with this line of thinking? First, it does not take account of the widely varying scarcity of water over time. When water is abundant, farmers may be prepared to sell water to urban water authorities if this is physically possible. Yet if urban catchment dams are at healthy levels, these authorities will lack the incen-tive to buy rural water. That is, the marginal value of additional water will be close to zero. What about when water is scarce? Rural water trading prices increase many-fold during drought. Among irrigators, there will be substantial movement of water away from annuals and irrigated pasture towards perennials. As water scarcity wors-ens, rice production drops dramatically. This is an effi cient outcome. Rice farmers quickly reach a point at which they can earn more income from selling water than using it in production. That is, as the marginal product of water rises by manyfold during drought, it will reach a point at which it exceeds the average product of water in rice production.

During drought, the average annual price of irrigation water rises substantially. In 2007–08, prior to the GFC, the average price in the Goulburn Valley reached $560 per megalitre, falling to around $340 per megalitre in the following year with a partial recovery in dryland productivity and falling commodity prices. Even though urban water users in Melbourne pay in excess of $1,000 per megalitre, the volume of water required to equalize the urban and rural prices (after including pumping costs) would be relatively small once irrigation water has already passed $300 per megalitre. Moreover, much of Melbourne’s fruit and vegetable produce is grown in the basin so that rising water prices for irrigators through urban-rural water trading would result in rising fruit and vegetable prices for Melbourne consumers – which would impose further upward pressure on the price of irrigation water. Water prices during drought appear to refl ect only small ineffi ciencies among users. Redressing these through moving water from irrigation to urban uses would do lit-tle, even if it were politically possible during drought. The actual pipeline con-structed between Eildon and Sugarloaf dams cost $750 million. 4 Only under

4 The costing comes from a Victorian government website http://www.ourwater.vic.gov.au/ programs/water-grid/sugarloaf (accessed 18 February 2011). Pipelines linking regional centres in Victoria to Murray-Darling Basin supplies have higher marginal benefi ts as existing town water supplies were relatively fragile.

http://www.ourwater.vic.gov.au/


exceptional circumstances, when water is abundant in the Murray-Darling Basin and scarce in Melbourne’s catchment, would there be substantial advantages in water trading between the two regions. There is no guarantee that pipeline construc-tion costs could be recovered through future allocative gains. In summary, water trading in the basin does more to ensure effi cient use of water than facilitation of trade between rural and urban users. Irrigation regions, through production of food for urban households, are already engaged in ‘virtual’ rural-urban water trading.

In the case of regional towns adjacent to the basin, the benefi ts from being con-nected to basin water appear more likely to cover costs. The recurrent droughts during the fi rst decade of the new millennium put the water resources of many regional towns in western Victoria under greater strain than those of Melbourne. Among a number of regional projects, the Goldfi elds Superpipe Project connected Ballarat and Bendigo to basin water. These two rapidly growing regional centres each have populations of around 90,000. Moreover, the pipelines were completed in 2008 before the drought had broken, providing an additional water resource at a time it was desperately needed (Offi ce of Water 2008 ) . Geelong, with similarly rapid growth, was connected to Melbourne’s water supply late in 2011. Ballarat, Bendigo, and Geelong are being reinvented as outer suburbs of Melbourne as trans-port links improve and Melbourne’s house prices put ownership out of reach of many. Many regional centres in western Victoria faced stage 4 restrictions, under which all outside watering is banned, for several years. In the case of the Hamilton-Grampians pipeline (connecting Hamilton to the Rocklands Reservoir), a similar problem as that of the Sugarloaf pipeline exists. The Rockland Reservoir fell to 2% of capacity in 2009, indicating that had it been open during the drought, it would have done little to supplement Hamilton’s water supply. 5

9.2.3 Data Required to Prepare a TERM-H2O Variant for Another Nation

One of the fi rst steps the TERM creators advocate in preparation of a multi-regional database is to ensure that the initial national input-output table has suffi cient repre-sentation of agriculture (Wittwer and Horridge 2009 ) . This simplifi es the task of undertaking further modifi cations to include water accounts in the model. As the data preparation tasks are detailed in Chap. 2 , in this section, we will concentrate on additional data required in TERM-H2O.

A useful fi rst step is to have water accounts available at the regional level for the country of interest. These may have a coarser regional split than we intend in our

5 The volume of water in Rocklands Reservoir remained above 8 GL in 2009 when capacity fell to 2% ( http://www.gwmwater.org.au/information/information-reservoir-levels ). This volume would have provided a useful supplement to Hamilton, with a population of less than 10,000, but water quality is an issue at such a low capacity level.

http://www.gwmwater.org.au/information/information-reservoir-levels

170 G. Wittwer

Table 9.1 Countries included in the UN survey on water accounts

Surveyed countries with or planning water accounts No plans

Andorra France Occupied Palestinian territory Azerbaijan Armenia Germany Peru Bosnia Australia Greece Philippines Brazil Austria Guatemala Portugal Chile Bahamas Hungary Romania Cyprus Belgium Iraq Singapore Czech Botswana Israel South Africa India Bulgaria Italy Spain Indonesia Canada Jordan Sweden Luxembourg China Lebanon Switzerland Malaysia Eastern Mauritius Trinidad and Tobago Kenya Colombia Mexico Tunisia Poland Denmark Namibia Turkey Serbia Dominican Republic Netherlands Ukraine Slovenia Egypt New Zealand United Kingdom United Arab Emirates Estonia Norway

Source : UN ( 2009 )

database. This is not an insurmountable problem. We can devise rules based on activity shares or data on cultivated hectares to split the coarser numbers into smaller regions. It might be the case that the sectoral split of water usage is also much coarser than we desire. If so, we may be able to borrow water requirements from those of another country.

It appears that water accounts data are gradually improving. A UN ( 2009 ) survey found that a majority of surveyed countries either have water accounts at present or intend to do so in the future (Table 9.1 ).

The compilation of water accounts is a complex process. The Australian Bureau of Statistics (ABS 2010 ) explains in detail how it matches water accounts to national accounts sectors. The water accounts devised by the ABS follow the System of Integrated Environmental and Economic Accounting ( 2006 ) . In practice, the process of estimating water accounts comes from combining many data sources. These include an array of ABS surveys, data from state and territory government agencies, surveys of industry associations, and annual reports of water providers.

In Chap. 5 , we mention that it would take considerable effort to devise a split between dryland and irrigation agriculture outside the Murray-Darling Basin in Australia. It follows that to attempt a similar split in the regions of other countries would be challenging. Nevertheless, it will be possible with time and perseverance to come up with defensible splits. For example, data are available on the hectares of dry land and irrigated land for 17 of the 31 provinces/municipalities of China. Putting together a database of key irrigation regions that includes water accounts is a process that could be revisited as better data emerge. The practitioner needs to write data preparation programs that can be modifi ed and run again as better data emerge. Often, writing programs ends up being a time-consuming task, but it is


necessary in highly complex data manipulation exercises. Improvements can be made to a sub-national database through experience and persistence.

9.2.4 Evolving Accounts and a Suitable Base Year

In Chap. 6 , we examine a scenario in which 1,500 GL of water is purchased by the Australian government from irrigators in the southern Murray-Darling Basin. It would appear that this volume arose early in discussions on environmental fl ows when basin-wide water accounting was in its early stages (see Wentworth Group of Concerned Scientists 2002 ) . That is, there was uncertainty as to the absolute volume of water used in irrigation in the basin. Later, the Murray-Darling Basin Authority (MDBA 2010 ) was examining targets of between 3,000 and 4,000 GL for environmental fl ows. Chapter 6 treats 2005–06 allocations as though they represent 100% entitlements. This was not the case. Following a severe drought in 2002–03 and a patchy recovery in the years that followed, basin-wide allocations in 2005–06 were still more than 25% below those of a typical year. The early buyback simulations run with TERM-H2O were towards the end of the 3-year drought (2006–07 to 2008–09) in the south-ern basin. At that stage, 2005–06 appeared to be more typical than the three following years in which allocations were much lower. The 1,500 GL of water removed from production in the Chap. 6 simulation, based on 2005–06 allocations, is equivalent to around 2,500 GL in a typical year. In subsequent modeling of buybacks undertaken for MDBA (Wittwer 2010 ) , volumes sold to the Australian government were expressed as proportions of entitlements rather than absolute volumes. The MDBA irrigation water entitlement base for the entire basin is around 10,900 GL, compared with allo-cations of 7,720 GL in 2005–06 (ABS 2008 , Table 3.20).

From the perspective of farmers and environmentalists alike, the greater the volume of water available, the easier it is to allocate water for all uses. The marginal product value of water increases manyfold during prolonged drought. This means that economic losses arising from moving entitlements to the environment during drought result are larger than is so during normal years. At the same time, smaller volumes of water are available for the environment during prolonged drought, as both farmers and environmental holders of water suffer cuts in water allocations.

9.3 China

Complex models may take many years to develop. However, this can be a positive experience. Improved data may emerge over time. Practitioners may develop an understanding of policy issues in the process of model development. They may make useful industry contacts. However, there are also pitfalls in developing a model as complex at TERM-H2O. The danger exists that substantial effort might go into model development without a matching effort on policy simulations and analysis of the results.

172 G. Wittwer

An early task to make model development tractable might be to determine which issues and which regions within a country are important. To explore such priorities, we turn to the example of China. On the surface, there are many water issues that could be modeled within China. Population numbers even at the provincial level dwarf those in most other nations, magnifying water policy issues.

These issues in part arise from the massive migrations of people from inland regions to urban centres on the eastern seaboard. Migrations in turn reduce farm land availability as urban demands grow. Growing urban water demand demands compete with farm demands for water. Another issue concerns climate change in the north. Reduced snowfalls, drought, and soil degradation are affecting farm pro-duction. Underinvestment in agricultural R&D has almost certainly compounded water pollution, as effi ciency improvements in fertilizer use over time, for example, might have done more to reduce pollution from runoff. Aquifer water is being depleted rapidly to maintain water supplies for agricultural and non-agricultural users. Underground water levels are dropping by a metre a year or more, threatening future supplies and raising the risk of contamination. At the same time, as economic growth and climate change are imposing immense strain on water resources in China, there are ongoing efforts to improve water quality. 6 In this context, we turn to the unprecedented engineering effort that has started in China to redress water shortages in the north, namely, the South-to-North Water Diversion Project.

9.3.1 South-to-North Water Diversion Project

Originally, there were three proposed routes for the project. The eastern route entails upgrading of the Grand Canal, a not-quite-contiguous artifi cial waterway almost 1,800 km long that runs between Jiangdu City and Beijing. 7 Water will need to be pumped uphill along parts of the canal. The project also includes construction of a tunnel under the Yellow River, from where it will fl ow downhill to Tianjin. Construction began in December 2002. Pollution from agriculture and other indus-tries has delayed the project, which authorities expect to be completed in 2013. 8 The eastern route will supplement water supplies in the coastal provinces of Anhui, Hebei, Jiangsu, and Shandong and the municipality of Tianjin.

The central route will divert water from Danjiangkou Reservoir on a tributary of the Yangtze River to Beijing. Construction commenced in 2004. Prior to the Beijing Olympics in 2008, the 307-km northern segment of the route was completed at a cost of US$2 billion. Although this provided a display of dams visible to visitors

6 Circle of Blue are documenting ongoing water pollution issues in China at http://www.circleof-blue.org/waternews/2011/world/infographic-chinas-water-pollution-events-and-protection-policies-2004-2011/ . 7 The Grand Canal starts at Hangzhou, almost 300 km to the south of Jiangdu City. 8 From http://www.water-technology.net/projects/south_north/

http://www.circleofblue.org/waternews/2011/world/infographic-chinas-water-pollution-events-and-protection-policies-2004-2011/



http://www.water-technology.net/projects/south_north/


fl ying into Beijing at the time, it entailed a transfer from water-stressed Hebei province to Beijing rather than from the relatively water-abundant south. 9 Farmers and industries in Hebei had to cut back on water usage to assist in the display, which was more mirage than miracle given the many years of drought. A tunnel to be built under the Yellow River is the greatest engineering challenge of the central route project. As the project has proceeded, environmental diffi culties associated with it have resulted in delays. The expected completion date will now be in 2014. A considerable expense in the project will involve the relocation of 330,000 people from around Danjiangkou Reservoir and along the water route. In October 2009, the affected inhabitants of Hubei and Henan provinces commenced relocation. 10 The central line will supplement water supplies in the municipalities of Beijing and Tianjin and the provinces of Hubei, Henan, and Hebei.

The Chinese government has delayed plans for the western route of the project. This entails seven dams and a 1,000 km of tunnels in western Sichuan and Qinghai to tap into Yangtze tributaries. Lu Jiagua, a retired economist from the Sichuan Academy of Social Sciences, spearheaded a campaign to stop the route. He twice sent letters to Premier Wen Jiabao in 2005. Among the diffi culties of the project, the planned route would cross fi ve fault lines in western Sichuan. Lu Jiagua feared that an earthquake that caused cracks in new dams would fl ood millions of homes. He estimated that diminished fl ow in the Three Gorges Dam as a consequence of the western route would reduce hydroelectricity output from the dam by the monetary equivalent of billions of dollars per annum. Then the Sichuan earthquake of 2008 appeared to halt the project. The central government cut off funding to the key plan-ning group (Mufson 2010 ) . Since then, plans for the western route appear to have been put back on the table. The intent of the western route is to supplement supplies so as to enable exploitation of energy resources, particularly coal, in the Yellow River basin provinces of Ningxia, Inner Mongolia, and Shanxi.

9.3.2 An Economic Perspective

A project as massive as the three routes of the South-to-North Water Diversion Project involves tiers of complexity. First, there is the intent in terms of geography. The eastern and central routes of the project will supplement water supplies in a number of cities in six provinces and two municipalities. The western route appears to be directed at development of energy resources rather than for urban water uses in the less densely populated regions of Inner Mongolia, Ningxia, and Shanxi (Ivanova 2011b ) . Second, there is the urgency for additional water particularly in

9 Local authorities rather than individual farmers receive compensation for water diverted from agriculture to urban use (Ivanova 2011a ) . 10 See http://community.travelchinaguide.com/forum2.asp?i=53796 (accessed 18 February 2011).

http://community.travelchinaguide.com/forum2.asp?i=53796

174 G. Wittwer

Hebei, Beijing, and Tianjin as a consequence of rapid population growth, rapid per capita income growth, and a drought that has prevailed from more than a decade. From a regional economic perspective, urban demands have taken water away from the rural poor at a time when Chinese authorities are becoming increasingly con-cerned by the disparities between the booming eastern seaboard cities and the rest of China. A similar divide exists between urban and rural households.

Since the projected was approved in 2002, the costs relative to potential alterna-tive projects have changed. Already, massive costs have been incurred, not least with the commencement of a massive relocation of villagers around Danjiangkou Reservoir. For the coastal cities, desalination as an option has become relatively more attractive, as the energy requirements and consequent unit costs have fallen over the past decade. At the same time, with China’s booming economy, labour costs have risen sharply, in turn raising the construction costs of the South-to-North Project. In addition, the costs of treating polluted water on the eastern route are much higher than fi rst anticipated. In summary, the costs of alternatives such as desalination have fallen, while the costs of the South-to-North Project have risen since the project was approved in 2002 (Jaffe and Schneider 2011 ) .

The creation of water infrastructure often refl ects concerns about supply risk. Water engineers in developed nations tend to be extremely risk averse when it comes to urban water supply management, which in turn implies that they are quite reluc-tant to consider demand-side management strategies. In many nations, environmen-tal impact studies are a mandatory part of major projects. Without such studies, engineers may have little motivation to minimize environmental consequences. Engineers arguably dominate economists even more so in China than the developed world when it comes to water management. Both the World Bank ( 2001 ) and Shalizi ( 2006 ) mention the underuse of pricing as a demand management tool in China. There is a long way to go.

Some reforms under way in Australia that may be some way from maturity have not even started in China. High on the Council of Australian Governments agenda in the mid-1990s in relation to irrigation water reform was the disentanglement of water rights from land ownership. The opportunity to trade water across regions within a river basin has a positive impact on agricultural effi ciency, particularly when allocations vary widely across regions due to variations in seasonal condi-tions. In the case of China, land and water became the property of the state under communist rule. Farmers do not have secure access to water, as was evident during the 2008 Olympics. Without ownership and with limited opportunities to trade water, the incentives to improve water effi ciency are lacking.

It is diffi cult to envisage China undertaking reforms in which land ownership and water rights are clearly defi ned and transferable. Ivanova ( 2011b ) notes that indus-tries seeking to obtain access to water for development of energy resources in Ningxia and Inner Mongolia must fi rst upgrade existing irrigation infrastructure so as to save water. From an economic perspective, this appears to be far from the best strategy for water management. Among various diffi culties with the current approach is that of reliably measuring realized water savings.


Probe International ( 2008 ) in conjunction with anonymous experts in Beijing has proposed a number of strategies for dealing with Beijing’s water needs that echo the World Bank ( 2001 ) and Shalizi ( 2006 ) . These include the introduction of full-cost pricing, the establishment of an independent water industry regulator, and assign-ment of tradable water rights. Full-cost pricing would encourage water users to increase water effi ciency. It would promote wastewater and sewage treatment. Some of the concerns of the Probe International ( 2008 ) report arise from elements of con-spicuous consumption that are far removed from sustainable water use. These include the construction of new golf courses and artifi cial ski fi elds in the Beijing region.

Given the state of China’s water systems, alternative investments to the South-to-North Water Diversion Project might include extending and upgrading water recy-cling and purifi cation plants and desalination plants in coastal cities, as indeed is already happening. A number of sewage treatment plants have been built in Beijing since the 1980s (MacLean 2008 ) . The central government of China has funded water treatment plants away from the capital. However, many remained idle as local author-ities attempted to cut costs (Mufson 2010 ). Tensions between the intent of the central government and the actions of other tiers of government are pervasive in China.

Economic tools including CGE analysis may help assess the potential contribu-tion and costs of alternative projects to stretch urban water supply in northern China. Already, Beijing appears to have acted to increase the rate of recycling, cited as only 15% of water used a few years ago (compared with 85% in many industrial nations) by Probe International ( 2008 ) . More recent estimates are that Beijing recycled around 60% of wastewater in 2010, with the rate expected to reach 75% by 2015 (Ivanova 2011a ) .

9.3.3 What Is a Suitable CGE Model for the Project?

Already, a TERM version has been developed for China but with limited applications (Wittwer and Horridge 2009 ; Horridge and Wittwer 2008 ). In the regional dimension, the modeler could represent the most affected provinces and municipalities separately while aggregating those provinces to the south and west not directly affected by the project. Model development would require estimates of water uses by crop and region, together with some estimate of water obtained from natural rainfall. There would be some advantage in representing some regions at the sub-provincial level. The yearbooks of some Chinese provinces contain suffi cient data to split pro-vincials at the input-output level of sectoral representation with some confi dence. 11

What use would water accounts be in a Chinese version of the model? Water accounts appear in TERM-H2O because the initial price of water may vary between users. As water availability changes, optimal usage of water changes markedly between irrigation activities. Rice producers, for example, cut production by a much

11 Our main access to such data is via the University of Michigan’s China Data Center.

176 G. Wittwer

larger percentage than the fall in water availability. In China, even if the massive water diversion project secures a constant supply of new water, water requirements will vary from year to year depending on rainfall. In theory, water could move between crops. A fi rst step in establishing the usefulness of water accounts would be to obtain data on water usage by crop in regions of interest. Such data from year to year would help us establish how mobile water is between users. In Australia, where water is quite mobile in the southern Murray-Darling Basin between users, economists still debate the short-run mobility of farm factors with other analysts. Further research is required to evaluate the mobility of water in particular regions of China. Alternatively, water accounts might be a way of running hypothetical sce-narios to examine what might happen if water markets were developed suffi ciently to allow relatively free trading.

A starting point for a CGE model of China with water accounts might be to con-centrate on the 11 prefectures of Hebei and the municipalities of Beijing and Tianjin, and then represent the rest of China with a single region. The Hebei Provincial Yearbook includes a chapter with some county-level data that might help estimate prefecture output levels by agricultural activity.

9.4 Conclusion

Whether a standard version of dynamic TERM or a version that includes water accounts is most suitable for analysis depends on the nature of the water issue at hand. A relatively standard version of TERM may be suitable for an urban water supply scenario. This assumes that new dam is being built, for example, and that water is not being diverted from irrigation purposes. If irrigation activity is nega-tively affected by the project, it might be possible to use a version of TERM without water accounts by estimating the productivity losses arising from a decline in water availability for farm production, as indeed was done in the fi rst ever application of TERM (Horridge et al. 2005 ). But this is a typical dilemma for a modeler: does one make do with an existing model, or can modifi cations lead to additional insights? At the heart of this dilemma is whether the modeler has the resources, confi dence, and energy to make further modifi cations.

For most practitioners, moving from a comparative static TERM to a dynamic version of TERM is daunting enough, and few have attempted this without assis-tance from the model’s main architects. Each of the enhancements made to TERM, in moving from a comparative static to dynamic model and fi nally to a model with water accounts, is non-trivial and time-consuming. Model building is too demand-ing to be an end in itself. An important reason for modeling is to gain insights into policy scenarios that otherwise may not emerge readily. The objective of CGE mod-eling should be to have some infl uence in policy debate. Such modeling may not appeal to those seeking novelty rather than usefulness among analytical tools.

In theory, one expects major projects in Western democracies to be subjected to different types of analyses in order to comply with various government regulations.


These might include cost-benefi t analysis and environmental impact analysis, for example. In practice, there are exceptions: politics often obstruct more objective anal-ysis. Section 8.1 shows that cost-benefi t analysis has played little part in Australia’s urban water management. On the other hand, concerning the Murray-Darling Basin, economic analyses including CGE modeling have played an important role.

In some respects, in nations that are not democratic or have weakly evolved democratic institutions, major projects are even less likely to be subjected to inde-pendent analysis. The justifi cation for the South-to-North Water Diversion Project in China appears to be based on outdated notions of nation building, albeit in response to massive water management issues, rather than any rigorous analysis. Academics and others who raise concerns about projects and their possible unin-tended environmental side effects struggle to be heard. It is not surprising, therefore, that CGE modeling to evaluate major projects is not yet in high demand within the Chinese government. Our aim as modelers should be that government departments in nations such as China eventually demand independent modeling as one way of assessing the social worth of major projects.

References

ABC News (2010) Tillegra Dam plans scrapped. http://www.abc.net.au/news/stories/2010/11/28/3078524.htm . Accessed 18 Feb 2011

ABS (Australian Bureau of Statistics) (2008) Water and the Murray Darling Basin: a statistical profi le, 2000–01 to 2005–06. Catalogue 4610.0.55.007. Australian Bureau of Statistics, Canberra

ABS (2010) Water account, Australia, 2008–09. http://www.abs.gov.au/AUSSTATS/[email protected]/Lookup/4610.0Explanatory%20Notes12008-09?OpenDocument . Accessed 18 Feb 2011

CIA (2010) The world factbook. https://www.cia.gov/library/publications/the-world-factbook/geos/xx.html . Accessed 18 Feb 2011

Food and Agriculture Organization (2010) AQUASTAT. http://www.fao.org/nr/water/aquastat/ . Accessed 18 Feb 2011

Hone S, Foster A, Hafi A, Goesch T, Sanders O, Mackinnon D, Dyack B (2010) Assessing the future impact of the Australian Government environmental water purchase program. http://www.abare.gov.au/publications_html/landwater/landwater_10/waterbuyback.pdf . Accessed 18 Feb 2011

Horridge M, Wittwer G (2008) SinoTERM, a multi-regional CGE model of China. China Econ Rev 19:628–634


Ivanova N (2011a) Off the deep end – Beijing’s water demand outpaces supply despite conserva-tion, recycling, and imports. http://www.circleofblue.org/waternews/2011/world/off-the-deep-end-beijings-water-demand-outnumbers-supply-despite-conservation-recycling-and-imports/ . Accessed 11 May 2011

Ivanova N (2011b) Water rights transfer and high-tech power plants hold off energy-water clash in Northern China. http://www.circleofblue.org/waternews/2011/world/from-agriculture-to-industry-effi ciency-upgrades-transfer-water-use-rights-on-china%E2%80%99s-yellow-river/ . Accessed 17 Aug 2011

Jaffe A, Schneider K (2011) A dry and anxious North Awaits China’s Giant, unproven water transport scheme. http://www.circleofblue.org/waternews/2011/world/a-dry-and-anxious-north-awaits-china%E2%80%99s-giant-unproven-water-transport-scheme/ . Accessed 12 May 2011



http://www.abs.gov.au/AUSSTATS/[email protected]/Lookup/4610.0Explanatory%20Notes12008-09?OpenDocument

http://www.abs.gov.au/AUSSTATS/[email protected]/Lookup/4610.0Explanatory%20Notes12008-09?OpenDocument

https://www.cia.gov/library/publications/the-world-factbook/geos/xx.html

https://www.cia.gov/library/publications/the-world-factbook/geos/xx.html

http://www.fao.org/nr/water/aquastat/



http://www.circleofblue.org/waternews/2011/world/off-the-deep-end-beijings-water-demand-outnumbers-supply-despite-conservation-recycling-and-imports/

http://www.circleofblue.org/waternews/2011/world/off-the-deep-end-beijings-water-demand-outnumbers-supply-despite-conservation-recycling-and-imports/

http://www.circleofblue.org/waternews/2011/world/from-agriculture-to-industry-efficiency-upgrades-transfer-water-use-rights-on-china%E2%80%99s-yellow-river/

http://www.circleofblue.org/waternews/2011/world/from-agriculture-to-industry-efficiency-upgrades-transfer-water-use-rights-on-china%E2%80%99s-yellow-river/

http://www.circleofblue.org/waternews/2011/world/a-dry-and-anxious-north-awaits-china%E2%80%99s-giant-unproven-water-transport-scheme/

http://www.circleofblue.org/waternews/2011/world/a-dry-and-anxious-north-awaits-china%E2%80%99s-giant-unproven-water-transport-scheme/

178 G. Wittwer

Mai Y, Peng X (2009) Estimating rural labour surplus in China – a dynamic general equilibrium analysis. https://www.gtap.agecon.purdue.edu/resources/res_display.asp?RecordID_3064 . Accessed 18 Feb 2011

MacLean J (2008) Country water action: People’s Republic of China Beijing beefs up sewage treatment. Asian Development Bank. http://www.adb.org/water/actions/PRC/Beijing-Sewerage-Plant.asp . Accessed 18 Oct 2011

MDBA (2010) Guide to the proposed basin plan. http://download.mdba.gov.au/ Guide_to_the_Basin_Plan_Volume_1_web.pdf . Accessed 7 Feb 2011

Mufson S (2010) As economy booms, China faces major water shortage. Washington Post, March 16. http://www.washingtonpost.com/wp-dyn/content/article/2010/03/15/AR2010031503564.html . Accessed 18 Oct 2011

Offi ce of Water (2008) Goldfi elds sugarpipe. http://www.water.vic.gov.au/programs/water-grid/goldfi elds . Accessed 18 Feb 2011

Probe International Beijing Group (2008) Beijing’s water crisis 1949–2008 Olympics. http://www.chinaheritagenewsletter.org/016/_docs/BeijingWaterCrisis1949-2008.pdf . Accessed 18 Feb 2011

Reuters (2008) Scientist who invented “virtual water” wins prize. http://www.reuters.com/article/2008/03/19/us-water-idUSL1850227220080319 . Accessed 18 Feb 2011

Sauder R (1994) The lost frontier: water diversion in the growth and destruction of Owens valley agriculture. The University of Arizona Press, Tucson

Shalizi Z (2006) Addressing China’s growing water shortages and associated social and environ-mental consequences. World Bank policy research working paper 3895

UN (2006) Integrated environmental and economic accounting for water resources. http://unstats.un.org/unsd/envaccounting/ceea/PImeetings/Handbook_Voorburg.pdf . Accessed 18 Feb 2011

UN (2009) Report on the global assessment of water statistics and water accounts. http://unstats.un.org/unsd/statcom/doc09/BG-WaterAccounts.pdf . Accessed 18 Oct 2011

UN Water (2009) Water in a changing world http://www.unesco.org/water/wwap/wwdr/wwdr3/tableofcontents.shtml . Accessed 18 Feb 2011

Watkins J (2006) Libya’s thirst for “fossil” water. http://news.bbc.co.uk/2/hi/science/nature/4814988.stm . Accessed 18 Feb 2011

Wentworth Group of Concerned Scientists (2002) Blueprint for a living continent. http://www.wwf.org.au/publications/blueprint_for_a_living_contintent/ . Accessed 1 Mar 2011


Wittwer G, Horridge M (2009) A multi-regional representation of China’s agricultural sectors. China Agric Econ Rev 1:420–434

World Bank (2001) China air, land and water: environmental priorities for a new millennium. World Bank, Washington DC

Worth R (2011) On Libya’s revolutionary road. http://www.nytimes.com/2011/04/03/magazine/mag-03Libya-t.html? Accessed 10 Apr 2011

Young M, Proctor W, Qureshi ME, Wittwer G (2006) Without water: the economics of supplying water to 5 million more Australians. Water for a Healthy Country Flagship report. CSIRO and Centre of Policy Studies, Monash University, Melbourne

https://www.gtap.agecon.purdue.edu/resources/res_display.asp?RecordID_3064

http://www.adb.org/water/actions/PRC/Beijing-Sewerage-Plant.asp

http://www.adb.org/water/actions/PRC/Beijing-Sewerage-Plant.asp

http://download.mdba.gov.au/%20Guide_to_the_Basin_Plan_Volume_1_web.pdf

http://download.mdba.gov.au/%20Guide_to_the_Basin_Plan_Volume_1_web.pdf

http://www.washingtonpost.com/wp-dyn/content/article/2010/03/15/AR2010031503564.html

http://www.water.vic.gov.au/programs/water-grid/goldfields

http://www.water.vic.gov.au/programs/water-grid/goldfields

http://www.chinaheritagenewsletter.org/016/_docs/BeijingWaterCrisis1949-2008.pdf

http://www.chinaheritagenewsletter.org/016/_docs/BeijingWaterCrisis1949-2008.pdf

http://www.reuters.com/article/2008/03/19/us-water-idUSL1850227220080319

http://www.reuters.com/article/2008/03/19/us-water-idUSL1850227220080319

http://unstats.un.org/unsd/envaccounting/ceea/PImeetings/Handbook_Voorburg.pdf




http://www.unesco.org/water/wwap/wwdr/wwdr3/tableofcontents.shtml

http://www.unesco.org/water/wwap/wwdr/wwdr3/tableofcontents.shtml

http://news.bbc.co.uk/2/hi/science/nature/4814988.stm

http://news.bbc.co.uk/2/hi/science/nature/4814988.stm





http://www.nytimes.com/2011/04/03/magazine/mag-03Libya-t.html?

http://www.nytimes.com/2011/04/03/magazine/mag-03Libya-t.html?

179G. Wittwer (ed.), Economic Modeling of Water: The Australian CGE Experience, Global Issues in Water Policy 3, DOI 10.1007/978-94-007-2876-9, © Springer Science+Business Media Dordrecht 2012

Prof. Peter B. Dixon had his Ph.D. awarded by Harvard University in 1972. After working at the International Monetary Fund and the Reserve Bank of Australia, Dixon joined the IMPACT Project in 1975 under the direction of Prof. A. A. Powell. With Powell, he was the joint recipient of the 1983 Research Medal of the Royal Society of Victoria given in recognition of the outstanding contribution of the IMPACT Project to social science research in Australia over the preceding 5 years. He was elected a Fellow of the Academy of Social Sciences in 1982 and is one of only ten economists working in Australia included in Who’s Who in Economics, A Biographical Dictionary of Major Economists, 1700–1986 . Inclusion in the Who’s Who is based on international citations. In 1990, he was the Giblin lecturer at the 59th ANZAAS Congress.

Dixon is known internationally for his work in computable general equilibrium modeling. Together with colleagues at the IMPACT Project and the Centre of Policy Studies, he created the ORANI model and its dynamic successor, MONASH. These models have been prominent in the Australian economic debate for 35 years and have been used as templates for the development of other models throughout the world. He is the principal author of the ORANI and MONASH books published in the North Holland Contributions series in 1982 and 2002. In recent years, he has led the development of the USAGE model of the USA which is being used by the US International Trade Commission and the Departments of Agriculture, Commerce, and Homeland Security.

Dixon’s publication list contains about 200 articles and 7 books, including three North Holland Contributions monographs. In 2009, he was appointed editor (with Dale Jorgenson) of Elsevier’s Handbook of Computable General Equilibrium Modeling .

In 2003, Dixon was awarded the Distinguished Fellowship of the Economic Society of Australia. In 2006, he was appointed Sir John Monash Distinguished Professor by Monash University.

Dr . Marnie Griffi th joined the Centre of Policy Studies as a Research Fellow in 2007. Prior to this, she worked in private consulting and later completed her Ph.D.

About the Authors

180 About the Authors

on irrigator decision-making under hydroclimatic uncertainty through the Department of Civil Engineering at Monash University. This research simulated water use and trading decisions in the face of supply and demand uncertainty. Since joining the Centre, she has combined expertise in water issues with CGE modeling.

Prof. Mark Horridge is a global expert on CGE databases and software. He has contributed substantially to or developed CGE databases for Albania, Brazil, China, Finland, Indonesia, Jersey, New Caledonia, New Zealand, Philippines, Senegal, South Africa, Taiwan, Thailand, and Vietnam. He is the Director of GEMPACK software, a specialist CGE package used by over 400 organizations in over 70 coun-tries. He is primarily responsible for developing a number of GEMPACK tools. He is the creator of the TERM model and its predecessor MMRF, and also contributed to the development of the dynamic MONASH model. He has developed a unifi ed database that underlies these Australian models. Professor Horridge has played an important role in development of teaching materials for short CGE courses, and is in high demand as an instructor at these courses around the world. For many years, he has also instructed at the GTAP short courses.

Prof. Maureen T. Rimmer is the author and coauthor of over 50 scholarly pub-lished articles, appearing in mathematics and economics journals and edited vol-umes. Her main area of expertise is in model development and application. She is the coauthor of numerous consultancy reports from the Centre of Policy Studies. With Peter Dixon, she is the codeveloper of the dynamic MONASH CGE model of the Australian economy and the coauthor of the MONASH book. In the last 10 years, Prof. Rimmer has been a key contributor in the development, application, and docu-mentation of USAGE. This is a 500-industry, dynamic model of the US economy, with facilities for generating results for the 50 States and 700 occupations. The model is used in Washington by the US International Trade Commission and the US Departments of Commerce, Homeland Security, Agriculture, and Energy. Apart from the design and implementation of USAGE, Prof. Rimmer has made major contributions in applications of the model to key policy areas such as: the replace-ment of imported crude oil with domestically produced biofuels, legalization of unauthorized immigrants, and an analysis of the 2008–2009 US recession with and without the Obama stimulus package. In 2009, she was joint winner of a Dean’s Award for Excellence in Research in the Faculty of Business and Economics at Monash University.

Dr. George Verikios has worked on developing and applying MONASH-Health, a detailed dynamic model of the Australian economy with special emphasis on eco-nomic aspects of health. He has also recently contributed to research estimating the regional and global economic effects of infl uenza pandemics and epidemics. Dr. Verikios has also undertaken research in the areas of multilateral manufacturing and services trade liberalization, unilateral trade liberalization by Australia, the distribu-tional effects of microeconomic reform of infrastructure industries in Australia, and

181About the Authors

productivity and alternative measures of real output for service industries in Australia.

Dr. Glyn Wittwer is a Senior Research Fellow at the Centre of Policy Studies, Monash University, Melbourne, Australia. He joined the Centre in 2001. His exper-tise is in regional CGE modeling. He has played a major role with Mark Horridge in developing databases for TERM versions for Australia and China. In addition, he has developed a TERM version of the US economy. In 2007, he was joint winner of a CSIRO Land and Water Strategic Excellence award for his modeling of future urban and rural water demands and supplies. He has been the major developer of dynamic TERM-H2O. He has used the model extensively in work commissioned by various state and Australian government departments.

183G. Wittwer (ed.), Economic Modeling of Water: The Australian CGE Experience, Global Issues in Water Policy 3, DOI 10.1007/978-94-007-2876-9, © Springer Science+Business Media Dordrecht 2012

A Abbott, Tony , 160 Aggregation

procedure , 43, 45, 49–52 regional , 31, 32 sectoral , 31, 43 weighted , 49–51

AnalyseGE , 53 Armington sourcing , 20 Asset value calculation , 137, 164 Australian Bureau of Statistics (ABS)

Agstats data , 28 census data , 28, 93, 94, 139 commodity cards , 28 government fi nancial statistics , 42 manufacturing census data , 28 state yearbooks , 28

Australian Water Market Report , 131

B Back-of-the-envelope ((BoTE)) calculations ,

103, 104 Balearic Islands , 71, 72 Ballarat , 169 Barnett, Colin , 7 Basic values , 17, 24 Beijing , 8, 172–176 Bendigo , 95, 106, 131, 169 Bligh, Anna , 159 Bottom-up , 3, 15, 41,

52, 80 Brazil , 9, 33, 170 Brisbane River valley , 153, 161 Brumby, John , 7

Bureau of Meteorology , 100, 119, 120, 126, 138, 160

C Capital accumulation , 39, 44, 47–48 Cenepa River , 166 CES. See Constant elasticity

of substitution (CES) CGE modeling , 13, 26, 37, 38, 52, 70–74, 121,

153, 163–171, 176, 177 China , 1, 8, 9, 26, 33, 38, 165,

170–177 Churchill County, Nevada , 71 Climate change , 38, 39, 61,

69–71, 172 Closure fl exibility , 25 COAG. See Council of Australian

Governments (COAG) Coleambally irrigation area , 61 Commonwealth environmental

water holder , 68 Constant elasticity of substitution (CES) ,

20–22, 24, 25, 82, 84–87, 121, 128 Constant elasticity of transformation (CET),

farm factor movements , 129 Consumer taste changes , 38, 45 Consumption, aggregate , 111 Cotton industry , 105 Council of Australian Governments (COAG) ,

7, 60, 61, 63, 67, 100, 101, 113, 115, 174

Crohamhurst , 161 CSIRO , 68 Culling of livestock , 130–131

Index

184 Index

D Dairy industry , 5, 67 Danjiangkou Reservoir , 172, 174 Data requirements , 14, 26 Delivered values , 17, 24 Demand management , 149, 152,

160, 174 Deniliquin rice mill , 115, 124 Desalination

Gold Coast , 143–145, 148, 154 Kurnel , 146 Kuwait , 166 Kwinana , 145, 148, 158 Port Stanvac , 146, 148 Wonthaggi , 7, 144, 147

Destocking , 130 Dimensionality of database , 16–17 Dormant rights , 65 Drought

2002–03 , 3, 5, 6, 122, 135, 136 drought proofi ng , 119, 160

Dynamics. See modelling

E Effective land , 84–86, 88, 89 Eildon Reservoir , 144 Environmental fl ows , 64, 66, 68–70,

100, 113, 120, 139–140 Equity , 7, 113 Excess capacity , 124–126

F Factor prices , 74, 80, 113 FAO , 6 Farm factor mobility , 4, 6, 97, 122 Final demand aggregates , 27 Finland , 9, 33 Fiscal accounts , 41–43 Fixed costs , 147, 149 Flood mitigation , 7, 153, 157, 159–161 Fruit , 62, 72, 92–94, 96, 107, 108, 122,

130, 168

G Geelong , 169 GEMPACK software , 14, 52, 53, 126 Gillard, Julia , 42, 100, 115 Goldfi elds Superpipe Project , 169 Goulburn-Murray Water , 67, 120 Goulburn Simulation Model ,

60, 65, 66

Grapes , 28, 62, 92–94, 96, 105, 107, 113, 115, 122, 130

Gravity formula , 30 Groundwater , 67, 71, 72, 96, 100,

144, 158 GTAP , 73, 75 GTAP-W , 25, 123, 126

H Hamilton , 169 Hebei, Hebei Provincial Yearbook , 176 Howard, John , 99 Hunter Valley , 167

I IMPLAN , 1, 2 Indonesia , 9, 33, 170 Infrastructure , 6, 7, 43, 60, 63, 64,

66, 100, 144, 146, 150–159, 164, 174

Inner Mongolia , 173, 174 Input demand functions , 82–89 Input-output database , 17, 27–28, 43 Inter-regional labour movements ,

25, 41 Inter-regional trade matrices ,

1, 2, 94 Investment behaviour , 40, 50–51 Irrigation sectors , 4, 41, 73, 80, 81, 93, 122,

128, 132, 154, 164

J Japan , 33 Jiangdu City , 172

K Keynesian theory , 125

L La Nina , 160 Leontief assumption , 21, 87 Linear expenditure system , 25 Linearization errors , 45 Linear programming models , 121 Livestock , 3–5, 71, 81, 82, 84, 86, 88,

89, 94, 97, 113, 119, 122, 124, 129–131, 134

Local commodities , 15 Lu Jiagua , 173

185Index

M Macroeconomic impacts , 114 Marginal prices , 149, 150 Margin goods , 20, 24 Mary River valley , 153, 161 MDB. See Murray-Darling basin (MDB) MDBA. See Murray Darling Basin Authority

(MDBA) Method of construction , 27, 33 Mining boom , 2, 3, 5 Missing data , 27 MMRF. See Monash multi-regional

forecasting model (MMRF) Modelling

comparative static , 37, 38, 40, 42–45, 48, 176

dynamic , 5, 14, 37–39, 43–50, 164, 176

LUTE , 15 partial equilibrium , 4, 80, 163 quarterly , 3, 45–48

Mollenkopf, Tom , 160 Monash multi-regional forecasting model

(MMRF) , 15–17, 32, 34 Morocco , 72, 73 MSM-BIGMOD , 65 Multipliers , 4, 26, 80, 112, 129 Multi-regional , 1, 4, 8, 9, 12, 14, 15,

27, 53, 74, 75, 81, 92, 121, 166, 169

Murray-Darling basin (MDB) , 5–8, 41, 43, 51, 80, 94, 96, 100, 113, 120, 121, 130, 137, 139, 144–146, 163, 166, 168, 169, 176, 177

Murray Darling Basin Authority (MDBA) , 52, 53, 68, 100, 114, 121, 127, 163, 170

Murray-Darling Basin Ministerial Council , 60

Murrumbidgee irrigation area , 61, 106

N National Bureau of Statistics , 8 Net foreign liabilities (NFLs) ,

40, 42, 44, 48 Nile River , 165 Ningxia , 173, 174 Non-price rationing , 147, 156 North Africa , 73 NSW Irrigators Council , 113

O ORANI model , 4, 14

P Partial equilibrium models , 4, 163 Pearson, Ken , 14, 53 Perennial cropping , 69, 81, 97, 122 Pipeline , 7, 144, 148, 149, 159, 168, 169 Poland , 33, 169 Policy analysis , 1–9, 37, 45, 94, 163 Port Hedland , 2 Price rationing , 149, 153 Pricewaterhouse Coopers , 43 PRIDE. See Program for Regional Demand

Estimation (PRIDE) Productivity losses , 38, 129, 176 Program for Regional Demand Estimation

(PRIDE) , 65 Property rights , 60–62, 66 Purchasers’ values , 17, 22

Q Qinghai , 173 Queensland water commission , 150, 159

R RAS procedure , 31, 94 REALM , 60, 65 Referendum , 7, 158 Regional labour market adjustment , 40 Regional shares , 27 Regional sourcing mechanism , 19 Regression, water prices , 107–110, 137–138 Rental prices , 89, 108–111 Resource constraints , 2, 4 Rice , 4, 5, 31, 67, 82, 93, 94, 97, 105,

107–109, 113, 115, 122, 124, 130, 131, 136, 168, 175

Risk aversion , 146 Rocklands Reservoir , 169 RunDynam software , 52–54

S Salinity , 62–64, 67–71 SAM. See Social Accounting Matrix (SAM) Samson Brook Dam , 144 San Joaquim Valley , 71 Satellite matrix , 20 Satellite water accounts , 151, 153, 167 Sectors

dry-land , 128 irrigation , 4, 41, 73, 80, 81, 93, 122, 128,

132, 154, 164 Sets, parameters , 48–50, 128 Shanxi , 173

186 Index

Sichuan , 173 Sichuan Academy of Social Sciences , 173 SinoTERM , 8 SMDB. See Southern Murray-Darling Basin

(SMDB) Social Accounting Matrix (SAM) , 20 South Africa , 9, 33, 170 South East of South Australia , 96 South-east Queensland , 143–161 Southern Murray-darling Basin (SMDB) ,

99–117, 119–140 South-to-North Water Diversion Project ,

172–173 Specifi c capital , 82, 86–87, 106 State-contingent theory , 69 Statistical divisions , 16, 31, 32, 143, 154 Sticky rental adjustment , 125 Sugarloaf Dam , 168 Supply shifts , 121, 123, 124, 167

T TABLO computer language , 49, 53, 54 Technologies , 1, 2, 27, 29, 30, 38, 94, 97, 121,

122, 153, 164 TERM , 1–9, 13–34, 37–54, 79, 121, 153, 163 TERM-DYN , 6, 39–43, 45–47, 151, 154, 159 TERM-H2O , 4, 6, 8, 39–43, 45, 51, 52, 54, 74,

75, 79–97, 100, 103, 104, 106–112, 120–123, 125, 126, 128–132, 134, 135, 137–139, 151, 163–177

Terms-of-trade effects , 126 Three Gorges Dam , 173 Tianjin , 8, 172–174, 176 Tillegra Dam , 167 Toowoomba , 7, 95, 148, 158 Top-down , 3, 14, 15, 54, 72, 164, 165 Trade matrices , 1, 2, 30, 31, 94 Traveston Dam , 146, 147, 153–157, 159,

161, 167 Truss, Warren , 113

U Urban water

crisis , 145 managers , 149, 150, 153, 160

USA , 1, 16, 21, 33 Use-supply table , 14

V Vegetables , 17, 21, 22, 24, 72, 92, 94, 96,

105–108, 111, 113, 122, 130, 168

Victorian Farmers Federation , 115 VIEWHAR software , 53, 54 Virtual water , 73, 168

W Walker River Basin , 71 Water

allocation , 5–7, 38, 41, 61–65, 67–69, 73, 97, 99, 101, 116, 120, 124, 126–129, 131–133, 135–138, 166, 171

asset value , 5, 137, 164 availability , 4–6, 38, 61, 65, 66, 69–71, 73, 74,

80, 84, 97, 101, 114, 122, 126, 128, 137, 138, 144, 147, 149, 152, 165, 175, 176

buybacks , 5, 7, 38, 41, 43, 80, 121, 136, 139, 163, 164

effi ciency , 146, 150, 174 elasticity of demand , 102, 103, 124, 149 infrastructure , 7, 150–157 management , 15, 146–147, 150, 159,

174, 177 price , 6, 7, 43, 61, 73, 91, 92, 95, 103, 107,

110, 113, 116, 136–138, 146, 147, 150, 163, 168

supply augmentation , 146, 149–151, 157, 159, 160

supply insurance , 147 supply variability , 61 sustainable diversion limits (SDLs) , 100 trading , 4, 8, 51, 72, 73, 80, 91, 92, 97,

122, 128, 129, 134, 146, 165, 168, 169 uncertainty , 61, 63, 69, 171 Water Act 2007 , 7 Water Corporation, Western Australia , 158

Water accounts, UN survey , 170 Welfare , 38, 41, 42, 66, 67, 69, 70, 73, 152,

153, 157, 158, 164, 167 Wen Jiabao , 173 Western Corridor Water Recycling Project ,

148, 158 Wokalup Creek Dam , 144 Wonthaggi , 7, 144, 147 Wyaralong Dam , 148, 158

Y Yanco irrigation area , 61 Yangtze River , 172, 173 Yellow River , 172, 173

Z Zero trade fl ows , 31

Date post:	16-Dec-2016
Category:	Documents
Upload:	glyn
View:	230 times
Download:	0 times

[Global Issues in Water Policy] Economic Modeling of Water Volume 3 ||

Documents