Modern Input-Output Models as Simulation Tools for Policy-Making
Ciaschini, M.
IIASA Collaborative PaperSeptember 1982
Ciaschini, M. (1982) Modern Input-Output Models as Simulation Tools for Policy-Making. IIASA Collaborative
Paper. Copyright © September 1982 by the author(s). http://pure.iiasa.ac.at/2062/ All rights reserved.
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted
without fee provided that copies are not made or distributed for profit or commercial advantage. All copies
must bear this notice and the full citation on the first page. For other purposes, to republish, to post on servers or
to redistribute to lists, permission must be sought by contacting [email protected]
NOT FOR QUOTATION WITHOUT PERMISS I O N OF THE AUTHOR
MODERN INPUT-OUTPUT MODELS AS SIMULATION TOOLS FOR POLICY-MAKING
* M a u r i z i o ~ i a s c h i n i
S e p t e m b e r 1 9 8 2 CP-82-56
* D e p a r t m e n t o f E c o n o m i c s U n i v e r s i t y o f u r b i n o P i a z z a B. S t r a c c a Ancona I t a l y
- - . . - d,, * - - z . 2 ; r ~ ~ c ; < : ~ a -,-, 1.9 r e p o r t work w h i c h h a s n o t b e e n
p e r f o r m e d s o l e l y a t t h e I n t e r n a t i o n a l I n s t i t u t e f o r A p p l i e d S y s t e n s A n a l y s i s a n d w h i c h h a s r e c e i v e d o n l y l i m i t e d rev iew. V i e w s or o p i n i o n s e x F r e s s e d h e r e i n d o n o t n e c e s s a r i l y r e p r e s e n t t h o s e o f t h e I n s t i t u t e , i t s S a t i o n a l Nember O r g a n i z a t i o n s , o r o t h e r o r g a n i - z a t i o n s s u p p o r t i n g t h e w o r k .
INTERNATIOSAL INSTITCTE FOR APPLIED SYSTEXS ANALYSIS 4 - 2 3 6 1 L a x e n b u r g , A u s t r l a
PREFACE
Current policy issues require economic models to play the role of rational decision schemes. The problems we face today are more complex than those of the past and the progressive fragmentation of the policy maker's role makes it increasingly necessary to have a coherent scheme forecasting and simulating alternative types of behavior.
There has been a tradition of 'macro' model-building in which the demand side is privileged. However recent events have focused interest on economic variables defined at a more detailed level, and have emphasized the need for policies to be specified at a greater level of disaggregation but consis- tent at the macro level. In fact macro models provided infor- mation on each final demand component but do not describe the structure of each variable. However the sectoral composition of these components is often crucial jn indicating the pattern of either technological or behavioral structural change in the eco- nomy.
The study to be presented is part of the research work by the INTIMO group to build a modern 1/0 model of the INFORUM type for Italy. Some results obtained in the estimation of the investment function and In the simulation of the real side of the model are presented.
Work on input-output modeling at IIASA begun in 1979 with Clopper Almon and Douglas Nynus. During this period there has been considerable progress in the construction, linking and use of input-output models. With substantial help from IIASA and the Inter-Industry Forecasting project (INFORUM) at the
Un ive rs i t y o f Maryland, a s e l f o rgan iz ing network o f co l labo- r a t i n g i n s t i t u t i o n s has been b u i l t up t o work on t h e develop- ment and l i nkage o f inpu t -ou tpu t models. A Task Force Meet- i n g is h e l d each y e a r a t IIASA t o draw t o g e t h e r t h e r e s u l t s ob ta ined by t h e c o l l a b o r a t i n g groups and t o d i s c u s s f u t u r e research .
Mauriz io G r a s s i n i REGIONAL DEVELOPMENT
GROUP and SYSTEM AND DECISION
SCIENCES AREA
Laxenburg, September 1982
ACKNOWLEDGEMENTS
The results presented in this paper constitute a part of the research effort of the INTIMO (Italian interindustrial model) Group led by Professor M. Grassini. The author is grateful to Professor D.E. Nyhus, University of Maryland, for his assistance with the simulation and to Ing. A. Alessandroni, National Energy Agency (ENI), Rome, for specifying the energy scenario. penelope Beck, Olivia Carydias and Judy Pakes are thanked for coeditorial, organizational and secretarial assistance.
CONTENTS
1 . Introduction
2. Macroeconomic and Input-Output Models
3. Integration of Demand: the Role of Investment Functions
4. Exogenous Information
5. Information Produced by the Model
6. Conclusions
APPENDIX
REFERENCES
- vii -
MODERN INPUT-OUTPUT MODELS AS SIMULATION TOOLS FOR POLICY MAKING
Maurizio Ciaschini
INTRODUCTION
Current policy issues require economic models to play the
role of national decision schemes (Caffe 1977, Rey 1965). Since
the problems we face today are more complex and the policymaker's
role more fragmented than formerly, it has become increasingly
necessary to have a coherent scheme for forecasting and simulating
alternative types of economic behavior. This naturally implies
that the methodological principles underlying economic model
building should be carefully examined. Many of the fundamental
dichotomies assumed in the past for the sake of simplification
appear to be inappropriate for present-day policy problems.
The main distinction between stabilization and growth models
is in their statistical and mathematical basis, from which it is
easy to find a unique mathematical generating trends and a unique
statistical cause generating fluctuations. However, when consi-
ering these models from an economic viewpoint, it is more diffi-
cult to find a unique cause generating trends and fluctuations
(Hicks 1965). Such a distinction, be it explicit or implicit,
is based on the idea that stabilization problems should be dealt
with by short run demand-oriented models and growth problems by
medium run supply-oriented models (Fox et. a1 1973). The ulti-
mate implication of such a methodological approach is to neglect
the interaction between stabilization and growth aspects, omit-
ting a consistency criterion coordinating short- and medium-term
policies.
There has been a tradition of 'macro' model building in
which the demand side is privileged. However, recent events have
focused interest on economic variables defined in more detail
and have emphasized the need for policies to be specified at
a greater level of disaggregation but consistent with the
macro level. Macro models provide information on each final
demand component, such as imports, exports, and domestic con-
sumption but do not describe the structure of each variable.
Yet the sectoral composition of these components is often cru-
cial in indicating the pattern of either technological or behav-
ioral change in the economy.
This issue seems to reveal the indadequacy of the concept
of the macro-variable (Pasinetti 1 9 7 5 ) . The internal dynamics
of such variables seem to compromise not only the very concept
of macro-variables but also their macro inter-relations (Spaventa
and Pasinetti 1970) . Nor is the solution to be found in disag-
gregating macro models in a nonsystematic way, such as by intro-
ducing additional sectoral .equations or splitting the macro re-
sults by means of a given set of weights.
To deal with these and other issues Almon ( 1982 ) proposes
that modern input-output models be used as rational decision
schemes for economic policy making. This implies changing the
way of looking at the economic process. Although it does not
mean that macro aspects of the economy should be ignored, they
are no longer considered central to the explanation of the indi-
vidual's economic behavior. Rather they are the result of an
aggregation of the behavior that has been defined and simulated
at a more detailed level, for example the level of the input-
output sector for total output and intermediate demand, the items
of expenditure from household budgets for final consumption, and
the appropriate disaggrecration for each particular item for the
remaining items of final demand.
Such a framework can be used to address a set of issues
that are currently relevant to policymaking. In the past these
issues were not tackled satisfactorily for a number of reasons.
F i r s t , a g r e a t p a r t of i n t e r e s t was devoted t o t h e aggregate
c o n t r o l of expend i tu re and t axa t i on . Second, t h e r e was a l ack
o f f l e x i b l e computing programs f o r es t ima t i ng s e c t o r a l behav io ra l
equa t ions and f o r ope ra t i ng m u l t i s e c t o r a l s imu la t i on models.
F i n a l l y , t h e o r e t i c a l advantages were n o t s o developed t o t a c k l e
conven ien t ly t h e i n t e g r a t i o n o f t h e input -ou tput s i d e w i th t h e
demand s i d e .
2 . MACROECONOMIC and INPUT-OUTPUT MODELS
Steady p rog ress i n economic modeling has been s t imu la ted
by t h e i n c r e a s i n g complexi ty o f economic problems. For some
t i m e a n a l y t i c a l t o o l s have been developed independent ly i n two
methodological frameworks: input-output models and macroeconomic
models.
T r a d i t i o n a l input -ou tput models have in f l uenced t h e f i e l d
of app l i ed modeling i n two ways. F i r s t , they have s t r e s s e d t h e
need t o r e f e r t o t he economic system by means of d e t a i l e d ca te -
g o r i e s . For such a purpose t he producing s e c t o r i s de f ined a s a
component o f t h e system having a homogeneous ou tpu t f o r a g iven
technology. Second, they made i t c l e a r t h a t p roduct ion must
s a t i s f y n o t on ly f i n a l demand b u t a l s o in te rmed ia te demand,
which can be i d e n t i f i e d when t h e t e c h n i c a l c o e f f i c i e n t s (such a s
those i n d i c a t i n g t h e in te rmed ia te demand f o r t h e ou tpu t of a cer- t a i n s e c t o r ) have been de f ined . The main c o n t r i b u t i o n of t r a -
d i t i o n a l input -ou tput models i s t h a t they a l l ow t h e l is t of
f i n a l demands t o be t ransformed i n t o a v e c t o r of s e c t o r a l out -
pu t s .
Given a vec to r x r ep resen t i ng n ou tpu t s , a vec to r f repre-
sen t i ng t he l i s t m o f f i n a l demands and a ( n x n ) ma t r i x A o f
t echn i ca l c o e f f i c i e n t s , t h e problem of t he supply/demand e q u i l i -
brium is so lved by f i nd i ng a va lue of vec to r x such t h a t t h e f o l -
lowing r e l a t i o n i s f u l f i l l e d :
the coefficients aij were traditionally considered as constants.
Less importance has been devoted to the vector f of final
demand. It represents the total final demand for the specific
good produced by each sector. Thus, the disaggregation of the
final demand components, in general, does not allow their behav-
ioral functions to be adequately specified.
Conversely, macro models have completely ignored the inter-
industrial aspects since they emphasize Gross Domestic Product
only. Nevertheless, they were able to specify the behavirol
functions for each demand component with great accuracy.
The supply-demand equilibrium macro-relation is represented
by :
where Y represents GDP (Siesto 1977), C(a) is the consumption
function, I(.) is the investment function, G(a) is public expen-
diture, and X(a) and M(*) are exports and imports, respectively.
Each final demand component is explained by a set of variables,
denoted by (a), among which Y may also appear. The only point
of intersection between the two schemes is:
The points of contact between the two approaches have steadily
increased and in particular input-output models have begun to ex-
plain the final demand formation process without compromising
on the multisectoral approach.
The Interindustrial Italian Model--1NTIMO (Ciaschini and Gras-
sini 1981)--is a modern input-output model of the INFORUM family
(Almon 1974 and 1981, Young and Almon 1978, Nyhus 1981). The final
demand components are explained by behavioral equations econo-
metrically estimated. Each final demand component is explained
at a level of disaggregation which allows for a correct speci-
fication of the sectoral demand functions. The disaggregated
consumption vector is composed of nl expenditure items accord-
ing to the items appearing in the household budget accounts.
In fact, the effects of the consumer's behavior through those
items can be correctly observed. Investments (F2) are explained
in terms of the n investing industries, and so on for the re- 2 maining components of final demand. In this way we obtain:
where F1 is the vector of disaggregated consumption functions,
F2 is the vector of disaggregated investments, and so on up to
the kth component of final demand.
The multisectoral supply-demand relation is to be fulfilled
at the input-output level. We therefore need to transform con-
sistently the F1,...,Fk demand vector and to do s.0 we make use
of bridge matrices Bl (t) - Bk (t) such that
The B m a t r i c e s e x p r e s s t h e cons i s tency between t h e i npu t -ou tpu t
accounts and t h e f i n a l demand accounts . I n t h i s model t h e e q u i l i -
br ium r e l a t i o n analogous t o ( 1 ) and ( 2 ) i s g iven by :
where
Equat ion ( 7 ) shows t h e s i m u l t a n e i t y i n t h e s imu la t i on o f
t h e model. The B, ,..., Bk br idge m a t r i c e s a l l o w t h e pu rchas ing s e c t o r s t o be connected t o t h e produc ing s e c t o r s . The supp ly
demand equa t i on i s so l ved a t t h e inpu t -ou tpu t l e v e l . Th i s means
t h a t w e can o b t a i n t h e s o l u t i o n f o r f i n a l demand accord ing t o t h e
purchas ing s e c t o r s and t o t h e inpu t -ou tpu t s e c t o r s . While t h e
f i r s t r e s u l t a l lows a change i n t h e demand s t r u c t u r e t o be ana-
l yzed e f f e c t i v e l y , t h e o t h e r p rov ides i n fo rma t ion on t h e d e s t i -
n a t i o n of o u t p u t a t t h e inpu t -ou tpu t l e v e l .
C 1 , . . . , C k m a t r i c e s r e p r e s e n t t h e pa ramet r i c s t r u c t u r e ,
economet r i ca l l y e s t i m a t e d , of t h e s imul taneous r e l a t i o n s h i p be-
tween t h e f i n a l demand v e c t o r s and s e c t o r a l o u t p u t . Equat ion
( 8 ) shows t h e lagged e f f e c t and equa t i on ( 9 ) t h e exogenous v a r i -
a b l e e f f e c t .
Such is the logical scheme that connects matrices and vari-
ables within the model. We now give a detailed example of how
demand'functions are introduced in the input-output structure,
of the type of a p r i o r i information that can be provided for the
model, and of the type of result that can be expected.
3. THE INTEGRATION OF DEMAND: THE ROLE OF INVESTMENT FUNCTIONS
The integration of interindustrial and demand aspects,
achieved by means of equation ( 7 ) , enables us to construct a
flow table between the intermediate and final sectors that is much
richer in information than traditional flow tables (Ciaschini 1982;
M. Grassini 1932; and L. Grassini 1 9 8 1 ) . Table 1 presents the
flow table for the INTIMO model.
Table 1. The flow table for the INTIMO model.
A
INTERME - DIATE
FLOWS
B
CONSUMP-
T I ON
E
PRIVATE
AND
PUBLIC
EXPEN-
DITURES
C
INVEST-
MENT
Each row of the table referes to a product of the input-output
list and each column refers to a purchasing sector. Such sec- tors, summing to 114, are specified as in Table 2.
D
I N v E N T 0 R
Table 2. The flow table for the INTIMO model: purchasing sectors.
MATRIX PURCHASING SECTOR CONTENT
A 44 Intermediate demands
40 Expenditure items in house- hold budgets
2 3 Investment by investing sector
D 1 Inventory change
Public administration and private social institution expenditures:
1. Health
2. Education
3 . Other public expenditures
4. Private institutions
F 1 Imports
G 1 Exports
Table 2 shows the type of item for which the INTIMO model
produces information for each year along the time horizon. The
computational algorithm constructs such tables by solving equa-
tion (7) iteratively. A given output vector for the input-output
sectors is transformed into a vector of total output consistent
with final demand equations x. With such a vector and with a
vector of exogenous variables y, the set of final demand vectors
Fir i = 1, ..., k is determined. These demands are transformed
into the input-output demand vectors f l , ..., fk. Then, using
the technical coefficients, we can determine the new vector of
total output GIO. If significant differences are found between
the vectors xIo and XIo, the procdure is repeated. Within such
a loop there exists a further loop that determines the total
output vector given the final demand vector.
Intermediate and final demand can be determined simultane-
ously on the basis of total output because some final demand equa-
tions, such as the investment equations, show total output among
their arguments. The logical scheme of such a process is shown
in Figure 1. The sectoral investment function used is the follow-
ing. Total investment I is given by expansion investment V and
substitution investment S, so that
Substitution investment is given by a replacement rate that is
r times the capital stock K .
where the capital stock K is determined as the capital-output
ratio k times the smooth output 5:
Figure 1 . Scheme of the simulation procedure.
- FINAL DEMANDS
E
-
0 u i BRIDGE Fi V E M X T - D E ~ N D S _ - . U E E N P P P
- u
R U S S
- 7
- X AGGR - X - -
..
INTERMEDIATE
Expansion investment is equal to the capital-output ratio k times
a distributed lag on changes in output:
where
The sectoral investment function is then given by:
At this stage the capital cost is not considered within the argu-
ments of the sectoral investment functions. Even if such an
element were to be taken into consideration, we do not have avail-
able reliable sectoral data on such a variable. Given the limited
length of the variable series, the hypothesis of equality between
the marginal and the average capital-output ratio was preferred
to a more elaborate one.
In the estimation
and
where w2 is not estimated but calculated according to:
The sectoral investment function estimated for 23 investing indus-
tries is then given by:
The statistical data base for the regression is given by:
1. Investment by producing and investing sectors
for 23 investing sectors from 1970 to 1979 in
constant and current prices (ISTAT 1970 - 1980).
2. Total output for 44 input-output sectors from
1966 to ,1979 determined on the basis of the
industrial production index and services!
value only.
The relation (18) was imposed on available data, assuming a
replacement rate of 10 percent and a distributed lag of the third
and fourth order. Selected plots of the regression are shown in
Tables A l -A4 of the Appendix. The results obtained are summa-
rized in Tables 3 and 4. The estimation was performed earlier
(Ciaschini 1981), but has been repeated since better information
on total output prior to 1970 for the industrial sectors is now
available. The goodness-of-fit, in terms of the average absolute percentage error (AAPE), is slightly better in the 4 period lag
estimation. However, in such a case the percentage of negative w i is higher. Some sectoral functions show a reasonable fit. For
one third of the sector the AAPE is less than 10 percent, in the
second third it is between 10 and 20 percent and in the final third
it is greater than 20 percent. All the capital-output ratios show
a standard error that makes the estimation look reasonable on sta-
tistical grounds, but in at least one third of the wi estimations
the capital-output ratio seems to be too low.
Additional estimations were performed allowing the value
of r to vary parametrically. The results relating to the good- ness-of-fit in terms of the AAPE are shown in Table 5.
The 23 sectoral investment equations are an example of how
a final demand component was introduced consistently in an input-
output scheme. For the remaining items of final demand see
Ciaschini and Grassini (1 982) and Alessandroni (1 981 ) . 4. EXOGENOUS INFORMATION
Having introduced the final demand components into the input-
output structure (Almon 1979, Nyhus and Almon 1977), we need to
define how the model deals with external information.
From a system's viewpoint external inputs may affect the -- exogenous variables,
-- endogenous variables,
-- parametric structure of the model.
n t r m Q O Y I ~ r e n e m N r L n n O n e o N r ~ ~ ~ ~ ~ ~ u o o ~ n ~ m m a o ~ m ~ m u e N N O ~ ~ ~ O O r t O O O O O O O O r r t O U . . . . . . . . . . . . . . . . . . . . 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 O N
I I a I I I I 7 1
P - N W O C ~ J ~ N C ~ ~ C N C C O O N C O O N n w o t r e Q n ~ 0 . 0 . - C Q m ~ C 0 - t o m . . . . . . . . . . . . O . . . . . . . . . . N U ~ ~ O N O N O ~ ~ O ~ ~ ~ ~ ~ ~ U O O ~ N
.-I d -4 k 0 c, 0 a,
m N N
0
o 0
Y ) N
L U O
W e n r 3 C , N ~ c ~ O Q m Q r . o O ~ ~ O ~ ~ m n e o u r O n n r o , u o ~ n . o ~ r ~ ~ Q m O r N N N r Q Y I r r - b Q m r O e U N r c m m . . . . . . . . . . . . . . . . . . . . . . . O U ~ ~ ~ . ~ N N ~ N O U N Q ~ - ~ Q ~ Q ~ Q *
n r ~ r r ~ r r r ~ N C r c
~ r c u m e n n Q ~ . o . o m ~ ~ ~ r e . o r - . o b o e u m r - n n - ~ m e a r - m ~ u n e r - e n r - m m o O n m e N - r - m N r 0 0 e ~ e u n . o n r m . o . . . . . . . . . . . . . . . . . . . . . . . b n U s . o ~ ~ ~ ~ o r n u ~ ~ b a . o n Q b ~ *
W r N r r N r r r r N r r c
- - . o r - ~ r r m e r - m m ~ n u r - m o o . o n ~ m Q r t N r b * m m & n n r - N r U Q O r r U n r - n r n o m r - e u e o u o m o ~ ~ w ~ o . . . . . . . . . . . . . . . . . . . . . . . . r - u m u e . o n n ~ t e m ~ n e r - - . o n a r c m e
n r r u r r ~ ~ r r N r c c
Q C + N ~ O ~ N ~ ~ N U ~ O ~ Q U ~ U Q U r - n o m u u o m ~ ~ u ~ ~ n e u n c v r U W Q r c
Q s N u m r r . o m N r - e N N r m n r - ~ m m b r c . . . . . . . . . . . . . . . . . . . . . . . r - u n u e . o u n n ~ e m m n o ~ o . o n o r c r c e
n r N r r N r r c N N c N C
m ~ . o n n ~ n ~ n n e u o . o r ~ r c r - n o . ~ U ~ O U O ~ O ~ ~ N U O ~ ~ O N ~ N @ ~ ~ m r ~ w f ( U ~ ~ m . o ( U m m r r c u r ~ r r ~ r V ~ o N m . . . . . . . . . . . . . . . . . . . . . . . Q m N u ~ . o ~ u u . o n ~ e ~ r - u ~ r c r c n
~ C N ~ C N ~ ~ c N N c N C
m n ~ ~ n m r c n u u ~ r - . o r - r n ~ e r - o m u ~ . o u c , u c m r c n w n . o o m n e o o o e n U ~ - U O N N ~ ~ r n n f . ? 9 9 . . . 4 ' : 9 S ? O ~ N ~ O ~ Q ~ Q O ~ ~ ~ ~ - ~ N O N O . ~ U ~ Q O
~ C W C C N ~ C c N N C C N r c
k m
r -
k tn 0 Q) kc, k d W k
Q) C
o
0 m c
o
m N c
~ 1 b 0 r c Q O ) O b O r u ~ u n r v r a N r o . o ~ v Q J U O C ~ ~ ~ " ~ ~ W o ~ U ~ O O N O ~ O U ~ C ~ f i ~
m m o ~ n m r U m a n o n N o ~ n . o r - O m u r n N . . . . . . . . . . . . . . . . . . . . . . . c, g m w o ~ C ~ ~ ~ N N N N C N C N N C C C *? r r
l u N
n k v . 4 a
3 a, F S d c,
3 4 rn
C a 0 S -4 B 0
rn
a, .-I
B
c,c, C d r o a, -4 O O k ~
h a a d
0
n
C.
lj
VI u Cl I U Y I m
0 z C+ t- m
> Z .+
m . r n a a m N r c ( ~ u a N n r n ~ o r ~ r ~ ~ o r v . . . . . . . . . . . . . . . . . . . . . . . r U V N O O O U C I O P C ~ O O ~ ~ ~ V ~ U ~ U ~ U F ~ ~ r r ~ ~ n r u ~ r . l ~ r r r u r ~ r ~ ~ ~ ~ r ~ . r r r
H C U
Y1 3 L r a, 0 C m L L o r e o P, B 0 L L C U 0 c Y 0, . 4 u C "l > YI UI -a c +* m m +' a L . d 3 0 r ( d C C L I . C U (Y C O L * I I G L O PI 0 lJ "l 0 C "l
L P C + ' U O U l E C C U B L L L C > U u a w d 4 L 3 L 0 O . Y E E " l O . d m C 2 r, " l 3 " l C c c c v O J O C C Y I C c c u ) a b
4 0 . U L Y U O ~ - L G r l C V1 O C U P l @J C L L L G P l r J 5 d + . d C O 0 . 4 4 . d G L o P O L L - ~ u c a 1 . 4 L c > 3 C d C P 1 0 C O 4 3 C C G U L C C E.. " H 4 C U L > " l L * * r u -1 L U O 0 - V ~ C ~ C . . ~ O P I ~ C I 3 0 . 4 1 V1 C ~ r 3 m u u a t ~ a n ~ : ~ ~ w ~ o a c c U U O E r l G U V C Y I 4 L C J C O d O > r t L L ~ L L E E ~ 4 ~ d U C V C B u " l u ~ C E P l B L P ~ L C B C L - P I ~ O X P ~ O C C ~ E C C S C ~ O C O ~ ( L O P I ~ O O L O L O ~ ~ ~ ~ B ~ C U C ~ ~ O ) C ~ C P L ~ U C C C U L \ O ~
) r t U Y I O & ~ b 3 r N . t r n O P - 4 ) P . , 3 r . C J l i l C C C . - . - . - . - . - C . - N - J N N
With respect to the exogenous variables, this consists mainly in
defining the trajectories of a set of exogenously determined
variables either as being under the control of a decision maker
or as being outside the set of variables the model can influence.
In this sense they constitute the traditional exogenous variables,
i.e. instruments and data, of the policy problem (Tinbergen 1 9 5 2 ) .
The effects on the endogenous variables consist in the pos-
sibility of substituting the simulated values with observations.
This turns out to be particularly useful for the forecasting
period that starts with the base year, i.e., the year in which
the forecast begins. As we approach the current year fron the
base year, the available statistical data become gradually less
numerous. Thus, the statistical data covering the current year
is incomplete.
For all such periods of the forecasting horizon, the model
takes the observed values and simulates those values for which
there are no statistical data. Only the total output vector
cannot be imposed on the observed values but should be simulated.
Thus, the initial values for the endogenous variables are always
the most recent ones. If there are no data available for a par-
ticular variable, it is simulated according to the most recent
observations on the other variables.
The effect of exogenous information on the parameter struc-
ture allows for a time-change in the technological coefficients
and bridge-matrices, This is possible because of the flexibility
of the computing routines, which enables us to include time-vary-
ing technological coefficients. The trajectories of changesin the
exogenous coefficients can be forecasted and imposed on the model.
For this purpose we can assume that technological coefficient C
varies over time so that the present change is proportional,
together with constant b, to the distance between the actual
value of C and a given constant value a. In algebraic terms:
which admits as a solution the logistic curve
where A is an integration constant.
For estimation purposes, equation (20) can be written as
log (g-1) = log A - bat, if a > 1 , (21 ) Ct - Ct
a a log log (-A) - bat, if - < 1 . t Ct -
Equation (20) is used for coefficients with increasing values,
whereas equation (21) is used for those with declining values.
Unfortunately, we have only one flow matrix for interne-
diate goods. We therefore apply (20) and (21) to a complete
row of the matrix rather than to each coefficient. In this
way we are able to identify the dependepent variable Cit
as an index that shows the volume of intermediate goods pro-
vided by a sector for the whole economy as a percentage of
the total volume of intermediate goods produced by that sector.
where
Such a method of introducing changes in the coefficients is not
exhaustive because the price substutution effect on intermediate
goods is neglected. This effect can be dealt with by means of
Leontief generalized production function ('Dewiert 1971), once
the price formation process has been modeled. Work on this as-
pect of the model, which is underway (Ciaschini 1982), is based
on Belzer (1978).
The effects of the exogenous information considered above
can be classified (Figure 2) in relation to economic policy
according to:
( 1 ) assumptions,
(2) demand controls,
(3) structural hypothesis,
(4) forecasting hypothesis.
@~ornestic P r i c e s I
a ~ x c h a n ~ e R a t e s
@ ~ x p o r t P r i c e s
Q ~ o r l d Demand
E M P L 0 Y M E N T
Figure 2. The impact of exogenous information.
INTERMEDIATE
DEMAND +
1
-
I @ ~ i s p o s a b l e Income
-
0 U
P U T
-
>
FINAL DEMAND
E
N aRIDGE
R U S S
N
T &
The assumptions a r e represented by t h e s e t of v a r i a b l e s
t h a t makes t h e ou tpu t sec t i on of t h e model independent of t h e
p r i c e and income s i d e . I f t h e former opera tes autonomously, we
have t o spec i f y t h e t r ends i n domestic p r i c e s 0, and i n d ispos-
ab le income a. We have a l s o t o f o r e c a s t t he labor f o r ce @. The demand c o n t r o l s mainly r e l a t e t o s imula t ion of t h e
e f f e c t s of d i f f e r e n t pub l i c expend i ture paths @. The dispos-
ab le income t r a j e c t o r y can a l s o be used i n s imula t ing d i f f e r e n t
t rends i n taxa t ion .
The s t r u c t u r a l hypothesis al lows exogenous changes i n t he
elements of t he in termedia te c o e f f i c i e n t s @ and i n t h e br idge
mat r ices @ . t o be taken i n t o account i n t he model.
The f o recas t i ng hypothesis a l lows us t o inc lude i n t h e - model in format ion on
i n t e r n a t i o n a l p r i c e s
0.
t h e
f o r
exchange r a t e a, t h e vec to r of
competing expor ts a, and world demand
A l l t h i s exogenous informat ion enab les us t o formulate a
d e t a i l e d scenar io which forms t h e b a s i s of t he f o r e c a s t . The
r e s u l t s obta ined a r e t hus a funct ion of t h e scenar io t h a t has
been chosen.
5. INFORMATION PRODUCED BY THE MODEL
The exogenous i npu t s a f f e c t t h e macro and s e c t o r a l va r ia -
b l e s i n t h e model. Having def ined a base scenar io t h a t takes
i n t o account t he hypothesis of change i n t h e techno log ica l
s t r u c t u r e of the economic system by means of ( 2 0 ) and ( 2 1 )
and a t r a j e c t o r y of energy demands cons i s ten t wi th t he na t i ona l
energy p lan, we obta ined t h e macro r e s u l t s shown i n able 6 -
This t a b l e p resen ts t h e f o r e c a s t s of t h e macro va r i ab les i n
t h e supply-demand equat ion f o r a 10-year per iod toge ther with
t h e assoc ia ted macro assumptions.
We should s t r e s s t h a t these macro r e s u l t s have been obta ined
us ing a procedure t h a t aggregates t h e s e c t o r a l r e s u l t s . F i r s t ,
t he s e c t o r a l f o r e c a s t s a r e obta ined; they a r e then aggregated
i n t o the rnacrovariables. This process i s dependent on t h e model
~ r e n ~ n r n o U U Y I . . . . . . U U N N N N
U I N ~ I Y I
o o o m m 0009.3 . . . . . N N N r r
r N N O O 00000 . . . . . NNNNN
P - ~ ~ P V I N ooqoo . . . . NNNNN
LIO 0 0 U U W D . . . . O r r r
N I
U U U O O lri . . . . I n v . ~ v \ I
O r :r 0 . N U * < , . . . . N r n, r..
6 U W ' J 0 ) r.1 PI N w - ,- . . . . . l r I P J NNN
+' a E 3 Vb :., . L , - I.'
0 3 " 8 , + u " 8 n .4
m 3 = > C O O L I " O 0 " J > b. ,.'B c -
operating at the sectoral level without the assistance of the
macro part that 'drives' it.
The unemployment variable is the result of the difference
between the labor force forecast, which is exogenously given,
and total employed. Such a variable provides a first check at
the macro level of the consistency of the model inputs.
A selection of plots of the sectoral results that generated
the aggregated information given in Table 6 is given in the Appen-
dix. These results show the sectoral structure of each macro
variable and the path followed within the forecasting horizon.
In particular Tables A5 to A7 in the Appendix show results
obtained for employment in 7 employment sectors and the consump-
tion for 4 items of household budgets in the 'base' scenario.
Sectoral results are also available at the input-output disaggre-
gation level; this is particularly useful when a detailed analy-
sis of a single input-output sector is required. Table A8 shows
the forecasts for total output, imports, exports, and consumption
for input-output sector 4. Tables A9 and A10 show the same items
for input-output sectors 1 1 and 27.
A detailed analysis of the tables produced in one simulation
thus shows the results on three different levels: the macro
level, the purchasing sector, and the selling sector.
The forecasting method using the scenario approach not only
enables us to evaluate the effects of exogenous inputs on the
set of specific trajectories, it also allows us to take full ad-
vantage of the information generated by comparing the results of
various scenario hypotheses. Table 7 presents the macro results
of the 'base' scenario ~ l u s scenarios ALT 1 and ALT 2.
In ALT 1 the energy hypothesis was maintained while assum-
ing a constant technological structure. In ALT 2 the technical
coefficient change was maintained while dropping the energy hypo-
thesis.
Table 7 summarizes the results obtained from the three
scenarios and compares the average growth rates for the aggre-
gated results over the periods 1985-1990 and 1975-1990.
Table 7 . The aggregated r e s u l t s obtained from the base scenario and ALTs 1 and 2 .
P r r v a t e C c n s b n p t ~ o n F a o u i t u f f s r ,oous S e r v l c e s
F i x e d i n v s s t m ? n t I n v o n t o r y C h a n g *
E x p o r t s G o o d s 5 e r v i c e r
I m p c r t s :oous S e r v i c e s
( b a s e ) ( base) ( base) ( h ~ s e ) ( a l t l ) ( s l t 2 ) ( b a s e ) ( a l t l ) ( 7 5 - 31 1 2 92- ?5 5 5 - 00 55- 90 85- 90 75- 90 75- 90 75- 90 ----- ----- ----- ----- ----- ----- ----- ----- -----
? . 4 1 -? .?2 4 . 5 6 2.53 2.50 2-49 2.96 2.96 2.95
D u b l i c a n d 3 r i v a t e 5 o c l ? l ' x p e n d i t u r e 2.17 2.C7 1 . O O 2 .OO 2.00 2.01 2.07 2.07 2.05 A d m i n i s t r a t i o n 2.19 ?.CQ 2.00 2.00 2.00 2.01 2.08 2.08 2.05 E d u c s t ~ c n 2.1: ?.C9 2.03 2.00 2.00 2.01 2.06 2.06 2.05 H ~ a l t h C r r e ?. 2C 2.05 1.97 1.98 1.98 2.01 2 2.07 2.05 j o c i a : F r i v a t c : x p e n e i t u r e s 2.14 Z . C ? 1 . V 4 2.01 2.01 2.01 2.05 2.05 2.05
ASSCl lPTICt iS 0 1 s p o s a b l e i n c o n e ( p e r c a p ) U n e m p l o y m e n t E x c h a n j e Q a t e F o r e i g n Cemano
Note that the average growth rate in consumption and total exports
is the same in all three simulations. This is because in the
model consumption depends on disposable income and relative domes-
tic price trend assumptions, which were kept the same for all the
simulations. The effects of prices on the consumption structure
can be simulated when the price side of the model is complete.
The interaction between output and prices can thus be adequately
taken into account. Exports depend on the forecasting hypothesis
related to the exchange rate, world demand, and the vector of
international prices for competing exports, which were also kept
the same for all three simulations. The average growth rate of
Gross Domestic Product throughout the forecasted period is almost
identical in all three simulations. Such a growth rate is com-
patible with the different unemployment growth rates of the three
scenarios and is explained by the fact that the sectoral struc-
ture of total output appears to be significantly more important
than GDP in determining employment through productivity equa-
tions.
The sectoral results, which are presented in aggregated form
in Table 7, are given in Tables All to A14 of the Appendix. They
show the growth rates for sectoral output, employment, consump-
tion and investment. The complete set of such tables is of par-
ticular interest for policy forecasting since it describes the
growth in the sectoral structure of the most relevant economic
variables. These results can also be used for defining new
scenarios and for verifying the consistency of those already
defined. The tables indicate how the dynamics of the macro
variables sectoral composition affects both their structure and
level. By measuring the time change in the sectoral composition
of the relevant economic variables, we are able to evaluate the
simultaneous effect of changes in the technological and behav-
ioral structure of the given economic system. This is one of
the main issues of present-day policy making.
6. CONCLUSIONS
In recent years economic policy problems have outgrown
the instruments designed to support the policy makers's acti-
vity. Modern input-output models constitute an attempt to
provide schemes for dealing with such problems.
The applications that can be made of the theoretical results
obtained go through two main stages: (i) the integration of the
input-output side of the model with the demand side so that sec-
toral demand equations can be consistently specified in the real
part of the model, and (ii) the formulation of the price side of
the model so that all information on sectoral prices and value
added components can be conveniently exploited.
In this paper some characteristics of a real part of a
modern input-output model for Italy have been described and
some results of the simulations presented. In particular it
has been shown how a simple investment theory was used for
estimating sectoral investment functions and under which assump-
tions the input-output technical coefficients were made to change
according to forecasted patterns.
A price side is also being developed for the Italian eco-
nomy so that the relative price vectors shall be simulated
simultaneously with the remaining endogeneous variables. This
shall improve the effectiveness of the whole scheme; for example,
in the case of the price substitution effect on technical coef-
ficients, the availability of a price side is essential for the
endogenous determination of the coefficient change.
The theoretical and applicative improvements that can be
attained are heavily influenced by the quality and coherence of the statistical data available. An increasing effort is
required to the data sources in order that a greater quantity
of information on input-output data, as well as sectoral demands
be provided in a greater detail and with an higher degree of coherence.
o o r,-8 O r P o w r u m c r D h O r 0 Q r
L O C I L V C , ' , . . .. L O O L i C , L
C - G f ! U C . t 4 U . C Y . P l 0 G C P . I yl L'I < ' . & < U P N N O 4 V I - II
o a . . . . . . . . . . ~ rn < 0 4 I J x, I I" L < a C L U l r C a O m Ouh m u , r . G '"<\I', ru Y I r n I C C C C I . , - V) CI V)
W J Y , > 4 I r E
1 1 1 . 4 L < E r
C' 0 1 C I r I *. .. 0 . . . C + I ~ ~ U C ~ U J U I I ' ~ ~ . ~ . ~ ~ + "- u UCILB 0 11 a C.) a() r- PJC Cr w s u 11 LIP G .. . . . . . . . I O U . ..,, h V) c .rl (., c u (\I 0, r o N C, o .CI ( ' o C, 09 o J - ( 7 r \ ~ I? .o Y, v F r.0 o o (:, II L C " ' a u - P J f ~ I C I n , O h r a, . . C1 01 C f \ , n, P.' P1 C l r.1 r r r N L L- V, C,
01 v O a J C. L - Tu
P C L ' C , C I C , U C I U C ~ U U r V) C I C . ~ ~ C ~ V V C ; ~ O C ~ C , * . . . . . . . . . . . . . . . . . . 3 " .O
C. nr * C, nl * 7 4 0
U 4 u V# .. a , . . . r 3 Y , O ' P S . r ( m c r,
n > F I1 I c Y .4 L u a r( L
L 6 C' C' O . . . L C,oo
r u Y - I VI d . 4 C . I V I V m L a w ~ o - . ~ ~ r ~ r ~ u u ~ a y c a > 2 ~ 6 n c r c r ~ ~ r - r - ~ k r r - r . r - r r- 5 > u v u (nu 0
m n d m o N C O O r n ru r> yc PB O E 0 0
m . . . . J * C , N U A N I I n > I C
C1 4 "I * O O N N .--or * o m , . -
A C < I ' ' L 7 L ' C 1 U C ' U L ) L ' C bl. C,OC,O CIC>C'CJ'JCI . *- J Cl ll . . . . . . . . . . II O 0 , A U L , - 0 m P - r i C l l . 0 1 . 4 0