West Bengal: Fiscal Decentralization to Rural Governments: ANALYSIS AND REFORM OPTIONS MAIN REPORT
February 28, 2007
71483
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Table of Contents
1. INTRODUCTION............................................................................................................................... 1
2. THE SETTING FOR LOCAL GOVERNMENT FINANCE IN WEST BENGAL ...................... 3
LOCAL GOVERNMENT STRUCTURE ............................................................................................................. 3 FISCAL AUTONOMY .................................................................................................................................... 3 STATE FINANCE COMMISSIONS. ................................................................................................................. 5
3. III. RURAL LOCAL GOVERNMENT FINANCES ....................................................................... 8
BUDGET SHARES .......................................................................................................................................... 8 EXPENDITURE STRUCTURE ........................................................................................................................... 9 CONCENTRATION BY POPULATION SIZE ....................................................................................................12 COMPOSITION OF EXPENDITURES BY POPULATION SIZE CLASS .................................................................13
4. REVENUE STRUCTURE ................................................................................................................15
OWN SOURCE REVENUES ..........................................................................................................................15 GRANTS AND TRANSFERS ..........................................................................................................................17 CENTRALLY SPONSORED SCHEMES ...........................................................................................................18 GRANTS AND TRANSFERS: STATE GOVERNMENT ...........................................................................................21
5. ANALYSIS OF THE DISPARITIES IN GRAM PANCHAYAT FINANCES ............................23
EXPENDITURE DISPARITIES .......................................................................................................................24 EXPLAINING EXPENDITURE VARIATIONS ..................................................................................................26 OWN SOURCE REVENUES ............................................................................................................................28 EXPLAINING THE DISTRIBUTION IMPACT OF INTERGOVERNMENTAL TRANSFERS ......................................31
6. FINANCIAL CONDITION ..............................................................................................................34
DO MANY GRAM PANCHAYATS HAVE A NEGATIVE RECURRENT REVENUE GAP? ....................................35 DO THE RESULTS FOR SURPLUS GPS INDICATE AN INABILITY TO ABSORB THE FUNDS? ...........................35 TO WHAT EXTENT DO “DEFICIT” GRAM PANCHAYATS DRAW ON THEIR CASH BALANCE RESERVES TO COVER
THE COST OF DELIVERING SERVICES? .........................................................................................................36 DO THE CLOSING BALANCES CHANGE OVER TIME, AND HOW DO WE EXPLAIN THIS? ARE THESE
CLOSING BALANCES “TOO LARGE”? .........................................................................................................37 WHAT DETERMINES FINANCIAL CONDITION? ...........................................................................................37
7. THE FISCAL POSITION OF BLOCK AND DISTRICT LEVEL GOVERNMENTS ..............40
EXPENDITURES ..........................................................................................................................................40 OWN-SOURCE REVENUES ..........................................................................................................................41 INTERGOVERNMENTAL TRANSFERS ...........................................................................................................42 FINANCIAL CONDITION..............................................................................................................................43 SUMMARY .................................................................................................................................................44
8. STATE GOVERNMENT AND FISCAL DECENTRALIZATION ..............................................45
STATE TAXATION ......................................................................................................................................46 REDIRECTION OF EXISTING FUNDS ............................................................................................................46 THE GRANTING OF MORE FISCAL DISCRETION..........................................................................................47
9. REFORM OPTIONS AND EVALUATION ...................................................................................49
DECIDE ON AN OPTIMAL TIER FOR LOCAL SELF GOVERNMENT ................................................................49 EXPENDITURE ASSIGNMENT ......................................................................................................................53 INCREASED OWN SOURCE REVENUE .........................................................................................................57
INCREASED STATE GRANTS: DETERMINING THE VERTICAL SHARE ...........................................................59 A Formula Approach. ..........................................................................................................................63 Increased, Untied Grants: A Formula Distribution .............................................................................63
10. FINANCING THE FISCAL DECENTRALIZATION PROGRAM ........................................67
REFERENCES ..........................................................................................................................................126
1. INTRODUCTION
1. This report is about the fiscal performance of rural local governments in the state
of West Bengal. Specifically, our goal is to develop a comprehensive fiscal information
system for all rural local governments, and to use these data to evaluate the
intergovernmental finance structure in the state. The work is of significant policy
importance, given the need to develop programs to respond to the constitutional
amendments mandating fiscal decentralization1
, and to support central and state
government initiatives to use the PRI system as an important part of its poverty
alleviation strategy. A more immediate need in West Bengal (and other states as well) is
to support the work of the State Finance Commissions to better integrate rural local
governments into the intergovernmental fiscal framework.
2. Many observers of fiscal federalism in India have pointed out that fiscal
decentralization to the third tier has not progressed very far since the constitutional
amendments of the early 1990’s. State governments have held on to budgetary control
and State Finance Commission recommendations to strengthen local government
finances have in large part been ignored. Underlying this state of affairs is the fact that
neither the central nor state governments have data that allow them to accurately evaluate
and monitor the fiscal performance of the Panchayat Raj Institutions (PRIs).
3. At present, there is no accurate, official record of the fiscal activities of PRI in
West Bengal. The existing data base is not comparable across local government units, is
not readily available in one place or in electronic form, is incomplete, and is not linked to
amenity or demographic databases. Every time the Department of Rural Development in
West Bengal (or the State Finance Commission) carries out an analysis of gram
panchayat finances, it must resort to doing a special survey2. The results presented in
this report, based on primary data for nearly all PRIs in West Bengal, give arguably the
first-ever look at the overall picture of rural local government finances in the state. The
situation as regards socio-economic data is somewhat better in that there is a 2001
census. But even here, census data are missing for a substantial number of gram
panchayats and the so-called “amenity data” have not been aggregated into a usable data
1 Article 40 of the Constitution directs states to endow village panchayats with such powers as necessary to
function as units of self government. This is reaffirmed in the 73rd amendment to the constitution (1992),
which defines the governance of the PRIs, and calls for the assignment of expenditure responsibilities and
taxing powers to them. 2 See, for example, SRD (undated).
2
base3. This is because the base measurement unit of the census is the sub-village level
and aggregation to the gram panchayat level has not been done properly in every case.
4. In this report, we use data from a newly developed fiscal information system to
study the fiscal performance of PRIs in West Bengal. The general outline of the data
base upon which this report is built, and the methodology used in collecting and
organizing these data are included in Annex A.
5. The first three sections of this report describe the structure of rural local
government finance in West Bengal. More specifically, we describe the expenditure
responsibilities and financing powers of local governments, and carry out an empirical
study of how they use these responsibilities and powers. In Section IV, we focus on own
source revenue and grants and transfers to PRIs from the central and state level
governments. In Section V, we study fiscal disparities among gram panchayats in West
Bengal and model an explanation of these variations. The concern in Section VI is with
fiscal balance among the gram panchayats, i.e., with the extent to which they run deficits
on current account and the extent to which they carry significant balances. A parallel
analysis for block and district level governments is the subject in Section VII. In Section
VIII, we examine the pivotal role of the state government in determining whether fiscal
decentralization will take place. In the final two sections of this report, we evaluate some
options for reform and the cost implications of these reform packages. While this work
does profile the fiscal activity of all three levels of PRI, the focus is on gram panchayats.
This is because of the view in West Bengal (and in many other quarters) that gram
panchayats are the best candidate for local self-government.
6. This study can be a significant compliment to the work of the Third State Finance
Commission in West Bengal, which is due to report in 2007. An added benefit from this
study, hopefully, is that it may lift the level of the discussion about PRI finances among
political leaders, technical experts, private sector leaders and voters in the state. It might
also be useful as a model for similar work being carried out in other states.
3 The “amenity” data include measures of community infrastructure -- such as school capacity, miles of
permanent roads, street lighting -- and measures of community development such as the presence of
commercial banks or telecommunication services. For an example of how these data may be used, see the
Karnataka and Kerala case studies in Sethi, ed. (2004).
3
2. THE SETTING FOR LOCAL GOVERNMENT FINANCE
IN WEST BENGAL
7. In terms of budget responsibilities and resources, each of the three levels of local
government within the rural sector is assigned a set of expenditure responsibilities,
revenue raising powers, and entitlements from the grant system. These assignments are
made by the State government under the guidance of the Constitution and the Panchayat
Act. The expenditure responsibilities of rural local governments are spelled out in some
detail in Government of West Bengal (2005). PRI finances also are impacted by the
recommendations of the State Finance Commission concerning the distribution of fiscal
resources between the state government and the local governments4. Finally, there are
direct, conditional transfers from the central government to the PRIs and these constitute
the major revenue flow to the local bodies. The impact of these three influences is
described below.
Local Government Structure
8. The PRI system in West Bengal consists of 18 districts (zilla), 341 blocks
(panchayat samatis) and 3,324 gram panchayats, as described in Figure 1. The urban
stream of local governments, which includes municipalities and city corporations, is
separate from the rural stream. The rural and urban local governments operate under
different enabling legislation, and each is allocated a specified share of the state
government revenue sharing pool.
9. Perhaps as befits a state with more than 80 million people, the structure of rural
local government is a hierarchical one. Local officials report up to the next highest level.
Moreover, the flow of most intergovernmental transfers passes through the district and
block levels before reaching the gram panchayats. Budget approval by PRIs, however,
does not have to be sought from a higher tier of local government.
Fiscal Autonomy
10. The data presented in this report describe the pattern of spending and financing by
rural local governments. But, data alone cannot reveal the degree of autonomy that local
governments have in making fiscal decisions. There are elected local governments in
4 The State Finance Commissions are required to sit once in every five year period.
4
West Bengal and each has the power to approve their budget. In fact, however, the
degree of fiscal discretion that is underneath the making of these budgets is limited by
expenditure mandates and by constraints on revenue-raising powers.
11. Local governments are given no powers to exceed their assigned complement of
employees, nor do they have any power to determine the rate of pay of their employees.
So, the wage bill of local governments is fixed from above.5 In addition, there are
expenditure mandates, e.g., the central schemes are conditional grants with narrowly
prescribed expenditure targets.
12. The gram panchayats (GPs) are thought by some to be more like autonomous
local governments than are districts or blocks. For decades, the idea of village
government with some degree of autonomy has been discussed in India. Gandhi’s vision
of village swaraj has influenced subsequent discussions of the need for local self-
governance (Alok, p. 207). Moreover, there is provision for this in Article 40 of the
Constitution. With the passage of the 73rd
constitutional amendment, local governments
were again recognized, and more explicit provision was made for planning and service
delivery responsibility and for revenue raising powers. In most states, the gram
panchayats have been given more fiscal autonomy than districts and blocks, including
some independent power to levy certain taxes6. However, in West Bengal, this has led to
only about 6 percent of GP expenditures being financed from local sources. On the side
of expenditure composition, GPs play a role in project selection; hence to some extent
they can be responsive to special needs in various locations within the gram panchayat.
Otherwise, most of their expenditure budget is driven by mandates from higher level
governments.
13. The blocks and districts have less fiscal discretion on the revenue side than do the
gram panchayats, even though they sit higher in the local government hierarchy. They
may raise revenues from fees and charges, and have the authority to set the rate for some
of these charges, but they have no taxing power. They rely primarily on grants and
transfers for their general purpose finances (only about 3 percent of their revenues are
raised from own sources). Therefore, the size of the block and district budget is mostly
determined by higher-level governments.
14. On the expenditure side, districts and blocks face the same restraints on their
budget choices as do gram panchayats. The expenditure discretion that they do exercise
is in the capital budget, and it relates mostly to project selection and implementation with
respect to centrally sponsored schemes. With respect to State grants, only a small
proportion is unconditional (untied). Most state grants and transfers in West Bengal are
conditional and give local governments little room to rearrange the priorities laid down
by the state government. Though governed by elected councils, the districts and blocks
appear to function largely as spending agents of the state and central governments.
5 An exception is that local governments may hire employees on a contract or a daily wage basis, without
permission from up-level governments. 6 This “discretion” is perhaps overstated. As is discussed below, the rates and bases of these local taxes are
prescribed by the State government.
5
State Finance Commissions.
15. The State of West Bengal has received reports from two State Finance
Commissions (SFC)7. The first SFC was constituted in 1994 and reported in 1995. The
“period of recommendation” was 1996-2001. The second SFC was constituted in 2000
and made an interim report in 2001, with a period of recommendation of 2001-2006. The
third SFC is now sitting with a report expected in 2007.
16. According to the West Bengal Panchayat Act and The Constitution, the charge of
the SFC is far-reaching. Its report is to contain recommendations:
“…on the principles that should govern the distribution of state revenue between
the state and the Panchayats and the allocation between the Panchayats, at all
levels, of their respective shares of such proceeds. The Commission was also
required to recommend the principle that should govern the distribution of the
revenue resources between the state and the Municipalities and the allocation
between the Municipalities of their respective shares. The Commission would
also determine the taxes, duties, tolls and fees which could be raised by the
Panchayats and the Municipalities. The Commission would also recommend
measures needed to improve the financial position of the Panchayats and the
Municipalities.”
17. In effect, the SFC is empowered to recommend a complete overhaul of the system
of state and local government finance. Though there were numerous recommendations
made by both West Bengal Commissions as regards intergovernmental finance, the focus
has been heavily, if not exclusively, on the revenue side. This is similar to the approach
that has been taken by State Finance Commissions in other states (Subrahmanyam, 2004),
and it is consistent with the charge given to the SFCs in the constitutional amendments.
Still, if the focus of the SFC is to be on ensuring adequate financing, a strongly implied
responsibility is to determine optimal expenditure assignments. Otherwise, how could
adequacy in “revenues” be determined?
18. Arguably, the most important proposal of the first two State Finance
Commissions was for a vertical share of 16 percent of state taxes as a grant entitlement of
local governments. Acceptance of this proposal would have led to a very significant
increase in fiscal decentralization. It was also proposed that these grants be “untied”, i.e.,
that the recipient local governments be given the discretion to spend this money in any
way they choose.
19. This primary recommendation of the Second SFC was not accepted by the state
government8. Instead, they decided on a program where a fixed allocation would be
7 The 73
rd Amendment made provision for the appointment of the State Finance Commissions.
8 The West Bengal State Second Finance Commission submitted their report to the government on
February 6, 2002. The corresponding ‘Action Taken Report’ was filed by the State Government on 15th
July 2005 before the State legislation.
6
made to a PRI revenue sharing pool each year, depending on the financial condition of
the state. In fact, even this approach was not acted on by the State government in 2004
and 2005. In these two years, the State government did not distribute finance
commission grants to the local bodies. In 2006, the vertical share was set at Rs 278 crore,
which is about one-half the amount recommended. The sharing of this pool of funds
among local governments, however, is done by formula, and in this regard, the
recommendations of the SFC have been followed.
20. Other important recommendations of the State Finance Commission were in the
area of own-source revenues of PRIs, where the need for a revenue increase and the need
for more autonomy were seen to be important. The First SFC proposed to devolve
entertainment tax revenues fully to local governments and to give rate setting powers to
the local bodies. The Second SFC called for full devolution of entertainment tax
revenues to local governments, but not for rate-setting powers. The State Government
has chosen to hold the power to set rates but to transfer 90 percent of the revenues raised
to local governments. Both State Finance Commissions argued for devolving the powers
to set for the level of various charges. State government has not moved very far in this
direction.
21. One could make the argument that the West Bengal State Finance Commissions
have not been very successful9. Their major recommendation, that a general revenue
sharing program with a 16 percent vertical share be established for local governments,
was not accepted. There are many reasons for this, with the precarious fiscal position of
the state government being the most common explanation. There were, perhaps, other
shortcomings. A reasonable critique of the principal recommendation is that the 16
percent share was not justified by a hard analysis of the expenditure needs of the local
governments. To date, neither of the SFCs nor the State Government has based their
recommendations for a vertical share on expenditure needs. Oommen (2006) studied the
outputs of five other State Finance Commissions and found that none estimated the
vertical gap based on a hard analysis of expenditure needs. Correcting this problem is
arguably the next important step that the SFCs should take to improve the acceptability of
their recommendations.
22. Another critique of the SFC work in West Bengal (though this problem was out of
their control) was that neither commission had access to a reliable data base of PRI
finances, as has been developed here. The extent to which the absence of fiscal
information constrained the work is well-described by the following quote from
“Recommendations of the Second State Finance Commission” (p. 26):
“Data regarding resources of the different tiers of the Panchayats were not
available at the State level. In view of this we circulated a questionnaire to all the
Panchayats. Responses to this questionnaire were not uniform. Seven hundred
and nineteen (719) out of 3362 gram panchayat sent replies but the data in respect
9 In fact, one could make the argument that very few of the SFCs have had a favorable impact on the
finances of PRIs. A good review of the recommendations and results of the State Finance Commissions is
in Subrahmanyam (2004). Also see Oommen (2006), and Rao and Singh (2006).
7
of only 170 gram panchayat could be used. We also received information from
142 PSs out of a total of 341 while 10 out of 17 ZPs responded to our
questionnaire. In spite of better administrative infrastructure available at the ZP
level, it is difficult to understand why larger number of ZPs could not respond.
We are aware of the inadequacy of the data and the problems of generalization
from the same. While in respect of PSs and ZPs, the data could be taken to be
fairly representative, in respect of gram panchayat this would at best give an
indication of their state of finance.”
23. The combination of this absence of data and very limited staff resources made it
difficult (impossible) for the SFC to fully evaluate the impacts of its proposed reforms.10
Whether a better evaluation of impacts would have led to a different proposal, or to a
different response from the state government, is something that we may only speculate
about. The Third Finance Commission in West Bengal faces this same data constraint.
10
The same problem has been emphasized in a number of other SFC reports. Subrahmanyam, 2004, notes
(section 1.4.3) that “ Absence of district/microlevel data on panchayat finances on which the State-level
data would be based could lead to the conclusion that the statistical particulars furnished by the States to
the Central Finance Commission are nothing short of a kind of ‘guestimates’ ”.
8
3. III. RURAL LOCAL GOVERNMENT FINANCES
24. The data base developed for this project is used here to investigate the fiscal
importance of rural local governments in West Bengal, the structure of their expenditure
outlays, and the sources of their finances. The goal is to develop a fiscal profile for
different levels of rural local government, with special attention paid to the gram
panchayats. Such information can be used by government to track rural local government
finances in the state, and to evaluate reform options. It is an essential piece of evidence
that heretofore has been missing from the policy analysis.
Budget Shares
25. The place to start this investigation of rural government finances is with an
understanding of the relative importance of the PRIs in the state fiscal system. In Table 1
we show the fiscal shares of different levels of government in West Bengal. Note that 72
percent of the state population is resident in rural areas, but less than 17 percent of
government expenditures are made by rural local governments11
. The point to be taken
from this comparison of population concentration and expenditure concentration is not
that rural residents are being discriminated against by the fiscal system (the State
government vertical programs may or may not emphasize rural benefits), but that rural
local governments are quite small players in the state fiscal system, and they are much
smaller players than are urban local governments. In total, expenditures by rural local
governments are equivalent to less than one percent of state GDP. One could safely say
that there is not much local self governance in rural areas of West Bengal.
26. The 17 percent share of expenditures managed through panchayats in West
Bengal is not so far out of line with that in at least some other Indian states where data
are reliable. Rao, et. al., (2004) estimate that about 20 percent of expenditures are
channeled through panchayats in Karnataka, while Oommen, et. al. (2004) estimate the
share at closer to 30 percent in Kerala.
27. West Bengal’s system of government finance is even more centralized on the
financing side of the budget equation: 96 percent of all revenues raised are at the state
government level. Rural local governments account for less than one percent of all own-
source revenues raised. Even this small amount is an overstatement of their importance
in revenue mobilization because local governments have limited discretion to determine
the amount they will raise.
11
All population data in this report are drawn from the 2001 Census.
9
28. Another way to describe the small role of rural local governments in public
financing is to note that local government expenditures are Rs 452 per person for urban
residents as compared to Rs 138 per person for those who live in rural areas.12
Voice in
fiscal decisions at the community level is a benefit that would appear to be more
important for urban voters than for rural voters.
29. We also may ask about the relative importance of the three tiers of PRI in terms
of expenditures and own source revenues. From Table 1, we can see that the district
governments account for about 45 percent of the spending of all PRIs13
. More important
is the finding that the gram panchayats in aggregate account for only about one-third of
spending.14
Thus about two-thirds of PRI expenditures are made by districts and blocks,
who under the present structure, might be better thought of as spending agents of the state
government. Under this interpretation, we may say that of all state and local government
expenditures made in West Bengal in 2005, only about one-third was made by the level
of government that that many would argue comes closest to approximating local self
governance15
.
30. On the side of own source revenues, districts raise approximately the same share
as the gram panchayats (who have access to both tax and non tax sources). The revenue
raised by districts is a mixture of licenses, fees and charges such as fees to oversee
tenders or the work of contractors, and revenues from district-owned enterprises. The
block level raises very little own source revenue (Table 1).
Expenditure Structure
31. How do rural local governments spend their money, and how much variation is
there across local governments? Interestingly, the budgets are dominated by spending for
capital purposes at all three levels (Table 2). District governments report that 85 cents of
every rupee spent is for capital purposes. The capital expenditure share in blocks and in
gram panchayats is closer to 60 percent. In a sense, this is a correct representation of the
expenditure pattern in that these reported capital expenditures are made to purchase or
create assets with a longer life than that of the annual budget.16
But, the percentages
shown in Table 2 are surely an overstatement. Much of this expenditure carries the
objective of job generation hence might be thought of as part of an income maintenance
package and therefore as a current expenditure. Moreover, the assets created by these
public works projects may be more akin to maintenance and repair than to new
12
These estimates of per capita expenditures in 2005 are based on 2005 fiscal data and 2001 population
data, hence are an overestimate. 13
Note that these samples are less than 100 percent coverage because of missing data for some districts,
blocks, and gram panchayats. These missing units are small enough, however, that we do not believe that
their inclusion would markedly change the results shown in Table 1. 14
An interesting comparison is Karnataka where Rao, et. al. (2004), estimate that gram panchayats
accounted for only about 5 percent of total PRI spending in 2003. 15
An interesting comparison is Karnataka where Rao, et. al. (2004), estimate that gram panchayats
accounted for only about 5 percent of total PRI spending in 2003. 16
Capital expenses are defined in the data base as follows: “Any expenditure resulting in the creation of
physical assets like land, buildings, machinery and equipment is regarded as capital expenditure. This
includes centrally sponsored schemes, state sponsored schemes and other schemes/welfare programs which
involve capital expenditure.” (See Annex A.)
10
construction. Also suggesting the temporary nature of some of these projects is the fact
that the use of machinery in the work is prohibited, as is the use of contractors
(Government of India, 2006).
32. Some flavor of this capital expenditure activity in the PRI can be gained from the
guidelines for the SGRY (employment generation) program. The SGRY specifies that
the grant must be spent for the creation of durable assets, particularly those that would
assist in drought-proofing such as soil and moisture conservation works, watershed
development, promotion of traditional water resources, and afforestation, as well as
construction of village infrastructure and link roads, primary school building
dispensaries, veterinary hospitals, and marketing infrastructure (Government of India,
2006). While the assets created under this program may or may not be long-lived, they
do seem to be more in the vein of development than consumption expenditures.
33. So, one can argue that rural local governments emphasize capital spending, but
the numbers reported in Table 2 should be interpreted in the context given above.
34. A second thing to note about the pattern of rural local government expenditures is
that there is specialization among the tiers of rural local governments in terms of
functional assignment. As shown in Table 3, block governments are more involved in
delivering social services through their budgets than are either districts or gram
panchayats. Health, education, and welfare programs account for one-third of block-level
budgets, by comparison to 10 percent or less for districts and gram panchayats. Block
spending in the health area covers a range of responsibilities such as maintenance and
upgrading of community health centers, supervision of primary curative services,
implementation of immunization and safe drinking water programs, and monitoring and
planning (Government of West Bengal, 2005). With respect to education, the blocks
manage the mid-day meal program and with respect to welfare they are responsible for
beneficiary selection for a number of programs. However, block level governments in
West Bengal have been little involved in housing programs. And again, it should be
emphasized that they have little discretion in deciding how to spend these funds.
35. The reported data (Table 3) show districts to spend primarily for housing,
infrastructure, and employment generation (77 percent of total expenditures), a finding
that seems reasonable for the level of government that can best handle service delivery
responsibility when the benefit zone covers multiple blocks and gram panchayats. The
budgetary expenditures of districts appear to be dominated by those programs financed as
centrally and state sponsored schemes. This seems consistent with the idea that district
governments are primarily spending arms of higher level governments.
36. The much smaller gram panchayat’s comparative advantage is in delivering
services and implementing schemes where the benefit zone is local. For the most part,
their expenditure assignments reflect this comparative advantage. Though there is some
responsibility for education, e.g., implementing continuing education programs, the
budgetary expenditures of gram panchayats are dominated by development programs (61
percent of total expenditures) that are mostly financed under the centrally sponsored
11
schemes. Again, there is the issue of whether these are capital expenditures related to the
creation of long-lived assets, or current expenditures to support an income maintenance
program. Arguably the most noteworthy observation to make about gram panchayats is
the 20 percent share of the budget devoted to administration and salaries (Table 4),
compared to less than 3 percent for districts and blocks. This reflects some combination
of the labor intensive nature of the services delivered by gram panchayats, the large
amount of responsibility for management type activities, and apparently a high fixed cost
associated with operating small local governments. This fixed cost effect is described in
Box 1.
Box 1
The Fixed Cost of Rural Local Governments
There is a fixed cost of local government that is independent of population size. All local
governments have in common the need for a secretary, bill collector, support for the local
assembly, a basic level of physical faculties, etc. In fact the State of West Bengal assigns
the same number of posts for each GP, and determines the compensation rate for these
posts. To the extent that all posts are filled, salaries may be viewed as a fixed cost of
local government, and do not vary much with the population size of the local
government.
To test this hypothesis, we have pulled a sample of the 20 largest (population) and 20
smallest GPs. The average population for the 20 largest GPs is over 10 times that of the
20 smallest. The calculations of the average fixed costs of each group are reported in the
table below.
The overall level of administration expenses is higher in the larger GPs, but on a per
capita basis, the fixed cost argument is very apparent. The smaller GPs spend 4 times as
much as the large GPs (per capita) on administration expense. The lower value for the
larger GPs reflects the ability to spread these costs over a larger resident population.
In the case of salary expenditures, the total amounts are about the same, for large and for
small GPs. The per capita amounts in the smaller GPs, however, are about 15 times
higher.
Table Box 1
Fixed Cost Effects: For 20 Largest and Smallest GPs
(in Rs)
20 Largest 20 Smallest
Administration Expenses Per GP 128,103 67,606
Salary Expenses Per GP 347,798 375,377
Administration Expenses Per Capita 6.8 27.7
Salary Expenses Per Capita 10.3 146.0
Source: Source: PRI-West Bengal data base: The World Bank (see Annex A)
12
Concentration by Population Size
37. An important question has to do with the concentration of expenditures among
local governments with different population sizes. In particular, we are interested in the
extent to which the expenditures of rural local governments in West Bengal are
dominated by their larger members.
38. One might begin with the null hypothesis that the small gram panchayats have too
little capacity to play a significant role in the rural local government fiscal system. If
this is the case, one would expect that the per capita expenditures of smaller GPs would
be lower. To test this, we examine the expenditure pattern for 2,959 gram panchayats,
divided into four population size categories, and report the results in the bottom panel of
Table 5.17
The percent of spending by gram panchayats in each size class is then
compared to the percent of population in that size class. These results (Columns 2 and 3)
do not support the null hypothesis. They show that the smallest gram panchayats (under
15,000 in population) are more important in the local fiscal structure than their population
share would suggest. The 849 gram panchayats in this size class account for only 18.6
percent of the rural population, but for 24 percent of the total spending by all gram
panchayats. In all other population size classes, the share of population equals or exceeds
the share of expenditures.
39. If more heavily populated local governments are more developed, and have the
better capacity to deliver services, this is not the pattern that one would expect. In fact,
the notion that more heavily populated GPs in West Bengal are more prosperous is not
fully supported by the data. As may be seen from the matrix of simple correlation
coefficients in Annex Table B-3, larger GPs tend to have significantly lower literacy
rates, suggesting less wealth. However, they also have significantly lower proportions of
SC/ST populations, smaller shares of female population, and lower percentages of
marginal workers. These patterns suggest that more heavily populated gram panchayats
have higher levels of income. Unfortunately, there are no income or product data
available for GPs, so we cannot directly test for a relationship between each of these
factors and the level of development of a GP.
40. Another explanation of the seemingly counter-intuitive result presented in Table 5
is that the transfers associated with centrally sponsored schemes are distributed away
from more heavily populated gram panchayats, and this is the source of the negative
relationship between population size and per capita expenditures. In fact, the data are
consistent with this hypothesis in that the simple correlation between population size and
17
Of the 3,354 gram panchayats in West Bengal, fiscal data are reported in this study for 3,016. However,
population data are available for only 2,973. Our maximum sample size, when using population data, is 86
percent of the total number of gram panchayats.
13
per capita revenues from centrally sponsored schemes is negative and significant (Annex
Table B-3).
41. Finally, we may attribute part of this result to an economies of size effect that
leads smaller local governments to spend more in per capita terms. This is explained in
Box 1.
42. In Column 4 of Table 5, we report the distribution of own source revenues raised,
according to population size class. Our expectation is that larger population suggests
agglomeration effects, such as regional markets or entertainment events, or more
developed infrastructure. All of these suggest a greater capacity to levy taxes and assess
user charges. Again, the result is surprising. The largest gram panchayats (over 25,000
in population) account for about 19 percent of the total population but for only about 15
percent of own source revenues raised. The fact that own source revenues are more
concentrated among the gram panchayats with fewer than 25,000 in population suggests
that either the more heavily populated gram panchayats are less developed than might be
supposed, or that they do not make as much effort in revenue mobilization as do the
smaller places.
43. We have repeated this analysis for blocks and districts, as shown in the top panels
of Table 5. For the 288 blocks examined, the results are much the same as for gram
panchayats. The smallest size class accounts for only about 22 percent of the population
but over 29 percent of total spending. The largest block governments (those with
populations above 250,000), by contrast, account for a greater share of population than of
expenditures. And, as in the case of the gram panchayats, the largest blocks account for a
small share of total own source (non-tax) revenues mobilized relative to their population.
The hypothesis that larger populations lead to better capacity to spend and to raise
revenues is not consistent with these statistical results.
44. For the 17 districts studied, the same pattern holds with respect to the
concentration of expenditures (Table 5). The four districts with the smallest populations
(less than 2.5 million) account for a disproportionately large share of expenditures. The
reverse is observed for the four largest districts. This result is almost certainly due to the
allocation formulae for centrally sponsored schemes. However, unlike the case of GPs
and blocks, these four largest districts account for about 40 percent of the population but
over 50 percent of own source revenues raised. In terms of user charges and license
collections, and perhaps revenues generated from assets owned by these governments,
larger districts would appear to have some comparative advantage and perhaps more
willingness to make use of their revenue-raising powers.
Composition of Expenditures by Population Size Class
45. Is there much variation across gram panchayats with different populations in
terms of the distribution of their expenditures by function? The issue of most interest
here is whether the smaller local governments are forced to restrict their spending to more
14
administrative functions, perhaps because they have a limited capacity to deliver services.
The answer we get from the results reported in Table 6 is that there is a population bias in
the functional distribution of gram panchayat spending. The share of capital expenditures
in total expenditures (mostly for employment generation and low income housing
assistance) rises with population size18
. Two possible explanations for this pattern are
that there are greater needs for housing in more heavily populated gram panchayats, and
that funds available for housing and other infrastructure investment are allocated
disproportionately to more heavily populated places. (These two possibilities are
examined in more detail in the discussion below.) A third explanation is that the fixed
costs of operating a local government force smaller gram panchayats to allocate a larger
share of their budgets to administrative functions. In fact, the data in Table 6 show that
the smaller gram panchayats do allocate a disproportionately large share of their budgets
to administrative and salary expenditures.
46. The results for blocks, and particularly for districts, also show a dominance of
capital expenditures in the budgets (Table 6). This pattern would be consistent with the
view that districts are primarily involved in the implementation of central and state
schemes, and are much less a local government than a spending agent of the state.
18
A somewhat surprising result revealed in this table is that the expenditures on the housing programs
(IAY) exceed the expenditures on employment generation programs (SGRY) for all population groupings
of gram panchayats. Further investigation shows that this is not the case for 2003 nor for 2004 -- in both of
those years, employment generation program expenditures are larger than housing program expenditures.
15
4. REVENUE STRUCTURE
47. “How do rural local governments finance their budgets? To what extent do they
pay for services from own sources of revenues and to what extent do they pay from
central and state grants and transfers?” About 6 percent of all revenue of gram
panchayats is derived from own sources (Tables 7 and 8). This is less than the 20 percent
estimated by Rao, et. al. (2004), for Karnataka for 2001, and the 17 percent estimated by
Oommen et.al. (2004) for Kerala in 1999.
48. The share of revenues raised from own sources (including tax and non tax
revenues) is less for smaller gram panchayats than for larger places (Table 7). From the
data in this table, we can see a general pattern of gram panchayats of all sizes raising
more revenue from non-tax fees and charges than from tax revenues. The dependence of
blocks and districts on intergovernmental transfers is slightly higher than that of gram
panchayats (Table 7). This is because districts and blocks have no taxing powers. These
measurements underscore the finding that there is a very strong degree of revenue
centralization in West Bengal state.
Own Source Revenues
49. The Constitution provides for gram panchayats to have the power to tax, i.e., to
determine the effective rate at which certain bases will be charged. However, it is left to
the state governments to determine which taxes a local government may levy, as well as
the nature of the autonomy that local governments will have in determining their level of
taxation. The broad-based taxes are denied panchayats, in all states, and this seems
appropriate given their limited administrative capabilities. Otherwise, states differ in
terms of the revenue sources they assign to local governments, and there is by now a long
list of taxes that are locally administered in India. In their review of the practice in
Kerala, Gujarat, and MP, Subrahmanyam and Annamalai (2004, p. 275-276) report 14
different categories of tax that are in use.
50. Rural local government taxes in most countries tend to be narrow-based and
administratively difficult. But even so, they have the potential to significantly increase
the level of revenues available to the local governments. By one estimate, rural local
governments in West Bengal raise a per capita amount of own source revenue that is
equivalent to about one-fourth the all-India average (Alok, 2006, Table 6.6). Based on
this interstate comparison, it would seem that there is significant potential for additional
revenue mobilization by local governments in West Bengal.
16
51. The term “own source revenue” requires some explanation in the case of rural
local governments in West Bengal. The normal definition requires that local
governments have the power to at least set the tax rate (Bahl and Linn, 1992, chapter 12;
Bird, 1999). For the property taxes and the entertainment tax in West Bengal, the rate
and base are prescribed by the State Government in the Panchayat Act. The gram
panchayats have the power to set a property tax rate, but may not exceed the ceiling rate
set by the state. Gram panchayats do have some discretion, however, in their
administration of the property tax and in terms of setting the rates of certain non-tax fees
and charges.
52. Property and Land Taxes. The gram panchayat level in India has been
empowered in most states to levy a land and building tax. In theory, the base of the tax is
the annual rental value of land and buildings.19
In practice, the tax is levied on an area
basis, or a housing unit basis, or on some other notional basis. At least one study in West
Bengal reports a “backward” method where the amount of tax is determined first and then
taxable property value is calculated as a residual (Pal and Adak, undated). It seems fair
to say that the current practice is one where there is no valuation process by which some
scientific method is used to estimate a market value. Neither is there a method to revalue
properties in a systematic way. The local governments are responsible for keeping the
tax rolls and for collections and enforcement.
53. In West Bengal, unlike in many other states, the tax base includes both
agricultural and non-agricultural property. Properties with a value less than Rs 250 are
exempt from tax. The maximum tax rates, set by the state government, are one percent
for properties with an annual value less than Rs 1000 and two percent for properties with
an annual value greater than Rs 1000. In addition, the gram panchayat may levy a
property transfer tax on immovable property. The tax rate is 2 percent and the base is the
selling price (consideration) in an arms length transaction.
54. We do not have data on property tax revenue. However, a reasonable guess is
that most of the tax revenue reported for gram panchayats in West Bengal is property tax.
Per capita total own source revenues for GPs is about Rs 8. Property tax performance is
very weak.20
55. One might offer a number of explanations for this poor revenue performance.
First, properties may be dramatically under assessed. We have no direct evidence on this
but can report a widespread belief in West Bengal that valuation practices are very ad
hoc. Second, collection rates can be very low. Pal and Adak (undated) estimated a ratio
of collections to demand (assessed liability) of 26 percent in 2001. A recent SRD study
(undated) made a similar estimate of the collection rate, based on data taken from a
sample of gram panchayats21
. Low collection rates may be traced to several factors:
19
Annual rental value is defined to be the equivalent of 6 percent of market value. 20
Rao, et. al. (2004) reports about the same level for Karnataka. 21
Rao, et. al. (2004), estimate a 69 percent collection rate in Karnataka in 2003, but noted that this is
against a base that has not been revised in 30 years.
17
lax or inept administrative practices
inadequate staffing and poorly trained staff
residents do not see the value of paying taxes in terms of the
services they receive from the GP, and therefore resist payment.
GP leaders are hesitant to enforce the tax on the local elite.
56. Entertainment Tax. The entertainment tax is levied on admissions to various
events. The Act defines entertainment as exhibitions, movies, performances,
amusements, and games or sports to which persons are charged admission. The tax rate
is 10 percent.
57. In West Bengal, the tax is levied and collected by the State government with 90
percent of the proceeds returned to local governments. The return is by formula: 80
percent to urban local bodies and 20 percent to rural local governments. Therefore, the
entertainment tax is really an intergovernmental transfer, i.e., the tax rate and base are
determined by the central government, collections are made by the central government,
and the central government decides on how the proceeds will be divided among local
governments.
58. The first SFC recommended turning this tax over to local governments, and
giving them discretion to set the rates. The state government did not accept this proposal.
The Second State Finance Commission did not call for full conversion to a local tax, but
did hold to the recommendation that revenues be fully devolved. The State government
has for the most part accepted the recommendation of the second SFC. It argues that an
intergovernmental transfer is a better approach than a local entertainment tax because it
(a) can reflect the superior ability of the state government to collect the tax, and (b) can
allow rural local governments (whose residents may travel to urban entertainment events)
to share in the revenue proceeds.
Grants and Transfers
59. About 94 percent of total revenues of the gram panchayats come from grants and
transfers. The share for districts and blocks is even higher. The composition of these
grants and transfers is described in Box 2, with revenue distributions for 2005
summarized in Tables 8 and 9.
Box 2
The Components of Intergovernmental Transfers
1 State Government
Finance Commission Grants
Untied Grantsa
Salary Grants
Other Grants
18
BEUP
State Sponsored Schemes (Minideep, PROFLAL, other)
2. Central Government
Finance Commission Grants
Centrally sponsored schemes (SGRY, IAY, NSAP, IMY, midday meal, SSA,
other)
MPLAD, other
a
Untied grants include funds devolved to the Panchayats from the Government, the use of which is not
restricted to any specific purpose. In some GP statements, untied funds are termed as ‘Shartaheen Fund’.
Untied funds include the following: (a) ‘Lump grant’ which is a lump sum amount received by the
panchayats from the Government and can be used by the panchayats for any development purpose, (b)
‘Matching grants’ which are funds received from the Government to bridge any deficit faced by the
panchayats in meeting their expenses; matching grants are also termed as ‘Paripurak Anudan’ or
‘Sampurak Anudan’ in some statements, (c) Other untied funds.
Centrally Sponsored Schemes
60. The centrally sponsored schemes are by far the largest share of revenues, as may
be seen in Table 8. About 75 percent of PRI transfers and 70 percent of PRI revenue is
from the centrally sponsored schemes. Over two-thirds of this total for gram panchayats
is accounted for by the IAY and SGRY schemes (Table 9). The share of centrally
sponsored schemes for blocks and districts is even higher (in the range of 80-90 percent
as indicated in Table 8), emphasizing again the agency role they play in implementing
central government programs.
61. SGRY. The SGRY is an employment generation programme for rural areas that
is targeted to benefit the poorest segment of the rural population (Government of India,
2006). The stated objectives of the program are to provide wage employment in rural
areas and to create a durable social and economic infrastructure. The program is
administered by the PRI, who have some discretion in deciding on the type and location
of public works projects that will be carried out. There are limits on this discretion.
Since this is a conditional grant, certain rules are prescribed on the use of these funds.
Moreover, the action plan of each PRI must be approved by the next level up, e.g., gram
by blocks, blocks by districts, etc.
62. The total funding for the programme for each year is decided on by the central
government in the course of the normal budget process. There is no fixed formula
defining the SGRY entitlement as a share of central taxes or of the central budget. In
recent years (2003-2005) the national plus state allocation has been Rs 6000 crore plus 50
lakh tons of food grains. The entitlement (vertical share) for each state is decided upon
by the central government, based on a formula that ranks states according to their share of
the rural poor population in India.
19
63. The Central Government amount22
for each state is paid directly to districts. This
is done according to a formula that gives equal weight to the proportion of SC/ST
population and to the inverse of production per agricultural worker. The Central
Government prescribes how the revenues will be shared among PRI within a district: 20
percent to the district governments, 30 percent to the blocks and 50 percent to the gram
panchayats. A matching requirement from the state government is mandated in the ratio
of 75 percent central and 25 percent state. The state government uses the same formulae
as the center in allocating its 25 percent share across rural local governments.
64. The districts make the allocations among blocks and among gram panchayats, but
again according to formulae laid down by the Central Government. The 30 percent
entitlement of blocks is distributed half by the proportion of SC/ST population and half
by the proportion of rural population, with both variables measured relative to that of the
district. The 50 percent share of gram panchayats is allocated by formula, subject to a
minimum entitlement of Rs 25,000 per local government23
. In fact, the gram panchayat
share of SGRY can be even larger than is indicated by these formulas. This is because
some district and block schemes are implemented at the GP level and the funds
associated with these schemes pass through the GPs. We cannot identify these amounts
separately.
65. The gram panchayat shares also can be smaller than these formulae suggest. This
is because some GPs do not spend the funds fast enough to trigger the full release of the
entitlement during the fiscal year.
66. Certainly the PRI have no discretion in determining the amount of their
entitlement under SGRY. Nor do they have discretion to shift this money toward
expenditures for other purposes. The discretion of gram panchayats in using this money
is also limited by the requirement that it must be spent on infrastructure projects that
employ unskilled labor, and that 50 percent of the funds be used for infrastructure
development work in localities with a larger SC/ST population.
67. In theory, the PRI do have discretion in project selection. They are charged with
developing an action plan that lays out the projects to be undertaken. (These plans must
be approved by the next higher level of government). In practice, the discretion that PRI
have in implementation may be very limited. One study argues, with respect to SGRY
that “… the role of PRIs in planning and implementation is insignificant” (World Bank,
2005, p6). It is argued that line department officials take on a supervisory role and that
central guidelines lead all key decisions about projects.
68. As may be seen in Table 9, SGRY transfers are a larger share of
intergovernmental transfers in districts than in either blocks or gram panchayats. This is
22
Actually, only 95 percent of the full entitlement is allocated in this way. The Ministry retains 5 percent
to allocate to areas of acute distress arising out of extraordinary seasonal conditions. 23
After 2004, the allocation to each gram panchayat was made in proportion to their previous year’s
allocation.
20
a surprising finding in that districts are supposed to retain only 20 percent of the total
received, and the gram panchayat share is 50 percent.
69. IAY. The Indira Awaas Yojana (IAY) is a flagship scheme of the Ministry of
Rural Development to provide houses to the poor in the rural areas (Government of India,
2006). The objective of the program is primarily to help construction/upgrading of
dwelling units of members of Scheduled Castes/Scheduled Tribes, freed bonded laborers,
others below the poverty line, and other needy groups24
by providing them with financial
assistance. The construction of the house is the sole responsibility of the beneficiary.
70. The IAY is a Centrally Sponsored Scheme funded on a cost-sharing basis between
the Government of India and the State Governments, in the ratio of 75:25.
71. Central assistance under the Indira Awaas Yojana was originally allocated to the
States/UTs on the basis of poverty ratio and housing shortage, with each of these
variables being given equal weightage. Since 2005-2006, the weightage is 75 percent
housing shortage and 25 percent poverty. The estimated poverty ratios prepared by the
Planning Commission are used for this purpose, while housing shortage is determined on
the basis of the most recent census data available. The proportions of rural SC/ST
population (25 percent) and housing shortage (75 percent) in a district, relative to that of
the State/UT, are the criteria for the allocation of the Indira Awaas Yojana funds within a
State/UT. The numbers of houses to be constructed for each block and gram panchayat
within a district are decided in the same way, but by the district government rather than
by a strict formula. Once the block or gram panchayat is notified about the number of
beneficiaries, it may decide on the specific beneficiaries, according to the guidelines of
the program. Hence there is some discretion for the local government to decide on the
distribution of benefits.
72. As may be seen in Table 9, IAY transfers are a larger share of intergovernmental
transfers in gram panchayats than in districts. Block level governments in West Bengal
were little involved in the IAY program in 2005. These results do not suggest a pattern in
terms of the involvement of PRIs of different population sizes.
73. Other Central Schemes and Grants. There are a number of other types of central
transfers to rural local governments. The “other schemes” category shown in Table 9
includes several assistance programs that are targeted on various sectors. Together, these
amount to about 10 percent of total central transfers in the case of districts and blocks,
and 7.5 percent in the case of gram panchayats.
74. The Member of Parliament Local Area Development Scheme (MPLAD) has the
goal of enabling members of parliament to suggest and facilitate execution of
development works based on local needs. The amount of MPLAD allocations are not
trivial, especially by comparison to state government allocations to PRIs, as may be seen
in Table 9. The funds go primarily to the block level.
24
The program benefits have been extended to the families of ex-servicemen killed in action, and to
physically and mentally challenged persons.
21
75. The Twelfth Finance Commission grants, earmarked for operating costs of water
and sanitation, account for a significant share of transfers. This award continues a
tradition begun by the Eleventh State Finance Commission. The distribution among
districts, blocks, and gram panchayats is shown in Table 9.
Grants and Transfers: State Government
76. The state government provides financial transfers to the PRIs through three types
of programs: State grants, State sponsored schemes, and BEUP. These assistance
programs, to the extent they carry conditions, allow the state government to influence the
pattern of spending by the PRI. State government transfers are a relatively small 25
percent of gram panchayat revenues and an even smaller share for districts and blocks.25
77. State Grants. There are several state government transfers in the financing
structure in West Bengal, but they take on most significance as a share of revenues in the
case of gram panchayats (See Table 8)26
. The State Finance Commission grant has the
feature of being unconditional, hence it is in step with a decentralized fiscal strategy. It is
this program that has drawn the attention of the two State Finance Commissions and led
to the recommendation of a 16 percent vertical share. But, as may be seen in Table 10,
the state government made no distributions under this program in 2004 and 2005. In
2006, the amount distributed was fixed at Rs 278 crore, and it is reported that the intent is
to distribute this same amount in 2007.
78. The most important component of state grants is the salary grant. This is a cost
reimbursement allocation to PRIs based on their number of approved posts. Since the
state government approves posts, and sets pay grades, this transfer leaves the local
governments with little discretion in terms of how the grant can be spent. The salary
grant is equivalent in amount to nearly 80 percent of all state transfers received by gram
panchayats and nearly 60 percent in the case of districts (Table 10).
79. Third, there are the so-called “untied grants”. The untied grants are a collection
of unconditional grants (e.g., the lump sum grant, matching grants) that are of primary
importance to district governments.
80. Finally, there are the BEUP grants made through state legislatures. As may be
seen in Table 10, these grants (similar to the central MPLAD grants) are most important
as methods of financing block level expenditures.
81. State Sponsored Schemes. In addition to the State Finance Commission and the
salary grants, the State provides funding to the PRIs for various specific schemes. These
schemes are targeted programs, for which the PRIs have little expenditure discretion.
The schemes include:
25
The state matching share for the CSS is not included in the state transfer category reported in Table 9. 26
This is because salaries comprise an important share of the gram panchayats budget, and these are
financed by the salary grant.
22
“Minideep”, which is an irrigation scheme to promote groundwater
irrigation. The mini deep tubewells that are supported by the program are
operated by electricity and all the operation and maintenance costs are
borne by the State government.
PROFLAL or Provident Fund for Landless Agricultural Laborers is a
social security scheme for the landless agricultural laborers
The fiscal importance of the state intergovernmental transfers is described in Table 10.
These constitute less than 3 percent of the total amount of state transfers received by
gram panchayats.
23
5. ANALYSIS OF THE DISPARITIES IN GRAM
PANCHAYAT FINANCES
82. There are significant variations in the socio-economic conditions existing in gram
panchayats, blocks and districts. These underlying conditions are the basic determinants
of expenditure needs, fiscal capacity, and the ability to attract central government
transfers, and should help us explain fiscal disparities among local governments. Most
empirical studies begin with per capita GDP as a standard for measuring variations in the
capacity to finance services. In West Bengal, however, we have no data on per capita
GDP below the district level. As an alternative, we might use the data available on socio
economic condition to justify some proxy measures of poverty and economic
development.
83. Various analyses in West Bengal have used four proxy measures of the
concentration of poverty in gram panchayats: (a) the percent of population in scheduled
castes and tribes, (b) the percent of the population female, (c) the percent of marginal
workers, and (d) the rate of illiteracy.27
In this study, we use population to measure size,
and to measure the level of economic development, we use the literacy rate and the
percent of employment in non-agricultural labor.
84. The variations across PRIs in these indicators are shown in Table 11. The mean
values are reported in the first column of each panel, with the number of units reporting
shown in parenthesis.28
The next three columns show the range and relative variation in
each indicator. For example, for the second row, we can see that for the 17 districts
reporting, the average share of SC/ST in the total population is 32.1 percent. The lowest
share is 13.3 percent and the highest is 55.6 percent, with a coefficient of variation of
37.2.29
85. Note from the minimum and maximum values in the table that the variation
across gram panchayats is quite large in the case of some of these measures. For
example, the literacy rate among gram panchayats ranges from a low of 12.5 percent to a
high of 79.4 percent. Such variations may be telling for variations in fiscal performance.
Literacy is likely to be associated with stronger economic development and therefore
revenue mobilization. We expect that gram panchayats with low levels of literacy will
27
See, for example, State Finance Commission (2002). 28
All means are unweighted. 29
The coefficient of variation is the standard deviation as a percent of the mean, hence it is a measure of
relative variation. By this measure, for example, there is more variation across districts in the percent of
SC/ST population than there is variation in the literacy rate.
24
obtain less own source revenue than those with higher literacy rates.30
The disparities in
literacy rates are also a reflection of expenditure need, which is likely to be much more
acute in those gram panchayats with low levels of literacy31
. This should be reflected in
the flow of revenue from central schemes for poverty alleviation. The percent of
marginal workers in the population also shows large disparities across gram panchayats.32
While the average percent of marginal workers among GPs is 11 percent, in one GP, it is
over 40 percent. In gram panchayats where there are larger proportions of marginal
workers, there tends to be a significantly larger share of SC/ST population, a larger share
of females in the population, and a lower literacy rate. The percent of marginal workers,
then, seems a reasonable proxy for the backwardness of a gram panchayat, and an
indicator of greater expenditure needs.
86. There also are substantial differences among districts in their level of economic
well being. Note from Table 11 that the richest district has a per capita GDP that is
roughly twice the level of the poorest district. We have developed an inter-correlation
matrix for districts on these same measures for the 15 of 18 districts for which data are
available. We find that in richer districts (measured by per capita gross product), there
tends to be a significantly smaller share of SC/ST population, and a significantly higher
literacy rate.
Expenditure Disparities
87. There are great disparities across local governments in West Bengal in the
amounts they spend. An accounting of these disparities will give a baseline for
determining the equalization job that the intergovernmental transfer system will be
required to do.
88. District Aggregates. The disparity between the highest and lowest income
districts in West Bengal is about 1.8 (Table 12). We examine whether rural local
government expenditure disparities are as large by computing an aggregate per capita
expenditure of district, block, and gram panchayats governments. The results, shown in
Column (2) of Table 12 show a disparity of 3.1 between the highest and the lowest
districts. Spending disparities are nearly twice as great as income disparities.
89. We examine the pattern of disparities in Figure 2 with aggregate per capita
expenditures on the vertical axis. The pattern we observe suggests equalization. The
four poorest provinces are among the highest spenders on a per capita basis and three of
the four richest are among the lowest.
30
The simple correlation between the literacy rate and per capita own source revenue is positive and highly
significant (see Annex B). 31
The simple correlation between the literacy rate and the percent of SC/ST population is negative and
highly significant (See Annex Table B-3). 32
A ‘marginal worker’ is one who has not worked in the past six months.
25
90. Gram Panchayats. To study the variation across gram panchayats, we have
computed Table 13, which shows the overall variation in per capita total expenditures of
gram panchayats. As above, we report mean values as well as variations.
91. Gram panchayats spend very little on a per capita basis, only about Rs 138 per
person on average, i.e., US $3.14 per capita. The level of per capita expenditures fell
from 113 rupees per person in 2003 to 102 in 2004 but then increased from 102 to 138
between 2004 and 2005. This continuing, very low level of expenditures suggests that
gram panchayats are “under-assigned” expenditure responsibilities with the result that the
services they do provide may not have much affect on the quality of life in the local
government area. The result is that voters might not be very interested in getting
involved to influence the level and structure of local government expenditures. In fact,
some studies point to the non-functioning gram sabha as evidence of little citizen
participation in budgeting decisions (Institute of Social Sciences, 2005). This is a major
obstacle to realizing the local self-governance outcomes sought by the constitutional
amendment. To address this problem, government could rethink the role of gram
panchayats in delivering services, in the direction of giving them additional assignments
of expenditure responsibility, as well as access to additional resources.
92. The large variation among gram panchayats in per capita spending level, in 2005,
is shown by the frequency distribution in Figure 3. We first rank GPs from lowest to
highest per capita expenditure (2005). We then create groups of per capita expenditures
in 20 rupee increments (0 to 20, 20 to 40, 40 to 60, and so on) and fit each GP into their
respective group. The second to last group is 480 to 500 rupees per person and the very
last group is “more than 500 rupees.” The level of per capita expenditure is plotted on
the X-axis and the cumulative frequency (number of GPs) is plotted on the right vertical
axis, while the absolute frequency (actual number of GPs in an expenditure category) is
plotted on the left vertical axis. For example, there are 591 GPs in the per capita
expenditure group “80 to 100 rupees per person.”
93. Over 83 percent of all GPs spend 180 or less rupees per person. The distribution
has a “long tail” meaning that there are a number of individual GPs that spend more than
180 rupees per person, but the level of spending for those GPs is not concentrated in any
one expenditure group. The higher spending amounts range from 180 rupees to 4,515
rupees. In such cases, the services delivered are likely to be meaningful. Understanding
the reasons for these wide disparities may be the key to developing a strategy to increase
the fiscal importance of gram panchayats. We argue that these disparities are due either
to the way in which grants and transfers are distributed, or to higher levels of own source
revenues in some gram panchayats, or to an inadequate capacity to absorb expenditure
responsibility on the part of some gram panchayats.
94. The distribution of per capita expenditures across population size classes, as
shown in Table 13, is surprising. On average, the smallest gram panchayats spend more
per person than do the more heavily populated gram panchayats33
. As noted above, this
33
A regression analysis on gram panchayats in Kerala also found a significant negative relationship with
population size (Oommen, et. al., 2004).
26
could be a fixed cost effect, i.e., local governments must spend a certain amount to satisfy
the basics of operating a local government and when this total is spread over a larger
population, the per capita amount falls. But, the fact that 80 percent of gram panchayat
spending is for capital and income maintenance purposes suggests that there are other
explanations. Chief among these is the possibility that the formulae to allocate grant
funding and other transfers may favor local governments that are less heavily populated.
95. We recalculate the relationship between per capita spending and population size,
with fixed costs removed from the expenditure variable. Since the state government
provides approximately the same salary grant for all gram panchayats, we might take
salary expenditures as a measure of fixed costs (See also Box 1). We subtract salary
grants received from the state government from total expenditures of gram panchayats,
reproduce the calculations from Table 13, and report the results in Table 14. The pattern
is much the same as reported in Table 13. The per capita expenditures of gram
panchayats are higher for the less heavily populated local governments. This result, then,
is due to more than the fixed cost effect. A hypothesis consistent with this finding is that
there is a bias in the allocation of central and state transfers that favors gram panchayats
with smaller populations.
Explaining Expenditure Variations
96. About 5 percent of the GPs report per capita expenditures that are very small—
averaging 50 rupees per person. The top 5 percent of GPs (in terms of per capita
spending) spend at least 2 ½ times that of the average GP. The very top of the
distribution—those GPs in the top 1 percent of the distribution -- spend over 584 rupees
per person. Understanding the determinants of this wide variation is a first important step
toward developing a financing structure that will allow the provision of a minimum level
of services in every gram panchayat.
97. As a first step, we calculate the simple correlation coefficients between selected
fiscal variables and selected socio-economic variables. This simple correlation matrix is
presented in Annex B (separately for districts, blocks, and gram panchayats). There are a
number of significant covariates. In particular, we can see that per capita expenditures for
gram panchayats are significantly higher where there are smaller populations, where there
is a larger percent of SC/ST population, where the percent of marginal workers is larger,
and where there is a greater share of agricultural population. Expenditure needs do seem
to matter. These simple correlations suggest that the joint impact of these population
characteristics on per capita expenditures is complicated, and involves more than first
order correlations. Therefore, we use a multi-variable regression to explain inter-
governmental differences in per capita expenditures.
98. We estimate OLS regressions separately for gram panchayats, blocks, and
districts, where per capita total expenditures is the dependent variable, and the
independent variables are justified as follows:
27
Population size should be negatively related to per capita spending, because for
smaller local governments, cet. par., the fixed cost effects will weigh heavily on
budgets. Moreover, central scheme revenues may be allocated in disproportionate
amount to local governments with a smaller population.
The percent of SC/ST population should be positively related to per capita
expenditures because this implies a heavier concentration of poor citizens who are
more costly to serve. For this reason, greater amounts of CSS transfers will flow
to local governments with heavier concentrations of SC/ST in the population and
these grants will result in higher per capita expenditures.
We do not have a prior on the marginal effect of variations in the literacy rate on
per capita expenditures. The effect should be positive if literacy signals more
willingness to pay on the part of residents and a greater capacity to deliver
services on the part of government. If a greater rate of literacy equates with more
poverty and therefore a greater inflow of transfers to support employment
generation programs, a negative effect might be expected.
The share of workers in the agricultural sector might indicate more need for
services because services currently provided in the more remote areas may be
more deficient and more costly to provide34
. It also may be positively related to
the level of per capita expenditures because of the likelihood that larger shares of
agricultural workers indicate a more agrarian economy and more demand for
employment generation programs. The simple correlations reported in Annex B
suggest that where the agricultural share of employment is high in a gram
panchayat, we can expect significantly less literacy, a greater percent of SC/ST
population, and a larger percent of marginal workers.
We have included a dummy variable to take account of district of location for
blocks and gram panchayats, and a dummy variable to take account of block
location in the gram panchayats analysis. This should pick up the impact of some
qualitative factors such as location, political power, and different attitudes and
levels of efficiency at the district and block levels. We expect that, all else being
the same, there will be “district effects” and “block effects”. To construct the
dummy variables, we have omitted Uttar Dinajur district and one randomly
chosen block in each district. The remaining district and block coefficients should
be interpreted as the “effect” of a particular district on the dependent variable,
relative to the omitted district (block).
99. The results of this OLS analysis are presented in Table 15. About 81 percent of
the variation in per capita expenditures among the 2,098 gram panchayats in the sample
can be explained by these variables. Population size exerts the expected significant and
negative size effect. The proxy measure for the concentration of poverty (SC/ST
34
Data on the share of workers in the agricultural sector are available for gram panchayats but not for
districts or blocks.
28
population) is significant in leading to higher levels of per capita expenditures. This is an
expected result because higher proportions of SC/ST population draw in more
intergovernmental transfers to address the greater expenditure needs. All else held
constant, a ten percentage point higher share of SC/ST population may be associated with
a 1.8 percent higher level of per capita spending by a gram panchayat35
. The literacy
rate exerts a positive marginal effect on spending. For any given level of population,
SC/ST population, etc., a higher literacy rate leads to a higher level of expenditures.
Better education leads to more demand for public resources and at the margin to a greater
willingness to pay for services. We might think of this as the impact of a higher rate of
economic development. A larger share in agricultural employment leads to significantly
higher levels of per capita expenditures, as hypothesized.
100. Gram panchayats in all districts spend more than those in Uttar Dinajpur, even
after account is taken of these explanatory variables. Seventy four of the 325 block
dummy variables were significant (23 percent), indicating that there are important
differences in spending levels within districts, even after we account for district effects
and for the socioeconomic characteristics included in Table 15. Based on this, one might
argue that there are important management and political factors to be considered in
explaining inter-GP variations in per capita expenditures.
101. One might conclude from this analysis that expenditures are significantly higher
in less populated and more backward GPs, suggesting that a considerable degree of
equalization is built into the system. At the margin, however, higher rates of literacy are
associated with higher levels of spending.
102. We repeat this analysis with the salary grant removed from the dependent
variable. This should allow an estimate of determinants, independent of the fixed cost
effects (See Box 1). The results, shown in Table 16, do not change the conclusions very
much. Population size remains a dampening influence on per capita expenditures, though
the elasticity is low compared to that reported in Table 15.
Own Source Revenues
103. Rural local governments in West Bengal raise very little revenue from own
sources. The average for gram panchayats is only about Rs 8 in per capita. However,
the variation is great. Some gram panchayats raise 20 to 30 times the average amount,
while 200 GPs report raising no own source revenue—with 140 of those GPs in the two
lowest population classes. This result is shown in Table 17 by population size class for
districts, blocks, and gram panchayats. Interestingly, the average level of per capita own
source revenue declines with population size for the gram panchayats36
.
104. We might turn to a more systematic approach to explain the considerable
variation in per capita own source revenues across gram panchayats, i.e., to estimate an
35
The alternate poverty measure, the percent of females in the population, was not significant, so was
dropped from the regression. 36
The simple correlation between population size and per capita own source revenue is negative, but not
significant, as is shown in Annex B.
29
OLS regression of the determinants of per capita own source revenues. Our question is,
“Why do some gram panchayats raise more own source revenues than do others?” The
answer to this question may be important to helping formulate an incentive policy for
stimulating revenue mobilization. The dependent variable in this analysis is per capita
own source revenues, including tax and non-tax sources. The independent variables are
included based on the following a priori reasoning:
Population size should be positively related to per capita own source
revenues because agglomerations of population suggest greater taxable
capacity. With a larger population, there should be greater levels of
economic activity, regional markets that would draw non-resident buyers
and sellers to the GP, more commercial dwelling units to tax, and more
services financed by user charges. On the other hand, if smaller GPs draw
more intergovernmental transfers, this may dampen enthusiasm for
revenue mobilization and a negative relationship might be expected.
A larger percent of SC/ST population suggests a greater concentration of
poor households, less taxable capacity, and a negative relationship with
per capita own source revenues. This might be reinforced if the
concentration of SC/ST population draws more transfers, and reduces the
incentive to mobilize more revenues.
A higher literacy rate suggests a greater taxable capacity and a positive
relationship with the level of own source revenue. This hypothesis is based
on the premise that more education leads to higher wages on the part of
the local population, and arguably to a greater willingness to pay taxes.
The percent of agricultural workers in the economy will signal more
difficulty in tax collection and arguably a weaker taxable capacity. We
expect a negative relationship with per capita own source revenue.
District and block effects are included in the GP regression and district
effects are included in the block regression.
105. The results of the analysis are reported in Table 18. About 55 percent of the
variation in per capita revenues across 2,067 gram panchayats can be explained by this
model. The literacy rate variable is significant and has the expected positive sign. This
would appear to measure the positive marginal effect of economic development and voter
awareness on the mobilization of own source revenues. The elasticity is relatively high,
i.e., if the literacy rate is 10 percent higher, the level of own source revenue collections
will be about 9 percent higher.
106. Greater percents of workers in the hard-to-tax agriculture sector do in fact dampen
the level of own source revenues, as hypothesized. Neither of the poverty variables are
significant determinants. The population effect is negative, i.e., there is no evidence of
agglomeration effects on revenue mobilization. However, the “population effect” may
30
have been obscured by the inclusion of the district and block dummy variables. There are
a number of district effects with Bankura and Hugli reporting higher per capita own
source revenues than the omitted district of Uttar Dinajpur (which is a relatively low
income district). In the case of several other districts a negative effect is observed (See
Table 18). Block effects are observed in about 17 percent of the cases, suggesting the
importance of qualitative factors as discussed above.
107. Intergovernmental transfers play so large a role in financing rural local
governments that they may have an impact on local revenue mobilization. Transfers may
dampen the enthusiasm of local governments to be aggressive about collecting tax and
non tax revenues. Alternatively, larger inflows of transfers might stimulate more local
revenue mobilization, i.e., they might generate a kind of local match to accommodate the
cost of larger government. To test this hypothesis, we repeat the regression analysis
described above but add per capita total grants and transfers as an (exogenously
determined) independent variable. The results (Table 19) show that for gram panchayats,
per capita grants and transfers are positively and significantly related to own source
revenues per capita. The latter results suggest that own source revenues and
intergovernmental transfers do not substitute for one another but rather are
complementary, i.e., the more grants received, the more own source revenue raised. The
result is not so far-fetched. The cash payment inflow under the schemes and grants may
well lead to increased tax and non-tax payments by beneficiaries and by those supplying
inputs in the implementation of programs. Note that when we introduce
intergovernmental transfers as an independent variable, neither the population nor the
poverty variables are significant.
108. In summary, we might interpret this analysis as showing that per capita own
source revenues in gram panchayats are higher where there is more literacy, where a
smaller share of the population employed in the hard-to-tax agricultural sector, and where
there is a greater inflow of transfers.
109. Measuring Tax Effort. One reason why this estimation of the determinants of
revenue mobilization is important is that we might use these results to monitor the
revenue mobilization efforts of gram panchayats. This would allow the State to identify
low performing gram panchayats where additional training and other technical assistance
is necessary to enhance collections. Or, an index of tax effort might be used in an
intergovernmental transfer distribution formula to provide an incentive for better tax
effort and a penalty for poor tax effort.
110. We may use the results generated above to develop an index of tax effort for gram
panchayats. The idea is to identify those GPs that are raising own source revenues at a
level above expectations and those that are performing below expectations. This is a
good alternative to using the average per capita amount raised as the “expectations” for
every gram panchayat. Fiscal capacities vary widely and some GPs should raise more
revenues than others, even if they exert the same effort. To control for this, we use the
following methodology:
31
From the equation shown in Table 18, we estimate the “expected” level of
own source revenues for each gram panchayat, based on the socio-economic
characteristics of that gram panchayat. For example, for the first gram
panchayat shown in Table 20 (Domahana), we would expect a per capita level
of own revenues of Rs 4.47.
The actual level of taxes divided by the estimated level is the index of tax
effort. This is shown as “effort” in Table 20. For example, the first gram
panchayat actually raises Rs 2.77 in per capita own source revenues, hence its
effort index is 0.62.
An effort index, for example of 0.62, indicates that Domahana gram
panchayat is raising own source revenues at a rate that is 38 percent below
expectations, based on a statewide comparison. By contrast, the gram
panchayat of Oldabari exerts a tax effort that is 107 percent above
expectations. The results for 10 of the more heavily populated gram
panchayats shown in Table 20, indicate an above average effort in three cases.
Explaining the Distribution Impact of Intergovernmental Transfers
111. The “rules” for vertical and horizontal sharing of intergovernmental transfers is
discussed at some length above. Now we turn to an analysis of the variations among
local governments in the actual amounts of transfers received. In particular, we want to
identify the “implicit formula” for the distribution of intergovernmental transfers across
gram panchayats and across districts and blocks. That is, we examine the actual
relationship between transfers distributed and measures of need such as the size of the
SC/ST population, population size, etc. We are especially interested in whether there is
any empirical evidence that this distribution is or is not equalizing.
112. To begin to answer this question, we have estimated an OLS regression with the
major components of per capita transfers and grants as the dependent variables. We study
the major components of the transfer system, because it seems to be the case that
different programs carry different objectives and are structured in different ways. We
would therefore expect the determinants of the amounts received by gram panchayats to
vary from one grant type to another. In each case, the idea is to tease out an “implicit”
grant formula, and in particular to see if per capita transfers are systematically allocated
to favor poorer jurisdictions.
113. Centrally Sponsored Schemes. We might begin with an analysis of variations in
the per capita amounts received from the two major centrally sponsored schemes: SGRY
and IAY. In the first regression analysis, the dependent variable is per capita SGRY
transfers. The explanatory variables include:
The percent of SC/ST population, which is an indicator of the
concentration of poor families. A positive association with per capita
32
transfers would be consistent with equalization. We expect a positive
association because of the poverty alleviation goal of the program and
because SC/ST population is a factor that government includes in its
distribution rules.
The literacy rate is expected to have a negative effect on the level of per
capita transfers received, because it is hypothesized that literacy is
associated with a smaller concentration of below-poverty-line population,
and therefore with a smaller inflow of transfers.
Population size is introduced as a control variable. We expect a negative
relationship with gram panchayats, in part, because there is a required
minimum allocation of Rs 25000 to each GP.
The percent of agricultural labor may indicate a larger rural sector and
therefore more demand for employment generation programs.
A district effect and a block effect. This might pick up some effects such
as better program administration, more political skill in attracting
transfers, etc.
114. The results for the SGRY scheme, presented in Table 21 show that the SC/ST
variable has the expected positive sign and is a significant determinant in the case of
gram panchayats. A one percent higher SC/ST population share, cet. par., is associated
with a 0.38 percent higher level of per capita SGRY transfers received. This result is
consistent with equalization. Gram panchayats with smaller populations also receive
significantly more. The unexpected finding is that the literacy rate is significant and
positive, suggesting a counter-equalizing influence in the distribution. The share of
agricultural labor is not significant. There are 8 significant district effects, and 12 percent
of the block dummies are significant, suggesting that other factors are at work in
determining the level of per capita SGRY receipts. In total, we can explain about two-
thirds of the variation in per capita SGRY receipts across gram panchayats.
115. We repeat this analysis for per capita IAY transfers, with the results shown in
Table 22. The percent of SC/ST population has the expected positive sign and is
significant. Per capita IAY transfers are also significantly higher in gram panchayats that
have smaller populations, and a larger percent of agricultural labor.
116. We find that, at the margin, a higher literacy rate is associated with a greater level
of receipts of IAY transfers. This runs counter to the backwardness hypothesis. About
65 percent of the variation in IAY transfers might be explained.
117. We carry out a similar analysis for total central schemes (Tables 23). The results
are much the same as those found for the IAY per capita grants. In the case of GPs, 79
percent of the variation is explained in the regression. Smaller GPs and those with more
33
backward populations (higher SC/ST) receive more central scheme funds per capita.
However, there is again a counter intuitive result regarding the literacy rate.
118. State Grants and Schemes. In Table 24, we analyze total per capita state
government transfers as a function of the same set of variables as reported in Table 22.
The dependent variable is the per capita amount received from all state grants and
schemes (Table 10). The goal here is to identify the determinants of per capita state aid,
and to compare these results against those for central schemes.
119. The independent variables in this analysis are the same as in the centrally
sponsored schemes. We are interested in the extent to which the distribution responds to
backwardness (percent of SC/ST population), population size, the share of agricultural
workers, and the literacy rate. A district and block effect are added to test for the possible
influence of intangibles such as location, political influence, or a more efficient public
administration.
120. The results of this analysis show that for gram panchayats, per capita state aid is
distributed significantly more toward those with smaller populations and larger shares of
agricultural employment (Table 24). The SC/ST, and literacy coefficients are not
significant. We may compare these results for state aid with those for centrally
sponsored schemes. In particular we might be interested in the question of whether state
transfers or central transfers are more equalizing. As the SC/ST coefficient is not a
significant determinant in the case of state transfers, we can conclude that the central
schemes are more equalizing. That is, if a gram panchayat has a 10 percent greater
concentration of SC/ST population (all else held constant) it does not receive a
significantly greater amount of per capita state aid but it does receive about a 3 percent
higher amount of per capita central scheme revenue.
121. Total Grants and Transfers. Finally, we use the per capita level of total state
grants and transfers as the dependent variable. This includes both central schemes and
state aid. The basic hypotheses are as above: per capita total grants and transfers should
be positively related to the percent of SC/ST population, and to share of employment in
agriculture, and negatively to the literacy rate, if the overall grant system features
equalization.
122. The results presented for gram panchayats in Table 25 show that about 82 percent
of the variation in per capita total transfers can be explained by this model. The percent
of SC/ST population is significant and positive, suggesting equalization in the
distribution of total grants and transfers. As noted above, this is due to the equalizing
influence of central transfers. There would appear to be a bias in favor of GPs with a
smaller population, and those with a larger share of employment in agriculture. These
findings are also consistent with the equalization hypothesis. As in most of this analysis,
these results show that the marginal impact of higher rates of literacy is positive.
34
6. FINANCIAL CONDITION
123. In this section, we examine the budgetary position of gram panchayats. We use a
simple pro forma (presented in Box 3) to describe budgetary position and assess the
magnitude of fiscal gaps, i.e., situations where annual revenues received (excluding
opening balances) do not cover expenditures made. We then turn to the issue of opening
balances. If these are adequate in amount to cover the recurrent fiscal gap, then the
opening balance can be viewed as a pre-funding of the annual budget. In this case, the
annual drawdown might be thought of as a recurrent revenue. In cases where opening
balances are inadequate to close the annual fiscal gap, the fiscal health of local
governments may be comprised. In studying this issue, we use data for three years (2003,
2004, and 2005).
Box 3
Pro Forma to Describe Budgetary Position
1. Total Revenues
1a. Own Source
Tax Revenue (property tax, amusement tax, other)
Non-tax Revenue (sale of property, leases, donations, other)
1b. Intergovernmental Transfers
Untied Grants
Salary Grants
Other Grants
Finance Commission Grants
Union
State
Centrally sponsored schemes (SGRY, IAY, NSAP, IMY, midday meal, SSA,
other)
State Sponsored Schemes (Minideep, PROFLAL, other)
Others (BEUP, MPLAD, other)
2. Less: total expenditures
2a. Current
2b. Capital
3. Equals: total annual surplus or deficit
35
4. Equals: difference between opening and closing balance
124. Based on this identity, we have calculated the indicators of financial condition
that are summarized in Table 26. These indicators are very simple, but might be useful in
profiling the financial condition of rural local governments in West Bengal. We draw on
these data to raise (and answer) four questions.
Do Many Gram Panchayats Have a Negative Recurrent Revenue Gap?
125. In fact, about one-half the gram panchayats reporting incurred a deficit in 2005
(Table 26). There was not much change in this number over the 2003-2005 period. The
deficits during this period were (on average) equivalent to about 13 percent of total
revenues. The other half of gram panchayats reported a surplus that also averaged about
13 percent of total revenues. There would appear to be a great unevenness in the fiscal
condition of gram panchayats, i.e., about half do not balance their current budgets and the
other half do not spend all of their recurrent annual revenues. In other words, about half
of GPs are able to add to their cash balances at the end of the fiscal year, and about half
find it necessary to draw from their balances to meet the gap between recurrent
expenditures and recurrent revenues available.
Do the Results For Surplus GPs Indicate an Inability to Absorb the Funds?
126. This is an especially important issue if a reform direction is to increase the flow of
resources to GPs. In fact, there are serious capacity problems at the GP level which may
indicate an inability to spend the money now available. An alternative explanation for
the accumulation of balances is that the conditionality and bureaucratic processes
associated with spending transfers from centrally sponsored schemes significantly limit
the chances that a rural local government can move the money in the year in which it is
received. Whatever the reason for this inability to absorb funds, it suggests the existence
of financial surpluses alongside deficiencies in public service levels in some gram
panchayats. We can note, however, that the same pattern of unspent balances was
observed for gram panchayats in Kerala and Karnataka (Sethi, 2005).
127. There is another possible reason for the surpluses. The receipt of
intergovernmental transfers late in the fiscal year makes it impossible to spend the money
in the year when the funds are received. Many GPs allege this to be a serious problem.
Some state officials see this timing issue as being overstated. The evidence presented
here points to the timing of receipts being part of the problem in West Bengal. Oommen,
et. al. (2004, p. 241-242), reach a similar conclusion in their study of Kerala.
“A rational transfer system should be predictable and ensure
an even flow throughout the fiscal period. This is especially important in view of
the existing situation in which grant receipts are bunched at the end of the fiscal
36
year, distorting the spending pattern and contributing to high closing/opening
balances. At present, the income-expenditure pattern is so uneven that the fiscal
balance fluctuates from a 200 percent deficit in August to a 40 percent surplus in
March. Closing treasuries for business on working days or imposing oral or
written instructions to restrict treasury transactions have exacerbated the problem.
Resolving the problems of the state-local transfer system in Kerala and resolving
the fiscal problems of the state government are, clearly, connected issues.”
128. The persistence of deficits is explored in Table 21a. During the 2003-2005
period, 199 GPs had deficits in all three years, but only a handful had deficits larger than
their opening balances. This suggests that while some GPs are regularly in a deficit
position, the size of the deficit relative to funds that are carried over is not large. About
one-third of the GPs had at least 2 years of deficit and another one-third had one year of
deficit. There does not appear to be any systematic relationship between the pattern of
deficits and spending levels. For example, we cannot see a relationship between “years
of deficit” (or surplus) and the average level of per capita expenditures for the GPs in this
category (Table 27). More GPs had surplus in all three years (257) than had deficits in all
three years. Failures in the area of financial indiscipline is not a widespread problem in
West Bengal.
129. If the deficit and opening balance problem is one of timing, the surplus should
remain about constant over a period of years (or should grow about in proportion to the
distributable pool). Local governments would bank the money received at the end of the
fiscal year, and then spend it down at the beginning of the next. Unless the flow of
transfers was reduced, one would expect balances to remain at about the same level over
time. This is the pattern we observe in West Bengal. Note from Table 26 that the size of
closing balances relative to total expenditures has remained between 20 and 30 percent
during this three year period. For those GPs running a surplus, the amount has stayed at
about 12 to 14 percent of total revenues. For those running a deficit, the amount has
ranged from 11.7 to 15 percent of total revenues. This analysis suggests that gram
panchayats are “pre-funded” by higher level governments and carry balances that are
sufficient for them to avoid year-end shortfalls.
To What Extent Do “Deficit” Gram Panchayats Draw On Their Cash Balance Reserves
to Cover the Cost of Delivering Services?
130. The approximately 1,500 GPs that annually operate recurrent account deficits, do
so in amount equivalent to around 11-15 percent of total recurrent revenue. As may be
seen for the reported ratio of deficit to balances in Table 28, these GPs have the resources
to balance their budgets.
131. We also can use these data to determine the extent to which opening balances are
adequate to cover the recurrent fiscal shortfall. In all but a handful of gram panchayats,
the balances are adequate. In most “deficit” GPs, the more accurate story would be that
certain recurrent expenditures undertaken in any given year were financed from transfers
received in the preceding year. From this analysis, we might conclude that financial
discipline problems, such as borrowing or deferring creditors to balance the budget, are
37
an issue for only a small number of gram panchayats in West Bengal. Dealing
individually with these is a manageable task for the fiscal monitoring and evaluation unit
of the State government.
Do the Closing Balances Change Over Time, and How Do We Explain This? Are
These Closing Balances “Too Large”?
132. The closing balances did not change very much in aggregate over the 2003-2005
period. They stayed in a range between 22 and 28 percent of total expenditures. We
might conclude from this, and from the analysis reported in Table 27, that, on average,
the closing balances are more than adequate to cover the shortfalls that deficit GPs face.
The question is whether these balances are too large and an indication that the gram
panchayats are receiving more in transfers than they can absorb. Balances between 20
and 30 percent of expenditures do seem large in a state where services in rural areas are
deficient and where the expenditures made by rural local governments are so meager.
133. Note from Table 26 that large balances are not the case in every GP. About 10
percent of all GPs closed their year with a balance lower than 5 percent of total
expenditures.
What Determines Financial Condition?
134. The state government should be interested in whether financial condition varies
among gram panchayats according to population size and concentration of poverty. Is
there a way to use data on the socio-economic makeup of GPs to develop an early
warning signal for fiscal distress? Or, might we use such data to classify GPs according
to whether they are “high-risk”? Such quantitative measures can help the state determine
where it might need to provide technical assistance in budget management, where it may
need closer surveillance, and/or how it might need to adjust intergovernmental transfers
to better fit the need for budget support. We might raise two questions in this research.
The first is, “What are the factors that lead some gram panchayats to incur a deficit?”
The second is, “What determines the size of the current account deficit?”
135. To analyze the propensity for GPs to be in deficit, we used a probit model, where
we define a “deficit GP” according to our pro forma in Box 3. The dependent variable is
(D=1) for deficit GPs and (D=0) for surplus GPs. We take two definitions of deficit GPs.
First, using data from Table 21a, we define “chronic deficit” GPs as a GP that has a
deficit for each year, 2003, 2004, and 2005. In the second case, we simply define a
deficit GP as any GP that posted a deficit for 2005.
136. The coefficients in Table 29 are estimates of the probability of incurring a deficit.
Our results show that the probability of choosing to run a current account deficit are
different for chronic deficit GPs than for all GPs that posted a deficit in 2005. We
hypothesize that a GP with a larger opening balance could “afford” to overspend against
38
current revenues, and so we might expect to see a positive correlation between opening
balance and the probability of a deficit. Interestingly, we see this for “regular” deficit
GPs (third column) but not for “chronic deficit” GPs. In fact, for chronic deficit GPs, we
observe an improbable relationship, that the probability of choosing a deficit is inversely
related to the size of the opening balance. The expectation is that the deficit choice for
the “chronic deficit” GPs would be driven more by underlying structural factors than by
cash on hand. For example, the percent SC/ST population might reflect a harsher
economic environment and therefore lead to higher deficits. We do observe this pattern
for chronic deficit GPs. Also, we note that in “regular deficit” GPs, own source revenues
per capita are associated with a smaller probability of deficit, which is an expected effect.
137. To answer the second question, we have estimated an OLS regression with the
dependent variable expressed as the per capita deficit (excluding any use of cash
balances). What we explain in this analysis is the variation in the size of the
revenue/expenditure imbalance, across those gram panchayats that ran a deficit in 2005.
“Surplus” GPs are excluded from this analysis. The independent variables and their
justifications are:
The per capita level of the opening balance. The hypothesis is that the
larger the available opening balance, the more likely is a gram panchayat
to incur a large current account deficit. Hence we expect a positive
regression coefficient.37
In estimating the relationship between the deficit and the opening balance,
we control for both population size and the percent of SC/ST population,
which we have shown above to be significant determinants of expenditure
levels. A priori, we would expect more backward gram panchayats to face
more budget pressures and to be more prone to run a deficit. To the extent
that more backward places are less able to absorb funds, there is another
rationale for expecting a positive relationship.
The per capita level of centrally sponsored schemes is introduced as an
explanatory variable. A positive coefficient would indicate that deficits
are driven up by higher transfers because of overspending. A negative
coefficient indicates that more scheme money dampens the deficit, a
finding consistent with the story that more transfers encourage increased
own source revenue mobilization.
There may be a “district effect” on financial condition. Some districts
may be more effective than others in tracking expenditure needs, training
local officials, and passing resources to GPs where needs are greatest. To
account for the district effect, we include a set of district dummy variables.
37
The dependent variable, per capita deficit (measured in rupees) is expressed as a positive number.
39
138. The results of the regression analysis (Table 30) confirm the null hypothesis. We
do find the hypothesized positive relationship between the size of the opening balance
and the size of the deficit, i.e., we find that gram panchayats that carry a larger opening
balances run significantly larger deficits. This finding indirectly suggests that larger
opening balances dampens the enthusiasm for higher levels of revenue mobilization. We
also find some evidence of a district effect, i.e., the size of the deficit in Darjiling and
Koch Bihar districts is significantly larger than in Uttar Dinapuri district. This finding
confirms the belief that the management of fiscal affairs in some districts is significantly
different than in others. The poverty variables are not significant determinants. About 55
percent of the variation is explained with this model.
40
7. THE FISCAL POSITION OF BLOCK AND DISTRICT
LEVEL GOVERNMENTS
139. Block and district level governments are more in the nature of implementing
agents of the state than they are local self-governments. Nevertheless, they manage about
two-thirds of PRI spending. The focus in this report is on the gram panchayat level.
However, to get a complete picture of PRI fiscal activity, it is also important to
understand the structure of finances of blocks (panchayats samitis) and districts (zillas).
140. The richest district in West Bengal has about twice the level of per capita gross
product as the poorest. They range in population size from 1.9 million to 9 million. The
simple correlations in Annex Table B-2 show that higher income districts tend to have a
significantly lower percent of SC/ST population and female population, and a higher
literacy rate.
141. The variation in the population of blocks also is large, from under 70 thousand to
over 400 thousand (Table 11). The simple correlations in Annex Table B-1 show that
larger blocks have significantly greater shares of SC/ST and female population, and a
larger percent of marginal workers. Larger blocks, apparently, are lagging in terms of
economic development and have a larger concentration of poor females.
142. The question we raise here is whether these variations in socio-economic
performance explain the different levels of fiscal activity.
Expenditures
143. On average, districts spend about 12 percent more on a per capita basis than do
gram panchayats. Block level governments spend well less than either (Table 31). The
variation in per capita spending across districts, however, is less than that across either
blocks or gram panchayats (note the size of the coefficients of variation). The pattern
found for gram panchayats, that smaller local governments spend more on a per capita
basis than do more heavily populated local governments, also holds true for districts and
blocks (Table 31).
144. We have carried out roughly the same analysis for districts and blocks as we did
for gram panchayats, with regression results reported in Table 32. The results for
districts are similar to those found for gram panchayats, as reported above. We use only
two variables in the district equation: population size and percent of SC/ST population.
41
About 63 percent of the per capita expenditure variation across 17 districts can be
explained, with the percent of SC/ST population the dominant determinant. Population
size has the expected negative coefficients, but is not significant at the .05 level.38
What
we can conclude from this is that per capita spending by district level governments is
significantly higher in districts where there is a greater concentration of poverty. As in
the case of gram panchayats, we might attribute this to the heavy weight attached to
backwardness in the distribution formulae for the centrally sponsored schemes. What is
interesting in these findings is that the percent of SC/ST population exerts a greater
marginal effect on per capita expenditures in the case of districts than in the case of gram
panchayats.
145. For the block level governments we can explain about 50 percent of the variation,
with both population size (negative) and the percent of SC/ST population (positive)
significant determinants. Note also that we find a significant “district effect”, with blocks
located in Koch Behar, Gopiballarpur, and Purulia districts spending significantly more
than those in Uttar Dinajur district. This suggests that some factors, unique to districts,
are important in explaining why some blocks spend more than others.
146. We might summarize these findings on the determinants of rural local
government expenditures in West Bengal with the following stylized conclusions:
Larger block governments spend less on a per capita basis than do smaller
block governments. There is no significant relationship between
population size and spending by district governments39
.
The greater the concentration of poor families, the higher will be per
capita expenditures. This conclusion holds for all three levels of rural
local government, and it suggests that the intergovernmental fiscal system
is dominated by an equalization objective.
Own-Source Revenues
147. Districts and block governments have no taxing powers, but can raise own
revenues from non tax sources. As may be seen in Table 33, the level of non-tax
revenues raised by block governments is very small. The average is less than Rs 3 on a
per capita basis and the largest amounts raised are less than Rs 40. The level of per
capita own source (non tax) revenues declines with the population size of the block and
the variation within each population size group is great. (Table 33).
38
The analysis in Table 32 was repeated with the dependent variable expressed net of salary grants. The
results were little different (See Table 32a). 39
To test whether this is due solely to a fixed cost effect, we reduced the dependent variable by the size of
the salary grant. There was very little change from the results (See Table 32).
42
148. With respect to districts, the pattern for per capita non tax revenues is
more in line with what one would expect. The more heavily populated districts, where
one would expect to see the greatest level of economic activity, show higher levels of per
capita collections than do the smaller districts. On average, however, only about Rs 5 per
capita is raised by district governments.
149. Since blocks and districts act as implementing agents of the central and state
governments, this poor revenue performance suggests little commitment to cost recovery
at any level.
150. There are variations in this generally low rate of revenue mobilization. We test
the hypothesis that the variation can be explained by the level of economic development
using an OLS regression analysis with per capita own source revenue as the dependent
variable. We use the various measures of economic development and backwardness that
are available, but cannot explain a significant amount of the variation for either blocks or
districts (Tables 34 and 35). We conclude that there is a large random element in the
pattern of variation in own source revenue per capita.
Intergovernmental Transfers
151. State and central transfers are distributed downward through the PRIs, first to
districts, then to blocks, and finally to gram panchayats. The question we raise here is the
extent to which central and state transfers are distributed in an equalizing way among
these PRIs.
152. Centrally Sponsored Schemes. We can explain about 60 percent of the variation
in per capita SGRY receipts for blocks and districts (Table 36). In the case of blocks,
those units that are more heavily populated receive significantly less in employment
generation grants. A higher share of SC/ST population draws more SGRY transfers per
capita. At the margin, the literacy rate is associated with greater SGRY transfers per
capita, which is a surprising result.
153. The results for districts, also reported in Table 36 are mostly in step with
expectations. On a per capita basis, SGRY transfers are allocated in greater amounts to
GPs with a heavier concentration of SC/ST population and a lower literacy rate. Oddly,
however, we find that the marginal effect of per capita gross product on per capita SGRY
is positive.
154. If we use the percent of SC/ST population as the basic measure of poverty, we can
say that the SGRY scheme is driven by equalization at both the block and district level.
155. In the case of blocks, we can explain a significant percent of the variation in IAY
transfers but cannot learn much about the determinants of the per capita distribution
(Table 37). At the district level, we can explain about 60 percent of the variation (Table
37). The implicit distribution seems heavily weighted toward indicators of
43
backwardness, i.e., percent of SC/ST population, and the illiteracy rate. Per capita IAY
scheme amounts are transferred with a bias in favor of districts with a larger population.
What we can say from this result is that the goal of the center -- to distribute housing
scheme revenue toward pockets of poverty -- seems to work better at the district level
than at the block level.
156. The next question is how this all plays out in terms of the per capita distribution
of total central scheme revenues. For blocks, we can explain only about 35 percent of the
variation. Both population size (negative) and SC/ST population (positive) are
significant (Table 38). Overall, there is an equalizing feature in the system.
157. For districts, SC/ST population and the literacy rate variables are significant with
directions of effect that are consistent with equalization. Again, however, we find that at
the margin, higher income districts receive more in per capita central transfers. Over 80
percent of the variation can be explained.
158. State Grants and Schemes. The regression results for per capita state government
transfers are reported in Table 39. For blocks, the variation would appear to be random,
except for a number of significant district effects. The explanation for why some blocks
receive more state government transfers may be driven more by political and
management factors than by the makeup of the population. For districts, we see some
conflicting equalization results. Literacy is positively correlated with per capita state
transfers, but a larger female population is positively related to state grants and per capita
GDP is negative.
159. Total Grants and Transfers. The last question is this all plays out in terms of the
distribution of total transfers from higher level governments. The results for districts and
blocks for per capita total transfers are shown in Table 40. The only significant
determinant of inter-district variations in per capita grants and transfers is the share of
SC/ST population, a result that is consistent with the equalization hypothesis. We find
essentially the same result for blocks, but note the much greater equalizing effect in the
case of districts. Overall, we find that the regression explains less of the variation for
blocks (48 percent) and districts (61 percent) than for GPs.
Financial Condition
160. The fiscal balance problem does not appear to be restricted to gram panchayats.
This is an unexpected finding since districts and blocks function more as spending agents
of the state than as autonomous local governments. As reported in Table 41, most district
governments closed the years 2003, 2004 and 2005 in a deficit position40
. The size of the
deficits, relative to total revenues of the district, were larger than in the case of gram
panchayats, as were the closing balances. The story seems to be much the same as for the
GP, i.e., districts carry large balances, presumably to pre-fund expenditures. We could
find only one district with a closing balance as small as 5 percent of total expenditures.
40
We use the pro forma in Box 3 to measure the deficit.
44
161. The financial position of blocks showed a great deal of variation. About half
incurred deficits, by our definition, and the amounts were in the range of 30 percent of
revenues. However, blocks carried balances that were substantially greater than that for
gram panchayats. The balances carried appear to be more than adequate to cover the
deficits. In general, the size of the deficit chosen is directly related to the size of the
opening balance that is carried into the fiscal year (Table 42). Again, the story seems to
be one of using allocations from the previous year to fund expenditures.
Summary
162. Though districts and blocks are more spending agents of the state and central
governments than local self-governments, they do exhibit variation in their spending and
financing patterns. Together, they account for a considerably greater share of
government expenditures than do gram panchayats.
163. The variations across the block and district levels show many of the same
characteristics that were found for gram panchayats. Per capita spending is higher in
blocks with smaller populations and in both blocks and districts where there is a heavier
concentration of SC/ST population. These expenditures are financed primarily by
intergovernmental transfers (97 percent) and there is no significant pattern of cost
recovery through fees and charges.
164. Centrally sponsored schemes are distributed among districts and blocks on an
equalizing basis, assuming that the percent of SC/ST population is an appropriate way to
measure equalization. There is less of a pattern of equalization in the distribution of per
capita state government assistance.
165. About half of all block level governments and nearly all districts show a shortfall
between recurrent revenues and annual expenditures. These gaps appear to be easily
covered by drawing on cash balances. As in the case of gram panchayats, the practice
seems to be one of pre-funding the budgets by permitting districts and blocks to carry
large balances.
45
8. STATE GOVERNMENT AND FISCAL
DECENTRALIZATION
166. From the above discussion, one can draw the conclusion that the fiscal system in
West Bengal is heavily dominated by the state government. By itself the state accounts
for 76 percent of direct expenditures, and raises 96 percent of all revenues. A recent
analysis by Oommen (2006) estimates that the local government expenditure share in
West Bengal is only about one-half of the national average.
167. It is also the case that the state government in West Bengal has chosen to limit the
amount of fiscal discretion given to its local units, so by this measure we also can
characterize West Bengal as a very centralized fiscal system:
Gram panchayats have some power to tax, but the tax rates and the legal
tax base are determined by the state government. Local governments have
discretion mostly in terms of how they administer the tax.
To a large extent, expenditures are dictated by the conditions associated
with receipt of state and central grants and transfers. The untied amounts
received by local governments are limited.
The vertical share of local governments -- their “entitlement” -- is
determined by the central and state governments, annually and in an ad
hoc way. This limits possibilities for efficient fiscal planning by local
governments.
168. How could the state begin to move this situation toward fiscal decentralization?
The PRIs in West Bengal might be “upgraded” in two ways: (a) by increasing the size of
their expenditure budgets, and (b) by giving them more discretion to make fiscal
decisions. The first of these paths will be costly, if an enhancement in the scope and
quality of public services is envisioned, and if a corresponding net increase in local
government funding is given. The second would also imply new costs in the form of
investments to upgrade the capacity of local governments and the transition costs
necessary to get PRIs “on the learning curve”. The transition cost might include a short-
term outlay to cover temporary service level shortfalls. One way or another, fiscal
decentralization will imply a drain on the state budget.
169. The two avenues open to the state government to find the necessary resources to
finance decentralization are: an increase in state government tax revenue, and a
redirection of existing funds.
46
State Taxation
170. The state could enact a new tax or a general tax increase with revenues dedicated
to the PRIs. For example, a specified amount could be dedicated to fund an increased
amount of unconditional grants. Though not explicitly stated, something similar to this
was implied by the SFC recommendation for a 16 percent share in the state taxes
dedicated to revenue sharing with local governments.
171. Whether the resources necessary to finance fiscal decentralization can be found
will depend in part on whether the economy of West Bengal has grown fast enough to
generate the necessary surplus or fiscal space. Though West Bengal is a poor state in
terms of average per capita GDP, it has grown well above the all-India rate over the past
decade, thanks in large to robust growth in the agricultural sector (CMIE, 2006).
172. There also would seem to be room for a tax increase. West Bengal’s rate of
revenue mobilization is low by comparison with other states, and has declined over the
1998-2004 period. A number of research studies have reached this conclusion
(Government of India, 2004; Coondoo, et. al., 2005, The World Bank, 2005a). West
Bengal’s relatively weak tax performance is longstanding (Jha, et. al., 1999).
173. One could also make the case that the West Bengal State Government is not
presently in a good place to finance such an upgrading. The budget situation is still
emerging from a significant imbalance. Over the past decade, West Bengal’s fiscal
position was one of the weakest in the country (Rajmal, 2006; Purfield, 2004,
Government of India, 2004). Though the fiscal deficit is down from its 9 percent level in
2000, even the projected 2007 deficit level of 5 percent of GDP presents a significant
hurdle to overcome. Some significant and perhaps painful discretionary measures will
need to be taken to reduce this deficit. As Rajmal (2006) has pointed out, the state
government may have limited discretion to address the issue. Between 2002 and 2004,
interest payments and pension payments together were equivalent in amount to about 70
percent of revenue receipts. Moreover, there are significant subsidy programs that are
not easily cut back or withdrawn. Finally, there is the commitment to a large wage bill
for state government employees and the possibility of another fiscal hit by the next pay
commission.41
Redirection of Existing Funds
174. Redirection as a strategy would involve shifting resources from other programs
toward the support of local governments. The issue here is less about funding new
monies than it is about rearranging priorities.42
41
One research report estimates that the next (sixth) pay commission has the potential to create more stress
on state budgets than did the fifth pay commission. Crisil (2006) estimates an impact of as much as 3
percent of state GSP by 2011, an amount that is well above the 2.6 of the previous commission and above
the 1 percent targeted by the Twelfth Finance commission. 42
We do not consider the alternative of replacing centrally sponsored schemes with state assistance
programs. For a discussion, see Rajamaram (2001).
47
175. Redirection may be a very difficult strategy to implement. The budget speech of
2006-2007 (Government of West Bengal, 2006) makes the case that there are many other
important and high priority claims on budgetary resources that will constrain any budget
redirection:
debt burden relief is a high priority,
rural unemployment is a major issue to be addressed, and
many social and physical infrastructure needs are pressing.
176. Fiscal decentralization certainly would address the last two items on this list. The
question at hand, however, is whether state government vertical programs would do a
better job of service delivery. If so, redirection would not be a good strategy choice.
Those who push hard for more decentralization will argue that some better balance
between state and local government involvement will give a better result. There does not
appear to be any hard evidence to support one position or the other.
177. Certainly the state government has not disowned the strategy of fiscal
decentralization. The 2006-2007 budget speech underlines the need to strengthen
panchayats, but no concrete, immediate plan is offered. An obvious choice is to redirect
some of the expenditures made by state line agencies toward the gram panchayats.
Though this “offloading” strategy would impose no revenue cost on the state, it would
raise other questions such as the capacity of the local governments to absorb the
additional responsibility and whether this offloading would be accompanied by
restrictions as to how the money should be spent. Otherwise, per capita expenditures of
the local governments would be increased but there might not be an improvement in
service levels or even budgetary discretion.
The Granting of More Fiscal Discretion
178. Another part of the reform package to address fiscal decentralization could be to
increase the spending and tax discretion of local governments, particularly gram
panchayats. This may be done in two ways. First, the state government could “untie” its
grants and schemes. This has already been done for the Rs 278 crore SFC grant. This
model could be extended to the salary grant which at present is no more than a cost
reimbursement to local governments for state-determined posts and compensation. An
unconditional grant would simply award the funds to local governments on a formula
basis and allow them to decide between salary and non salary expenditures. Under this
scenario, a local government might decide to move funds from teachers (where salaries
might be deemed too high) to health professionals (where they might be deemed too
low). In fact, all state grants and schemes might be converted to an unconditional status.
This strategy is considered in the next section of this report.
179. The unconditional grant route is not without risk. This strategy will impose a cost
on the state, in terms of the direction over public investment that it will give up. It will
carry a risk in that the money might not be spent “wisely” and the result may be
48
deterioration in service levels and a loss in confidence on the part of taxpayers. If some
gram panchayats fell into a pattern of fiscal indiscipline, it would be left to the state to
cover this.
49
9. REFORM OPTIONS AND EVALUATION
180. As the government of West Bengal moves toward the goals of better service
delivery and more local self governance laid down in the Constitutional amendments, it
will need to continuously rethink the structure of its intergovernmental fiscal system.
The work of the Third State Finance Commission will be an appropriate opportunity to
begin such a rethinking. In this regard, it is important to note that reform choices that
would lead to a functioning third tier of government in West Bengal are not being held
hostage by central laws and regulations. To the contrary, the Constitution strongly
endorses fiscal decentralization. The wherewithal to implement changes in the system is
clearly in the hands of the state government.
181. In the discussion below, we suggest and evaluate a number of reform directions
that are in step with the intent in the Constitutional amendments. Some are consistent
with ideas that have been offered by the first two State Finance Commissions, though in
each case we take a different approach than did the SFC. Some are more far-reaching,
and would require major structural changes. However, bringing some less feasible
options to the table might be a useful way to enrich the debate about fiscal
decentralization in West Bengal. Some other reform options hold promise but are not
fully developed here because they involve changes in the method of financing urban local
governments, which is a subject that this paper does not directly address.
182. We evaluate three general directions for reform: (a) a change in the structure of
local government, (b) enhanced expenditure assignment and fiscal autonomy for gram
panchayats, and (c) a revamped system of state government transfers.
Decide on an Optimal Tier for Local Self Government
183. The Constitutional amendments call for a third tier of government where voters
can have both a voice in the decision about the quality of local public services that they
will receive, and a responsibility for directly financing some of these services. While the
intent is to involve the PRI as autonomous local governments there is no pronouncement
in the amendments about the balance among the three tiers of PRI in terms of the degree
of fiscal autonomy that they should be given. The costs and benefits of the various
options should be weighed. The state government may retain the present hierarchical
arrangement, or move the focus to gram panchayats (or to either of the other two levels).
The principal instruments for shifting the emphasis from one level to the other are
revenue assignment, expenditure assignment, and the distribution of intergovernmental
transfers. All of these can be changed at the discretion of the state government. So, it
50
falls to the state government to decide whether and how the roles and relative importance
of the three levels of PRI ought to be changed.
184. The right structure of local government in West Bengal depends on the goals that
government most wants to accomplish. In terms of capturing the economic efficiency
benefits of decentralization (which requires moving government decisions closer to the
people), the gram panchayat is the best candidate for autonomous local government. It is
small enough to force elected leaders to pay attention to the preferences of voters, and
this relatively small size gives the local population the sense that their vote will matter.
In West Bengal, the average population size of the gram panchayat is 14,254 versus
181,000 for blocks and 4.4 million for districts. The gram panchayats in West Bengal are
about three times larger than the all-India average, and are not small by world standards
for local governments (See Box 4). Unlike the case in many Indian States, the gram
panchayats in West Bengal may be small enough to move government close to the people
but large enough to avoid some of the diseconomies of small size, and the efficiency
losses due to spillover effects.
Box 4
Gram Panchayats in India
The average population size of a gram panchayat in West Bengal is not small,
either relative to that in other states in India or relative to many other countries. See the
comparison below.
Average Population of Total Population
Local Governments (in millions)a
All India (gram pachayats) 4,386 1028.6
West Bengal State 14,254 80.2
Karnataka 8,872 52.8
Kerala 26,793 31.8
United Kingdom 126,128 59.7
Denmark 18,760 5.4
Poland 15,561 38.5
Finland 10,870 5.2
Spain 4,930 43.1
Hungary 3,242 10.1
a
Source: Non-India data from Fox and Gurley (2006). India results are author estimates based on data
from various services.
51
185. An argument that favors the choice of the gram panchayat as the best candidate
for local government is that it is a longstanding choice of many analysts and decision
makers. In fact, under the present system, the gram panchayats already have some fiscal
independence. Budgets are approved by the elected local councils, rather than by a
higher level government, and this is perhaps the most essential element of fiscal
autonomy. GPs have some budget discretion on the expenditure side, but they do not
have the power to hire, fire, or determine the compensation rates of their employees.
Moreover, they are saddled with a significant number of budget mandates in that most of
the intergovernmental transfers that they receive are conditional upon a particular use of
the funds. Gram panchayats receive little by way of untied grants43
. The gram
panchayats also have some independent powers to raise revenues through taxation, as
well as through user charges and fees. However, gram panchayats do not have discretion
in setting the tax rate or determining the tax base. Such discretion will be important if the
GP is to become a focal point for local self government. Districts and blocks are saddled
with the same restrictions on the expenditure side of their budgets, and neither has taxing
power. Some have argued that districts and blocks behave more like spending agents of
the state than like autonomous local governments.
186. On balance, a strong case can be made on efficiency grounds for the gram
panchayats as the principal unit of local government. However, economic efficiency is
not the only criteria that may be used in choosing an optimal size local government, and
other choice criteria may point to advantages of emphasizing the block or district level as
an autonomous local government. Four advantages of larger local governments would
seem particularly important.
187. First, both districts and blocks are large enough to capture the cost savings from
economies of scale in the delivery of services. For many public services, gram
panchayats may not be large enough to take advantage of size economies. This is
especially the case for services that require large capital costs, because there is not an
adequate population over which to spread these costs. To allow delivery of such services
by gram panchayats, in such cases, might be to invite cost increases, and perhaps a lower
quality of public services. However, the methods of service delivery for many types of
rural service (e.g., water supply, sanitation) do not require the level of capital investment
that is the case in urban areas.44
Unfortunately, there is little evidence on size economies
in the delivery of local public services in rural areas in LDCs and the results of studies for
industrialized countries likely are not relevant45
. Still, it seems reasonable to believe that
consolidation to the block level would eliminate some duplication and reduce
administrative costs.
43
The general purpose grants that they receive -- untied grants and State Finance Commission grants --
accounted for less than one percent of their revenue in the years studied for this report. 44
For a discussion of this aspect of service delivery by rural local governments in South Africa, see
Schroeder (2003). 45
For a review of the available evidence, see Fox and Gurley (2006).
52
188. Second, block and district governments have a larger benefit zone and therefore
can internalize spillover effects from certain public services better than can gram
panchayats. Examples are environmental and transportation services. This suggests that
such services should be delivered by districts, blocks, or by the State government. There
is an alternative to shifting assignment of such functions to a higher level government. A
conditional grant could be given to gram panchayats to stimulate their spending on
functions with spillover benefits. However, this can be an expensive way to deal with the
externality issue. Moreover, there is the issue of estimating the expenditure response of a
local government to a particular level of conditional grant funding, and the high cost of
effectively monitoring local compliance with the conditions. So, for a number of
important services, delivery by gram panchayats is not justified, even on efficiency
grounds.
189. Third, many gram panchayats have not shown themselves to have good capacity
to deliver services. Districts and blocks are larger, and likely could offer more
specialization in work assignments, and might be able to recruit a more skilled
managerial staff.
190. Fourth, there are 3,354 gram panchayats in West Bengal, and this is an unwieldy
number to control from the state government level. Even in a decentralized government
system, there is need for such controls. The necessary controls might involve, for
example, audit, enforced accounting standards, civil service rules, grant distribution
formulae, following up on compliance with mandates and tax limitations, etc. The 341
block level governments, or the present hierarchical arrangement, would seem a better
choice for a workable structure of local government, when the span of control issue is
considered.
191. There is no easy answer to this question. There has long been a debate about the
optimal size government, and the debate has not led to the conclusion that any one
population size is “best”. The optimal size government depends on what objective one
wants to emphasize. If the spirit of the Constitutional amendments is read as calling for
more emphasis on the economic efficiency objective of fiscal decentralization, the case is
strong to upgrade the role of the gram panchayat and make it the primary unit of
autonomous local government.
192. If strengthening the gram panchayat as an autonomous local government turns out
to be the best choice for West Bengal, should the block and the district tiers of
government take on a different role? There are two general approaches that might be
taken in response to this question.
193. A first approach would be to create a unitary system with two levels (state
government and gram panchayat) and formally designate districts and blocks as
deconcentrated arms of the State administration. Under this scheme, gram panchayats
would be the only sub-state government unit that represented local voters.
53
194. In some ways, this would not involve much change. Districts and blocks already
act like agents of the state. On the other hand, if this is to be a restructuring, it suggests
major changes in revenue and expenditure assignment, and in the degree of autonomy
assigned to gram panchayats46
. In particular, responsibility for some of the expenditures
presently made by blocks and districts (about two-thirds of total PRI expenditures) would
be either shifted downward to the gram panchayats, or in the case of services
characterized by significant scale economies or externalities, would become
deconcentrated state expenditures (vertical programs). The net impact of this approach to
decentralization would be to heighten the fiscal importance of the gram panchayats
relative to blocks and districts. In the case of services presently administered by blocks
and districts, there is not much autonomy in any case, so this solution would likely lead to
a net gain in welfare. In order to move toward this approach, the expenditure mapping
presently in place (Government of West Bengal, 2005) would need to be redone with an
increased emphasis on the gram panchayat level.
195. The main advantage of this approach is that it clarifies which PRIs are local
governments and which are state and central government implementing agents. With the
roles clarified, the stage would be set for an assignment of expenditure responsibility that
the state government felt would best match its decentralization objectives.
196. A second approach is to leave things as they are, with three levels of PRI, and to
assign functions to the up-levels according to factors such as service delivery capability,
economies of scale and the possibility of spillover effects. Certainly this would be the
least disruptive approach, and the expenditure mapping could be redone to enhance
economic efficiency by assigning more “local benefit” functions to gram panchayats. The
gram panchayats could be given increased autonomy in the areas of both taxation and
expenditure delivery, much as proposed in the case of the unitary program above.
197. The main differences in these two approaches are that in the case of the present
structure, gram panchayats will likely end up with less responsibility than under the
unitary regime, and that a hierarchical arrangement among gram, block and district
panchayats will remain in place. In short, there will continue to be less gram panchayat
participation in service delivery. Another drawback is leaving in place the likelihood of
overlapping service responsibility among three levels of PRI.
Expenditure Assignment
198. Expenditure assignment should be a priority reform concern for the State
government and for the State Finance Commission. If gram panchayats are to play a
meaningful role as local self governments, their expenditure responsibilities must be
upgraded. In 2005, GP expenditures accounted for about one-third of the PRI total, about
Rs 138 per capita ($US 3.13), and about 5.3 percent of all government expenditures. It is
not likely that gram panchayats will deliver services that will matter greatly to the local
quality of life when the amounts are so small. Nor is it likely that voters will get deeply
involved in budget decisions when the amounts involved are so small. The situation is
46
It also may suggest changes in the political structure that are beyond the reach of the state government.
54
not much better for the other levels of PRI. Districts account for only about 7 percent of
government spending in the state, and blocks for only about 3.6 percent. All together, the
PRI manage about 17 percent of state and local government spending in West Bengal.
199. To move toward the type of fiscal decentralization envisioned in the Constitution,
it will be necessary for the State government to ratchet up the spending of gram
panchayats (or all PRIs) and to involve them more heavily in delivering services that
matter deeply to their constituent populations. This might be done with a strategy that
involves the state government taking three types of action (a) the assignment of increased
expenditure responsibility and fiscal autonomy to the local governments, (b) assistance in
developing an enhanced capacity of local governments to deliver these services, and (c)
provision for higher levels of funding to enable delivery of these assigned
responsibilities. The state government might look to the SFC for leadership in
developing a strategy to achieve this objective.
200. Clarify and Upgrade Expenditure Responsibility. There is need for more clarity
in the assignment of expenditure responsibilities. In theory, this is not an issue among
local governments in West Bengal, because the state has done an extensive expenditure
mapping (Government of West Bengal, 2005) that defines the sub-functional
responsibilities of each level of PRI. The mapping contains a detailed and clear
statement of how the state government believes that expenditure responsibilities ought to
be assigned among the three tiers of PRI. As is so often the case, however, practice
departs from theory and the actual division of responsibilities has lead to confusion and
overlap.
201. The even bigger problem in West Bengal is the murkiness in the division of
responsibilities between the state government and the local governments47
. The
Constitution defines a list of 29 objects of expenditure that may be either assigned to
local governments or may be concurrent responsibilities with the State government. The
State of West Bengal has chosen the concurrent route for all 29 functions, and this has led
to three problems: (a) too limited an involvement of PRIs in service delivery, (b) some
confusion as to whether the state or the local governments are responsible for certain
services, and (c) an inability at the State level to rationally determine the amount of
resources necessary for PRI to deliver assigned services at an adequate level.
202. The World Bank (2005) offers an interesting approach to resolving the
expenditure assignment issue. The argument is that for each expenditure sector, there are
five activities that are (better or worse) candidates for decentralization: policy and
standards, planning, asset creation, operation, and monitoring and evaluation. In the case
of schools, water, sanitation, and employment generation, they argue for assignment of
operational responsibility to the gram panchayat level. This would include a transfer of
functionaries to the control of the local government in order to increase the accountability
of local officials.
47
This vagueness in the assignment of expenditure responsibility is a problem throughout India. For a
discussion, see Subrahmanyam and Annamaloi (2004, p. 223-224).
55
203. Neither of the first two SFCs in West Bengal have taken on the issue of
expenditure assignment in a comprehensive way. Both Commissions have argued to
increase the share of untied grants in the State budget. While this would give PRIs more
resources, and more autonomy, it would (presumably) leave in tact their expenditure
responsibilities and their discretion in making decisions about how these funds should be
used. The third SFC might consider addressing this larger issue directly, by considering
the need to upgrade the expenditure assignments of local governments. The first step
would be to determine for each of the 29 objects of expenditures, which sub-functions or
sub-activities will be assigned exclusively to rural local governments. Then it would
redo the mapping exercise to assign subfunctions of these expenditure categories to each
of the three levels of PRI. Somehow, delivery would need to be monitored in order to
ensure proper involvement by the PRI and to avoid duplication.
204. The second step would be to clarify the assignments that go with this
classification of expenditure responsibilities by sub-activity. For example, if The World
Bank (2005) model were to be followed, school operations and teachers would be shifted
to gram panchayats, and water supply and sanitation in rural areas would become a gram
panchayat function. Such changes would significantly upgrade the place of rural local
governments in public service delivery.
205. The third step would be to define a target level of spending for GPs that would be
great enough to stimulate voter interest and involvement in the process of budget making,
and at the same time would enhance the quality of services available. The target level of
spending would reflect both the minimum standards of service desired, and the fiscal
limitations faced by the state government.
206. Fiscal Autonomy. The goal of enhanced local self governance requires more than
just assigning new expenditure responsibilities to local governments. So long as the State
of West Bengal (and the Central Government) continue to dictate how the money should
be spent, local governments will not have the ability to adjust service delivery to match
citizen preferences. As noted above, if the local population thinks of their elected local
government as being little more than a spending agent of some higher level government,
they will not hold their elected local council responsible for the quality of services
delivered. Local governments must be given more autonomy to make budget decisions if
local government officials are to be accountable to their constituents. Unless this
happens, local voters/constituents will have little incentive to be involved.
207. There are a number of areas where increased autonomy might be given. First,
those officials involved in the delivery of local public services could be assigned as local
government employees. Their hiring, firing and the determination of their compensation
levels can be made local government decisions. The power of this approach is illustrated
by the case of education in Madhya Pradesh where teacher control was shifted to the
local governments (The World Bank, 2006). The PRIs were given the power to hire and
fire teachers (but not to determine wage rates). The result has been a remarkable
improvement in the rate of teacher absenteeism and an expansion in the coverage of the
school system.
56
208. Second, conditionalities could be removed on the distribution of state grants to
local governments. These could become untied funds with expenditure made at the
discretion of the local governments. Whenever this reform option is proposed (in nearly
all countries), the criticism raised is that government funds will be diverted from what the
upper levels of government consider the highest priority, and that corruption and waste
will necessarily follow. The response to these observations is, first, a diversion of funds
to projects with a heavier orientation toward local benefits will almost certainly happen.
As local politicians and local voters learn about accountability, the budget will begin to
show more “local choice” investments, and a better implementation of these projects.
This is the essence of the fiscal decentralization story. In fact, the shift in the package of
service delivered would be evidence that the fiscal decentralization strategy is working.
209. Unfortunately the corruption and waste story is also probably a valid one. Gram
panchayat officials are not skilled in service delivery and local management. The books
of account may not be kept well, staff may be untrained and there is probably little by
way of skill in the area of tax administration. As local officials learn their new
responsibilities, money may not be spent as wisely as if it were in more experienced
hands. On the question of whether decentralization leads to more corruption, there is not
yet a general agreement. In the scholarly research, Brueckner (1999) finds a direct
relationship, but Fisman and Gatti (1999) find the opposite. They argue that corruption
under decentralization can be lower if both revenue raising powers and expenditure
responsibility are decentralized.
210. The spirit of the Constitutional amendments and of the recommendations of the
State Finance Commissions is that of local self government. The implication here is that,
in so far as is possible, grants should be unconditional and expenditure mandates should
be removed. Therefore, the reform goal might be to move towards elimination of
conditions associated with State government grants to GPs, in favor of unconditional
grants with a defined vertical share48
. Such a reform would be consistent with the
recommendations of both the First and the Second State Finance Commissions in West
Bengal. Both recommended a dramatic increase in untied grants to local governments.
The state has not followed this recommendation and the present level of untied state
grants is equivalent in amount to less than 2 percent of state government taxes.
211. Third, local governments could be given more independence in choosing their
level of taxation, rather than have the tax rate and the tax base determined at the state
level. This is an important part of the reform package. Local government officials could
be made more accountable to their constituents if services were financed at least partially
by taxation. In this regard, it would be important that the local government has the power
to set at least the rate of a local tax. The state could set a minimum tax rate for GPs, but
could leave open the ceiling rate. Under this arrangement, the local council will be more
responsible to the local population for the quality of the service delivered.
212. Enhanced Capacity at the GP Level. It would be self-defeating to assign new
responsibilities to GPs if they could not deliver these services in an effective way. And,
48
The vertical sharing dimension of such a grant system is discussed in detail in Bahl and Wallace (2007).
57
every indication is that there are serious capacity problems at the local level. But not to
assign more responsibility to the GP because their capacity is weak is to create a self-
fulfilling prophesy. If GPs are not put on a learning curve as regards service delivery,
they will never develop the capacity to absorb more responsibility.
213. The state could take a number of steps to improve the service delivery capacity of
gram panchayats. Though this subject is beyond the scope of this work, a stylistic listing
of some possibilities that have been offered elsewhere are below:
In many gram panchayats, there is a need to increase the size of the gram
panchayat staff and upgrade the quality of the staff.
Panchayat secretaries often are not trained in the rules, procedures and
statutory provisions of the panchayats. Short training courses should be
provided (Subrahmanyam and Annamalai, 2004).
GPs should be given authority to buy services from the private sector or
contract with higher level governments in cases where they do not have
adequate capacity to deliver services. Resource pooling by several gram
panchayats for a specialized service might be considered. (World Bank,
2005).
Automation in the case of more advanced GPs, should be encouraged, at
least for record keeping purposes.
Provision should be made for annual audits of local government books of
account by external parties.
Re-assessment for purposes of local property tax should be made by
external valuers.
214. The first step to be taken here is to assess the need for training and staffing at the
GP level. The goal would be to assess the capacity of GPs to assume new responsibilities
and to design the technical assistance necessary to put this capacity in place. Second, a
transition plan should be put in place whereby the new expenditure responsibility of gram
panchayats would be passed to them when they are deemed ready. Until that time the
services, and the finances that go with these services, would be delivered by blocks and
districts. Third, a system to monitor and evaluate the performance of gram panchayats
would be put in place.
Increased Own Source Revenue
215. Much has been made of the need for gram panchayats to increase their own
source revenues (Rao and Singh, 2001). Success in this area will come from three kinds
of initiative. First, some potentially productive tax bases must be assigned to gram
58
panchayats, along with the power to set the tax rate49
. There are many options for rural
local government taxation that either are now used in West Bengal or are used in other
states and might be considered in West Bengal. These include:
Land and building tax. This is the mainstay of local taxation in rural India
and in most countries. The needs in West Bengal are to remove any rate
ceiling in favor of a policy of local governments setting their own rates, and
state support for an upgrading of property tax administration.
Property transfer tax. Local governments may levy a tax against the
consideration in the transfer of property. In many countries around the world,
this is an important source of local revenue. However, administration of a
property transfer tax is beyond the reach of rural local governments. A better
route is for the local government to be allowed to set a sur-rate against the
state government stamp duty 50 . A sur-rate is preferable to a share of
collections, because it forces the local government to make a tax choice and to
be accountable to taxpayers for the expenditure of that money.
Entertainment tax. In West Bengal, the entertainment tax is actually an
intergovernmental transfer to local governments. The tax rate and base, and a
sharing rate on collections with local governments, is set by the state
government. Since most/all revenues are shared locally, the state government
has little incentive for vigorous collection efforts. The entertainment tax
might be a candidate for local taxation (or at least local rate setting). It is a
local tax in some Indian states, e.g., Rajasthan, Karnataka, and MP (though
rates are still set by the state).
Professions tax. A tax on professions, trades, and employment is levied in
several states. In at least Rajasthan, Assam, Punjab, and Bihar the tax is
administered by the panchayats. This is a kind of rudimentary income tax that
reaches those in rural areas with a greater ability to pay. Administration of the
tax is difficult, especially for rural local governments.
Land cess surcharge. State governments levy a cess against land for a
variety of purposes. In some states, panchayats are allowed to levy a
surcharge against these cesses, with the revenues returned on an origin basis.
This approach to taxation is consistent with the accountability maxim of
decentralization, so long as the panchayat sets the sur-rate.
216. Second, the capacity of states to better administer local taxes should be upgraded
so that tax effort might be increased. With respect to the property tax, a number of steps
should be taken:
The state could develop a circular of guidelines on assessment and
collections.
49
Analogous reforms should be put in place for user charges, licenses, and fees. 50
See for example, the practice in Kerala as described in Subrahmanyam and Annamalai, 2004.
59
The creation of an independent valuation authority should be
considered. It should be staffed with trained values.
GP Secretaries will be trained so as to develop basic valuation
skills.
Tax mapping, showing all properties and land uses in the local
government area, should be completed.
217. The ability to improve the collection rate on other tax and non tax sources should
be strengthened. In this regard, the staff support of the panchayat secretary could be
upgraded. The collection of non tax revenues may have been pushed aside in order to
comply with the ever increasing compliance issues associated with the schemes. With
respect to the assignment of other taxes and more efficient collection of these, a number
of possibilities for improvement have been suggested:
More complete tax rolls for each levy (e.g., professions and trade
tax, entertainment tax, land cess) should be compiled and annually
updated.
Enforcement should be more vigorous and proper sanctions should
be applied. The names of delinquents might be publicly posted.
Where permitted, surcharges on state government levies should be
imposed. This allows the GP to bypass the entire administrative
process.
Where there is an organized sector, and where the GP is permitted
to levy a tax on professions and trades, tax deductions at source
should be allowed, as is done in Kerala and A.P.
218. Third, rural local governments might be given some significant incentives to
upgrade their level of tax effort. This might be done in three ways. One is to put no
restrictions on how the money can be spent. A second is to increase the taxing powers of
gram panchayats either by assigning them additional taxes or removing any tax rate
limits. The third approach is to reward tax effort through the system of transfers with a
kind of incentive grant. The programme would work by matching local collections (or
increases in local collections) with some amount of state grants. Such an approach has
been tried in both Tamil Nadu and Goa (Subrahmanyam, 2004).
Increased State Grants: Determining the Vertical Share
219. Even if gram panchayats are assigned more responsibility, and even if this
capacity is upgraded, it will come to naught unless adequate financing is provided.
Increased own source revenues are an important element of the reform, but will at best be
60
only a small part of the financing needs. If local governments are to provide a
meaningful level of services, a significant increase in state grants must be provided.
Moreover, the local governments must have some discretion over how these funds are
spent; otherwise they will tilt toward being spending agents of the state rather than
autonomous local governments.
220. Both the First and Second State Finance Commissions recommended an increase
in state funding of local governments. The recommendation of the first SFC was to
institute a program for rural and urban local bodies with a vertical share equivalent to 16
percent of State government taxes. At the time of this recommendation, and at present,
there is no specified percent of State taxes that is guaranteed for revenue sharing with
local governments. The vertical share is determined annually on an ad hoc basis. It can
vary from year to year depending on the financial circumstances of the State
Government.
221. A second dimension of grant policy is whether the additional funds will be
conditional upon use for certain types of expenditures, as might be prescribed by the
State, or whether they will be untied (unconditional) funds. The SFCs have
recommended untied grants as a way of giving local governments more discretion in
establishing spending priorities. The view here is that the general spirit of the State
Finance Commission recommendations is correct. We propose and evaluate a reform
package that is similar in intent to that of the SFC, but is structurally different.
222. We begin this analysis by establishing a target level of gram panchayat
expenditures to be financed. Properly calculated, the target amount will depend on the
expenditure assignment decided upon. It also will depend upon the minimum level of
expenditures that is set by the state. As a matter of good policy, it is essential to assign
expenditure responsibility and service level targets before determining the vertical share
of transfers to be allocated51
. This follows the well-traveled principle that “finance
follows function”. The failure to follow this maxim might have been a problem with the
model developed by the earlier State Finance Commissions52
. The 16 percent proposal
for the vertical share of all local governments appears to have been more of an arbitrary
decision than one based on objective analysis.
223. It is not in the scope of work here to develop a new set of expenditure
assignments or to develop a set of minimum expenditure requirements53
. Rather, we
simply assume a target level of per capita local government expenditures in order to
demonstrate the impact of a new state grant system. We set an illustrative target for GP
minimum per capita expenditures of Rs 414 (three times the current average level of per
capita expenditures) and alternatively at Rs 276 (two times the current average level).
We make the assumption that all of this increment above the present level of per capita
51
For a discussion of the rationale for this sequencing, see Bahl and Martinez-Vazquez (2006). 52
To be fair, it almost certainly was the case that neither SFC had the resources or the data to do this
expenditure requirements analysis. 53
Expenditure assignments for various tiers of PRI under the present system are laid out in detail in
Government of West Bengal (2005).
61
expenditures (Rs 138) would come from state government grants and transfers to gram
panchayats. We further assume there will be no reductions in central transfers. Note that
this analysis is portable to different expenditure targets.
224. In this first example, we propose a minimum expenditure approach to revenue
sharing where the vertical share for gram panchayats is based on a guaranteed minimum
level of spending for each GP. This might be viewed as a transition to a more traditional
formula-based system, as is described below54
. The revenue sharing rules for the
transition system -- which do not feature a formula based on socio-economic variables --
are as follows:
Each gram panchayat will be brought up to a minimum per capita
expenditure of Rs 414 (or, alternatively Rs 276 under the more
modest target).
Any gram panchayat with a present spending level that is higher
than Rs 414 (Rs 276) will be held harmless at their present level.
However, those GPs will not receive additional state grants.
225. The calculations for this hypothetical level of local government expenditures and
its financing are shown in Table 43. Reaching the target minimum expenditure level of
Rs 414 per capita would require that GP expenditures be higher than at present by Rs
15.8 billion (row 3 of column 1). The result is that the transfer from the state government
would rise to over Rs 17 billion (row 5), or to an amount equivalent to about 10.6 percent
of state government own source revenue collections (row 7). If the target was a less
ambitious doubling of minimum per capita expenditures by GP, the vertical share would
be 5.7 percent of state government tax collections (see the calculations in Column (2) of
Table 43). In either case, the amounts are well above the present level (there was no
distribution in 2005), but well below the 16 percent of state taxes recommended by the
State Finance Commission.
226. Had this program been enacted at the higher level in 2005, the majority of GPs
would have seen an increase in per capita total state grants of over 90 percent (Table 44).
As may be seen in Table 25, 591 of the 3,074 gram panchayats in the sample would
realize an increase in state aid of more than 95 percent. If the less ambitious minimum
expenditure were used, most GPs would witness an increase in per capita state grants of
80 percent or more (See Box 5).
54
There are several reasons why a government might want to transition to a new formula system rather than
do it in one “reform” year. The transition period gives the opportunity to set up a system of hold-harmless
where no local government receives less than under the previous system, and none receives too large an
increment. This minimizes the shock associated with the new system and gives some time for preparing to
absorb the phase-in of the new system where the distributive shares will change.
62
Box 5
Explanatory Note on Simulation Exercise
This simulation is to show the impact of an unconditional state grant program,
where the vertical share is equal to the target amounts described in Table 24, and the
horizontal shares are according to the formula described. The data used in the simulation
are for the year 2005.
The counterfactual in this analysis is the amount of state government transfers
distributed in 2005 as unconditional grants. Note that the state did not distribute the State
Finance Commission grant in 2005, hence the comparison here is against a very low
base.1
We can say, however, that the target level in this analysis implies a vertical share
of Rs 1729 crores in 2005 under the high scenario, and Rs 938 crores under the low
scenario. The 2005 amounts for all state grants used in this baseline were equivalent to
Rs 147 cr in this baseline. Therefore, we are evaluating an increase of 1582 crore (791 in
lower case) for gram panchayats in the total grant pool.
1The State did distribute the Finance commission grants in 2006, with a vertical share of Cr 278, but data
describing the actual distributions are outside the data base available here, which is for the years 2003-
2005.
227. With respect to the horizontal distribution, we ask if the proposed minimum
expenditure system is more or less equalizing than the present system. The answer
depends on how one defines equalization. Certainly the expenditures of most gram
panchayats are dramatically increased under this approach. Moreover, the gap between
the high spending GPs (who are held harmless under this scenario) and the low spending
GPs is narrowed considerably. Many would see this as a favorable outcome and one that
is consistent wit the idea of reducing disparities.
228. On the other hand, equalization might also be defined as developing a system that
compensates GPs with low fiscal capacity or with high expenditure needs, so that they
are on an equal footing with better situated local governments. Proponents of this vision
of equalization might ask, for example, if the relative revenue shares of the more
backward GPs greater or smaller than under the present system? If the latter, the
minimum expenditure approach would not pass this version of the equalization test.
229. To compare the distribution effects of the present system with the minimum
expenditure system, we have computed the simple correlations shown in Table 45. In
column 1, we show the correlation between selected variables and the per capita
distribution of the 2005 state aid system, and in column two with the proposed minimum
expenditure system. If the percent of SC/ST population is the barometer for poverty, the
63
proposed minimum expenditure system is no more equalizing than the present system.
Neither shows a significant correlation with the percent of SC/ST population.
230. What we may conclude here is that a minimum guarantee program will eliminate
low spending gram panchayats. However, this system will not favor poorer gram
panchayats, over those that are less poor, with larger per capita allocations. Another
disadvantage of the minimum expenditure approach is that it gives local governments no
incentive for revenue mobilization.
A Formula Approach.
Increased, Untied Grants: A Formula Distribution
231. The analysis above describes a program with a guaranteed minimum allocation
and with a hold harmless provision to protect any gram panchayat from losing revenues
at the time of the introduction of the new system. Otherwise, it has not built-in features
that would enable it to address specific issues, such as equalization or allocating funds
more heavily to gram panchayats with a particular population make-up.
232. An alternative approach or perhaps the second step in the transition would be to
adopt a new formula-based system but with the new vertical share maintained. In this
section, we analyze the impact of one version of a formula distribution with the vertical
share based on an entitlement of 5.7 percent of state government own source revenues.
The question becomes one of settling on a formula that fits the goals of the state towards
its rural local governments. The SFCs have suggested an objective formula, which is
reported to have been mostly adopted by the State government for distribution under its
present grant program. There are two ways in which the formula distribution among GPs
might be improved over this system. First, it could be made simpler. Second, the
revenue mobilization incentive might be made more effective.
233. There is no single “best” formula for distribution of a grant. The allocation
formula chosen should be driven by the goals that the State government most wants to
accomplish with its system of grants. There would seem to be three likely candidates for
the appropriate goals:55
ensuring a resource flow that would allow an adequate (minimum) level of
local public services to be delivered,
equalizing for differences in expenditure needs and fiscal capacity, and
providing an incentive for revenue mobilization.
Moreover, the choice of a formula should also be influenced by acceptability at the local
level, and by transparency. Both of these considerations would seem to argue for a less
complex formula. Finally, any new formula must be politically acceptable, a
consideration that will significantly constrain the policy choices.
55
See also the discussion in Singh and Srnivasan (2006, p. 349).
64
234. Note that some of the goals outlined above reinforce one another, but others may
conflict. For example, the goals of equalization and ensuring a minimum level of
services in all GP areas probably are reinforcing. The revenue mobilization goal,
however, may draw resources away from poorer jurisdictions and reduce the equalization
emphasis of the formula. Likewise, the minimum service level objective may conflict
with the goal of formula simplicity. Finally, any proposed change will produce winners
and losers, even if only in relative terms, and therefore will have political opponents.
235. Deciding on a new formula is no easy matter. In the end, it will come down to
weighing the balance between equalization and investing in GPs with greater immediate
growth potential, and to the outcome of a political bargaining process that always plays a
major role in formula determination. The State should begin with a clear statement of the
objectives to be accomplished, and design a formula that best matches these objectives.
Then it should move on to the art of political compromise. The casual search for
formula variables based on data availability and guesswork about what they might mean,
should be avoided.
236. There are many possible formula distributions that might be considered. The
proposal we offer here is for purposes of illustration, but it does suggest how a new
system might be developed. Developing such a formula involves four considerations.
237. Eligibility. If the GP is chosen as the primary unit of autonomous local
government, State government grants in aid could be allocated directly to gram
panchayats, bypassing the districts and blocks. Under this method, districts and blocks
would be part of the vertical programs of the state. They might have responsibility for
monitoring the performance of the gram panchayats, but they would not participate in the
grant system as local governments. This is the approach we take in this example.
238. Vertical Sharing Pool. The vertical sharing pool would be established as 5.7
percent of State government own source revenues. Of this total amount of state grants,
approximately 75 percent would be distributed by general formula, and 25 percent would
be distributed from an incentive fund for local revenue mobilization. The state would
commit to full distribution of the entitlement in a timely matter. All funds would be
untied.
239. General Fund Distribution. The general fund (75 percent of the total) would be
distributed by a formula with equal weights for two variables: population size and SC/ST
population. The first element of this formula would recognize general expenditure needs
while the second would recognize the special needs of more backward GPs and would
add an equalization component. Data used to make this allocation would be drawn from
the census. This approach is transparent, easy to understand, and is objective.
240. Revenue Mobilization Component. Of the vertical share for State grants (5.7
percent of State government own source revenue), 25 percent (1.44 percent of State
government own source revenue) would be allocated to a fund that would be used as an
65
incentive to encourage revenue mobilization by local governments. The distribution of
the revenue mobilization sharing pool among GPs would be 50 percent according to the
per capita amount of tax and non tax revenue collections in the preceding year, and 50
percent according to the increase over the preceding year.
241. Ideally, the distribution would be based on an index of tax effort, i.e., a measure
that makes allowance for one GP having a greater capacity to tax than another. (Such an
index is estimated and reported above.) But these indexes are difficult to use and to
defend, so we resort here to a shorthand measure. The inclusion of per capita tax
collections rewards those who make a greater effort, irrespective of their capacity to tax,
and provides an incentive to maintain that effort. The inclusion of the increase in per
capita tax collections rewards those GPs who respond to the incentive by raising their
efforts at revenue mobilization.
242. Minimum Guarantee. A constraint would be placed on the distribution of this
revenue-sharing pool. No gram panchayat would receive an amount less than Rs 276 per
capita. This level would be taken to represent the guaranteed minimum level of support
for per capita expenditures. If the minimum was not attained after distribution of the
general pool and revenue mobilization components, the general fund pool would be
reduced to ensure that the minimum per capita expenditure supported reached Rs 276.
The maximum that any GP could receive is Rs 1500.
243. Simulation Results. We have simulated the distribution of the full system--
general grant, revenue mobilization, and minimum expenditure for GPs --, with the
results shown in Tables 46-48.
244. An important feature of this proposal is that there are no “losers” among the gram
panchayats. No local government would receive less than it presently receives because
the total vertical share is ratcheted up from 1 percent to 5.7 percent of State government
own source revenue, and because a minimum guarantee is put in place.
245. The “fully phased in system” would give a different distributional result than the
minimum expenditure approach simulated above in that it includes the general
component, the revenue mobilization component and the minimum expenditure
component. We have simulated the distribution according to this formula system across
all gram panchayats for which we have data. Less than 2 percent of the GPs would
receive an increase in per capita grants of less than 10 percent. About 80 percent of the
GPs will see an increase in per capita state grants of 50 percent or more by comparison
with the current system (Table 46). The per capita distribution of the general fund
component is similar for small and large GPs (Table 47) but the revenue mobilization
component provides more grants per capita to the smaller GPs. The implication here, that
smaller gram panchayats have been more active in revenue mobilization, is surprising.
246. This formula system has several features that fit with the goal of stimulating the
involvement of GPs in service delivery. First, the base of the proposed state transfer
system would be more elastic, i.e., the vertical pool will automatically increase with tax
66
collections. The growth in tax collections in recent years in West Bengal has been
robust. Second, the new system would be relatively simple, with only a two factor
formula and a guaranteed minimum expenditure level. It would give government the
capacity to periodically upgrade the minimum per capita expenditure guarantee, or to
increase the weight of the equalization component, in a simple way.
247. Third, the proposed system would have an equalizing component. We estimate
the relationship between selected measures of equalization and the new grant and report
the results in Table 48. The formula system does not favor gram panchayats with a
higher proportion of SC/ST population, hence is no more equalizing than the present
system. Both, however, favor less populated GPs. The literacy rate, which may be an
indicator of relative wealth in a GP, is positively correlated with state government
intergovernmental transfers under the current system but is unrelated under the proposed
system. What we can conclude here is that it will probably take a significant movement
away from the revenue mobilization incentive to develop a formula with a strong
equalization component.
248. Finally, the proposed system has a revenue stimulation package that is large
enough to provide a significant incentive. Based on the results of this simulation we
show a wide disparity in the amounts received for achieving higher levels of revenue
mobilization. On average, a Rs 10 per capita higher level of revenue mobilization is
associated with a Rs 3 higher level of grants received.
249. There are major disadvantages to this proposal. First, about 3 percent of gram
panchayats do not gain from this scheme compared with the present system, and there are
wide variations in the increases received by various GPs. One would need to understand
this pattern and any anomalies that might be hidden here. Second, we have all of the
necessary data for only 2,098 gram panchayats, but any new grant system would have to
cover all 3,324 of the gram panchayats.56
Third, the census data that populate this
formula are available only every ten years. Fourth, there is a question about whether
some gram panchayats could absorb the very large increases in resources that they would
see under this new system. Fifth, exclusion of blocks and districts from this program
presupposes that the expenditure assignments of GPs are most in need of increased
emphasis. Finally, no parallel program for upgrading the finances of urban local bodies
is offered here.
56
SC/ST population is the most limiting factor in terms of data availability.
67
10. FINANCING THE FISCAL DECENTRALIZATION
PROGRAM
250. Where would the West Bengal State government find the additional funds to
cover such a program? Based on 2005 levels, the gap to be financed is equivalent to
about Rs 16 billion under the high scenario, and Rs 8 billion under the low scenario.
There are several possibilities for raising these additional funds, six of which are
discussed below:
251. The government could encourage (require) GPs to generate more own source
revenue. If this strategy were successful, it could reduce the amount of state grants
necessary to reach the target level of expenditures. However, the revenue potential of
this option may not be very great. The average level of GP own source revenue is about
Rs 8 per capita. If this were increased by 50 percent, the result would be an additional Rs
201 million. This would be equivalent to about 1.2 percent of the Rs 16 billion gap to be
covered under the high scenario and 2.5 percent under the lower scenario. Certainly this
does not cover a major share of the cost of this fiscal decentralization program, but it
does significantly increase the PRI contribution to rural local government financing, and
so is well in step with the goals of the constitutional amendments. A good argument can
be made that this is an essential part of the reform program.
252. A second scenario is to assign additional taxing powers to urban local bodies, and
shift a commensurate amount of the distributable grant pool from urban local
governments to gram panchayats. In 2005, urban local bodies in West Bengal received
Rs 698 million in state grants. If Rs 349 million (50 percent of this amount) was
redirected to gram panchayats, it would cover about 1.6 percent of the Rs 16 billion
required under the scenario of a threefold increase and 3 percent in the case of a twofold
increase. Urban governments might make up for this Rs 349 million revenue reduction
by increasing property and land taxes and the entertainment tax, if such powers for
making tax increases were granted to them57
. This reform is in the right policy direction
in that it moves municipalities and municipal corporations toward local self governance,
but does not go very far toward covering the necessary funding for PRI upgrading.
Moreover, it could exacerbate the financing problems of urban local bodies, which are
already acute, and would have significant political costs.58
57
The revenue potential of urban local bodies in India is significant, and they are less prominent in public
financing than is the case in many large countries. For a review of the fiscal performance and potential of
selected urban local bodies, see World Bank (2004). For a discussion about organizing a fiscal package for
urban local bodies, see Mathur (2001, Chapter 5). 58
For a discussion of urban government financial condition in India, see Mathur (2006).
68
253. The state government could increase its tax effort and earmark this additional
amount to the GPs. If the State government increased taxes by one percent of GSP, it
would raise an amount equivalent to 95 percent of the needed revenue for the high
scenario case and nearly twice that needed for the low scenario case.59
254. Certainly there is tax space in West Bengal. Own source revenues in the State
were about 4.26 percent of GSP in the 2000-2003 period, and this ratio has declined since
the 1993-1996 period (Government of India, 2004). As the Twelfth Finance Commission
reports (Government of India, 2004, p. 44), this ratio increased in most Indian states. The
policy question to be considered here is whether fiscal decentralization would be a
priority expenditure, even if tax effort were to be increased. West Bengal is one of the
most heavily indebted states, and in the 2000-2003 period, over 35 percent of revenue
receipts were allocated to interest payments.
255. Part of the necessary funding for a decentralization program could be covered by
a reallocation of expenditure responsibility and revenue receipts to the gram panchayat
level. For example, if 20 percent of district and block responsibilities were allocated to
the gram panchayats, and if the supporting revenues followed this reallocation, the
additional amounts would cover about 17 percent of the financing necessary for the high
scenario program and 33 percent for the low scenario program. There would be no net
cost to the state if this were done, but neither would there be an increase in the vertical
share of the PRI sector. This would be simply a reallocation of expenditure
responsibilities to gram panchayats. A good argument can be made that this is an
essential part of the reform program.
256. There is a fine line between the offloading of expenditure responsibilities by a
higher level government and the reassignment of expenditure responsibilities to lower
level governments. In some countries, the offloading has come without funding and is a
not too veiled shifting of the deficit.60
What we propose here is a reassignment that would
be accompanied by full funding, probably in the form of central or state transfers.
257. Funds could be directed away from state vertical programs to the untied grant
pool for gram panchayats. This would imply that gram panchayats would assume
responsibility for certain programs that are now delivered as state vertical programs. This
strategy would not amount to a net increase in funding of local governments, or a drain
on the state budget. Basically, it would be an offloading of expenditure responsibilities
from the state to the local governments. However, it would shift the locus of expenditure
responsibilities from the state government to the rural local governments, and in that
sense would be consistent with a fiscal decentralization strategy. It would differ from the
strategy outlined immediately above in that it would make the money available on an
59
This is based on data from the India Economic Survey, 2004-05 that lists West Bengal state domestic
product at Rs173,674 crore (http://indiabudget.nic.in/es2005-06/esmain.htm, Table 17). 60
An interesting case is Russia, where there was a significant “offloading” of expenditure responsibilities
by the federal government. These responsibilities included capital investment, social welfare, and price
subsidies for social goods. The funding for the offloading was to be a combination of privatization and
subnational government budgets (Martinez-Vazquez, Timofeev, and Boex, 2006, chapter 4).
69
unconditional basis and the gram panchayats could move the expenditures toward what it
considers higher priority areas.
258. In order to make an estimate of the extent to which this form of redirection would
contribute to the Rs 16 billion target, it would be necessary first to do the detailed work
on expenditure assignment. It seems clear, however, that a noticeable part of the gap
could be covered in this fashion.
259. The government might find external assistance to support the state budget during
the period of transition to a decentralized structure. There are a number of issues here.
First, all gram panchayats will not be prepared to assume all responsibilities in the reform
year. A period of training and other capacity building activities will be required. Even
when responsibilities are decentralized, there is a possibility of service breakdowns in the
“hand-off” years, therefore a contingency fund will need to be put in place. The
contingency fund and the capacity building could be candidates for external support.
260. Second, the enhancement of the own source revenue capabilities of local
governments will take time and technical assistance from the state level. It may take a
period of years before the new, higher levels of revenue can be reached. External support
might be sought to backfill during the transition.
261. Third, the redirection strategy cannot be accomplished quickly. This involves
moving some functions and functionaries from the state, district, and block levels to the
gram panchayat level. To minimize the chance of a service level breakdown, this
redirection will need to be phased in, perhaps over a five-year period. Even then, a
contingency fund may be required.
262. Fourth, it seems clear that some form of tax support will be required to finance the
increase in state transfers implied by this program. Even if the government were to
commit to this, it would take a period of years to reach the target. External financing
could cover the gap in the interim.
70
Annex
71
Annex
Table A-1
Matrix of Simple Correlation Coefficients:
Blocks a
POP SC/ST FP LR MW
Population (POP) ---
Percent of SC/ST
Population
(SC/ST)
.0762*
(2142)
---
Percent of Female
Population (FP)
.1328*
(2463)
-.2221
(2142)
---
Literacy Rate
(LR)
.1333*
(2456)
-.2052*
(2142)
-.0330
(2456)
---
Percent of
Marginal Workers
(MW)
.2425*
(2142)
0.1729*
(2142)
.0906
(2456)
-.0950
(2456)
---
Percent of
Agricultural Labor
(AL)
-0.020
(2453)
.2546*
(2139)
.0321
(2453)
-.2876*
(2453)
0.247*
(2454)
a>
* denotes significance at .05 level. The number in parenthesis is sample size.
Table A-2
Matrix of Simple Correlation Coefficients:
Districts a
POP SC/ST FP LR
Population (POP) ---
Percent of SC/ST
Population
(SC/ST)
-0.4492
17
---
Percent of Female
Population (FP)
-0.2826
17
0.3316
17
---
Literacy Rate
(LR)
0.3701
17
-0.2750
17
-0.2054
17
---
72
Per Capita GSP 0.181
15
-0.515*
15
-0.088
15
0.832*
15
a>
* denotes significance at .05 level. Number in parenthesis is sample size.
Annex Table A-3
Matrix of Simple Correlation Coefficients:
Gram Panchayats a
POP SC/ST FP LR MW
Population (POP) ---
Percent of SC/ST
Population
(SC/ST)
-0.0649*
2100
---
Percent of Female
Population (FP)
-0.1505*
2409
-0.0320
(2100)
---
Literacy Rate
(LR)
-0.1384*
2456
-1969*
(2100)
-.0330
(2456)
---
Percent of
Marginal Workers
(MW)
-0.2408*
2456
0.1658*
2100
.0906
(2456)
-.0950
(2456)
---
Percent of
Agricultural Labor
(AL)
-0.0287
2453
0.2503*
2099
.0321
(2453)
-.2876*
(2453)
0.247*
(2454)
a>
* denotes significance at .05 level. number in parenthesis is sample size.
73
Figure 1
Government Structure in West Bengal
a Percent of total population
b Not including 3 notified areas
State Government
18 Districts 6 Municipal corporations
117 Municipalitiesb
341 Samitis
3,324 Gram
Panchayats
Urban (28)a
Rural (72)a
74
Figure 2
Per Capita GDP and Per Capita PRI Expenditures:
By District
0
200
400
600
800
1000
1200
1400
1600
1800
UTTA
R D
INAJP
UR
KOCH B
IHAR
BIRBHUM
PURULI
A
MURSHID
ABAD
NORTH
TW
ENTY
FOUR P
ARGAN
AS
DAK
SHIN
DIN
AJPUR
MALD
AH
JALP
AIG
URI
BANKU
RA
SOUTH
TW
ENTY F
OUR P
ARGANAS
NAD
IA
DAR
JILI
NG
HUGLI
HAO
RA
District
Pe
r C
ap
ita
Ru
pe
e A
mo
un
t
Per Capita Gross State Product
Per Capita Expenditures
75
Figure 3
Distribution: per capita expenditure
0
100
200
300
400
500
600
700
20 60 100 140 180 220 260 300 340 380 420 460 500
per capita expenditure
freq
uen
cy
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
cu
mu
lati
ve %
Frequency cumulative %
76
Table 1
Distribution of Public Finances Among Levels of Government in West Bengal: 2005
Percent of Total
Expenditure
Percent of Own
Revenues
Percent of
Populationa
Exhibit: Per
Capita
Expenditures
State
Government
75.8 95.8 100 1218.2
Urban Local
Governments
7.89 3.7 28.0 452.5
Rural Local
Governments
16.26 0.5 72.0 138.4
Of which
Districts
45.0 41.8 155.8
Blocks 21.9 14.2 94.8
Gram
Panchayat
33.1 44.0 137.8
a Total population in Kolkata Metropolitan Area is 14.96 million in 2001; of which 88.4 percent is urban
(Administrative Report of the Municipal Affairs Department, page 2).
Sources:
Population: Census of India, available at:
http://www.wbcensus.gov.in/DataTables/02/FrameTable4_1.htm
Expenditures:
o PRI-West Bengal data base, The World Bank (See Annex A): accounts for 94 percent
(17 out of 18) of the districts, 84.4 percent (288 out of 341) of the blocks, and 87.7
percent (2,941 out of 3,354) of the gram panchayats.
o Urban – Administrative Report of Municipal Affairs Department, 2005, Government of
West Bengal.
o State – Finance Department, Government of West Bengal, Budget Publications 2006-07
(http://www.wbfin.nic.in/) total expenditure heads (revenue account) minus line 3604
compensation and assignments to Local Bodies and PRIs (page 14)
Own Revenue:
o PRI- West Bengal data base, The World Bank (see Annex A)
o Urban – Administrative Report of Municipal Affairs Department, 2005, Government of
West Bengal (pages 25 and 26).
o State - Finance Department, Government of West Bengal, Budget Publications 2006-07
(http://www.wbfin.nic.in/): tax revenue plus non tax revenue (pages 1 and 3)
77
Table 2
Capital and Current Expenditure Shares by PRI: 2005
Capital Current Total
Districts
(N=17)
85.50 14.50 100
Blocks
(N=288)
62.41 37.59 100
Gram
Panchayats
(N=3,016)
63.16 36.84 100
Total 63.4 36.6 100
Source: PRI-West Bengal data base: The World Bank (See Annex A).
78
Table 3
Expenditure Shares by PRI Type: 2005
Education Housing Infrastructure Employment
Generation
Panchayat
Administration
Poverty
Alleviation
Health Other
Districts
(N=17)
1.6 18.1 28.6 30.6 3.9 1.3 1.0 14.9
Blocks
(N=293)
15.2 2.9 20.8 28.3 3.8 10.0 5.6 13.4
Gram
Panchayats
(N=3,082)
0.9 30.1 6.2 24.8 25.4 6.7 2.8 3.1
Total 2.1 27.7 7.6 25.1 23.5 6.9 3.1 4.0
Source: PRI-West Bengal data base: The World Bank (See Annex A).
79
Table 4
Administration and Salary Costs:
by Level of Rural Local Government in 2005
Districts Blocks GPs Total
Administration 0.5 0.9 0.6 0.6
Salary 2.2 1.4 19.2 17.6
All Other
Expenses
97.3 97.7 80.2 81.8
Source: PRI-West Bengal data base: The World Bank (See Annex A).
80
Table 5
Expenditures and Revenues of Districts, Blocks, and Gram Panchayats:
by Population Size Group: 2005
Population
Size
Number in
sample
Percent of
Total
Population
Percent of
Total
Expenditures
Percent of
Total Own
Source
Revenue
Districts
Less than
2,500,000
4 11.8 15.2 6.8
2,500,001-
5,000,000
9 48.8 56.1 40.4
Greater than
5,000,000
4 39.4 28.6 52.8
Total 17 100 100 100
Blocks
Less than
150,000
93 21.5 29.1 27.0
150,001-
250,000
156 56.4 53.1 60.1
Greater than
250,000
39 22.1 17.8 13.0
Total 288 100 100 100
Gram Panchayats
Less than
15,000
849 18.6 23.9 20.9
15,001-20,000 1,075 34.7 34.8 38.4
20,000-25,000 681 28.1 26.1 25.7
Greater than
25,000
356 18.7 15.3 15.1
Total 2,961 100 100 100 Source: PRI-West Bengal data base (See Annex A).
81
Table 6
Percent Distribution of Expenditures by Gram Panchayats:
by Population Size in 2005
Source:
PRI-West
Bengal
data base
(See
Annex
A).
Population
Size
Number
in
Sample
Percent of
Population
Education Housing Infrastructure Employment
Generation
Panchayat
Administration
Poverty
Alleviation
Other Total
Districts
Less than
2,500,000
4 11.8 2.2 13.6 26.1 39.9 4.3 1.5 12.4 100
2,500,001-
5,000,000
9 48.8 1.0 22.4 26.9 30.2 3.3 1.4 14.8 100
Greater
than
5,000,000
4 39.4 2.1 13.1 34.4 22.4 4.8 0.7 22.5 100
Blocks
Less than
150,000
93 21.5 14.3 2.3 19.7 29.7 3.8 10.9 19.3 100
150,001-
250,000
156 56.4 15.7 3.0 21.0 26.6 3.9 9.8 20.0 100
Greater
than
250,000
39 22.1 14.8 3.6 22.9 32.1 3.4 8.5 14.7 100
Gram Panchayats
Less than
15,000
849 18.6 0.8 29.0 5.5 23.9 27.9 6.5 6.4 100
15,001-
20,000
1,075 34.7 1.1 28.3 6.3 24.6 27.3 7.0 5.4 100
20,000-
25,000
681 28.1 0. 30.9 7.0 25.7 23.4 6.8 6.2 100
Greater
than
25,000
356 18.7 0.7 33.5 6.4 26.4 21.7 6.6 4.7 100
82
Table 7
Percent Distribution of Revenues of Gram Panchayats by Source
and by Population Size in 2005a
Own Source Revenue Intergovernmental
Transfers
Population
Size
Percent of
Total
revenue
OSR Tax
as a
Percent
of Total
Revenue
OSR Non-tax
as a Percent
of Total
Revenue
Percent of Total
Revenue
Less than
2,500,000 2.6 0 2.6 97.4
2,500,001-
5,000,000 3.2 0 3.2 96.8
Greater than
5,000,000 7.1 0 7.1 93.0
Total 4.0 0 4.0 96.0 Less than
150,000
2.9 0 2.9 97.1
150,001-
250,000
3.1 0 3.1 96.9
Greater than
250,000
2.5 0 2.5 97.5
Total 3.0 0 3.0 97.1
Less than
15,000
5.4 2.2 3.2 94.6
15,001-20,000 6.4 2.8 3.6 93.6
20,000-25,000 6.1 2.6 3.5 93.9
Greater than
25,000
6.2 2.9 3.3 93.8
Total 6.0 2.6 3.4 94.0 a Unweighted means
Source: PRI-West Bengal data base: The World Bank (See Annex A).
83
Table 8
The Composition of Intergovernmental Transfers by Source: 2005a
Population
Size
Total Grants
and Transfers
as a Percent of
Revenue
From Central
Government
From State
Government
Districts 94.3 87.6 6.7
Blocks 96.0 80.9 15.6
Gram
Panchayats
93.8 68.8 24.9
Total 94.0 70.0 24.0 a Unweighted means
b The data file contains a category of “other transfers.” This category includes “Other Continuing
Education”, “transfers for the Village Education Committee”, and “Other development programs”.
The file does not include these as separate data entries so they cannot be distributed between
central and state transfers. They are therefore excluded from the entire analysis.
Source: PRI-West Bengal data set, World Bank (See Annex A)
84
Table 9
Percent Distribution of Central Transfers to Local Governments: 2005
Schemes
Population
Size
Total SGRY IAY Other Union
Finance
Commission
MPLAD
Districts
Less than
2,500,000
100 68.1 21.3 4.5 4.9 1.2
2,500,001-
5,000,000
100 48.1 30.4 12.3 8.7 0.6
Greater
than
5,000,000
100 40.5 19.2 17.9 21.2 1.2
Average 100 51.0 25.6 11.8 10.7 0.9
Blocks
Less than
150,000
100 35.2 0.9 46.1 13.0 4.9
150,001-
250,000
100 36.9 0.3 46.3 9.2 7.3
Greater
than
250,000
100 42.5 0.0 41.5 12.4 3.6
Average 100 37.1 0.5 45.6 10.9 6.0
Gram Panchayats Less than
15,000 100 32.7 40.4 19.9 6.8 0.2
15,001-
20,000 100 32.9 38.6 20.4 7.8 0.3
20,000-
25,000 100 31.6 40.9 19.5 7.8 0.2
Greater than
25,000 100 31.2 43.1 18.3 7.3 0.02
Average 100 32.4 40.2 19.8 7.5 0.2 Source: PRI-West Bengal data set, World Bank (See Annex A)
85
Table 10
Percent Distribution of State Transfers to Local Governments: 2005
Population
Size
Total Untied Salary State
Finance
Commission
State
Sponsored
Schemes
BEUP
Districts
Less than
2,500,000
100 27.4 41.6 0.0 0.0 15.3
2,500,001-
5,000,000
100 28.1 61.1 0.6 0.5 0.4
Greater
than
5,000,000
100 13.1 70.6 0.0 1.1 1.2
Average 100 24.4 58.8 0.3 0.5 4.1
Blocks
Less than
150,000
100 1.3 25.5 0 2.1 46.0
150,001-
250,000
100 1.4 20.7 0 1.8 51.8
Greater
than
250,000
100 3.7 23.9 0 4.8 44.0
Average 100 1.7 22.7 0 2.3 48.9
Gram Panchayats Less than
15,000 100 0.8 80.1 0.2 2.1 1.3
15,001-
20,000 100 1.0 78.8 0.0 2.6 2.2
20,000-
25,000 100 0.9 77.8 0.01 2.7 2.4
Greater than
25,000 100 1.2 77.4 0.08 1.9 2.6
Average 100 0.9 78.7 0.07 2.4 2.0 Source: PRI-West Bengal data set, World Bank (See Annex A)
86
Table 11
Characteristics of Districts, Blocks, and Gram Panchayats: 2005 Districts Blocks Gram Panchayats
Indicator Mean
(N=17)
Min Max CV Mean
(N=288)
Min Max CV Mean
(N=2,432)
Min Max CV
Total
Population
3,986,591 1,503,178 8,934,286 50.0 180,663 69,587 418,461 33.8 18,198 1,065 47,360 31.4
Percent of
population in
Scheduled
caste and
tribe
32.1 13.3 55.6 37.2 33.0 1.4 72.7 48.8 36.0 0.6 96.7 53.4
Percent
Female
Population
48.5 47.5 49.1 0.73 48.7 45.4 51.1 1.15 48.8 42.4 52.9 1.51
Literacy rate 57.0 37.8 80.0 19.8 54.2 24.2 75.9 18.0 54.5 12.5 79.4 19.67
Workers as a
percent of
population
40.7 32.5 87.1 31.7 38.0 27.1 59.0 16.9 38.8 23.1 69.4 19.6
Percent of
total workers
in
agriculture
34.0 1.0 67.1 38.8
Percent of
total
marginal
workers
11.1 1.1 40.7 58.2
Gross State
Product Per
Capita
11,595 8,682 15,621 16.8
Source: PRI-West Bengal data set, World Bank (See Annex A)
87
Table 12
Per Capita Income and
Per Capita Expenditures of PRI: By District in 2005
District Name Per Capita
Per Capita
Expenditures
Gross State
Product
For all levels of
PRI
Uttar Dinajpur 868.2 419.49
Koch Bihar 961.2 725.88
Birbhum 1028.4 407.12
Purulia 1036.1 682.82
Murshidabad 1073 231.18
North Twenty Four 1094.1 278.89
Dakshin Dinajpur 1098.3 430.48
Maldah 1102.9 273.18
Jalpaiguri 1111.4 676.64
Bankura 1115.4 403.63
South Twenty Four
Parganas 1200.7 292.26
Nadia 1219.4 255.98
Darjiling 1449.2 490.53
Hugli 1472.2 310.73
Haora 1562.1 231.7 Note: The following districts are not included due to lack of gross state product data: Paschim
Medinipur, Purba Medinipur, and Barddhaman
88
Table 13
Per Capita Expenditures by Population Size for Gram Panchayats: 2005
Population
Size
Per capita
expendituresa
(in Rs)
Minimum Maximum Coefficient of
Variation
Gram Panchayats Less than
15,000 183.6 30.9 1,423.3 83.5
15,001-20,000 125.9 19.3 580.1 59.8 20,000-25,000 117.1 17.7 538.2 67.1 Greater than
25,000 102.8 14.9 373.1 61.4
Total N=2,961 137.7 14.9 1,423.3 78.2 a Unweighted mean
Source: PRI-West Bengal data set, World Bank (See Annex A)
89
Table 14
Per Capita Expenditures (Minus Salary Grant) by Population Size, 2005
Population Size Per capita
expenditures
excluding fixed
costsa
(in Rs)
Minimum Maximum Coefficient of
Variation
Districts Less than
2,500,000 176.7 103.9 225.5 29.2
2,500,001-
5,000,000 169.7 68.7 288.3 44.3
Greater than
5,000,000 103.7 79.3 116.7 16.6
Total N=17 155.8 68.7 288.3 42.0
Blocks Less than 150,000 121.8 14.6 394.5 62.7 150,001-250,000 83.0 15.7 349.6 55.8 Greater than
250,000 70.4 20.0 178.3 55.4
Total N=288 93.8 14.6 394.5 64.2
Gram Panchayats Less than 15,000 150.5 11.0 1,145.8 92.0 15,001-20,000 105.2 2.7 559.4 71.5 20,000-25,000 101.1 5.7 521.7 77.5 Greater than
25,000 90.4 7.3 360.6 69.4
Total N=2,961 115.4 2.7 1,145.8 70.0
a Expenditures exclude reported salary grant.
Source: PRI-West Bengal data set, World Bank (See Annex A)
90
Table 15
OLS Estimation of the Determinants of Variations in Per Capita Total Expenditures
of Gram Panchayats: 2005
(Log Per Capita Expenditures)
Level of Government Gram Panchayat
Constant 9.96**
(41.64)
Log Population -0.432**
(14.62)
Log Percent SC/ST
Population
0.184**
(13.7)
Log Literacy Rate 0.291**
(5.50)
Log Percent of
Agricultural
0.090**
(7.81)
Labor
R2 0.81
N 2,098 a Coefficients on district and block dummy variables not reported.
* Significant at the 95% level or higher
** Significant at the 99% level or higher
Notes:
Significant District Effects:
Positive: Darjiling, Jalpaiguri, and Koch Bihar
Negative: Bankura, Birbhum, Haora, Pachim Medinipur, Purba Medinipur,
Nadia, Purulia, North Twenty-four Parganas, South Twenty-four Parganas,
Dakshin Dinajpur
The omitted district is Uttar Dinajpur.
Seventy four of the 325 block dummy variables were significant (23 percent).
This indicates that they are important, indicating that there are differences in
spending levels within districts, even after we account for socio-economic
characteristics included in Table 13. A random block was omitted for each
district.
91
Table 16
OLS Estimation of the Determinants of Variations in Per Capita Total Expenditures
(minus salary grants) of Gram Panchayats: 2005
(Log Per Capita Total Expenditures)
.
Level of Government Gram Panchayat
Constant 8.85**
(30.63)
Population -0.335**
(13.22)
Percent SC/ST
Population
0.240**
(15.07)
Literacy Rate 0.282**
(4.40)
Percent of Agricultural 0.042**
(2.39)
Labor
R2 0.80
N 2,079 a Coefficients on district and block dummy variables not reported.
* Significant at the 95% level or higher
** Significant at the 99% level or higher
Notes:
Significant District Effects:
Postive: : Darjiling, Jalpaiguri, and Koch Bihar
Negative: Bankura, Barddhaman, Birbhum, Haora, Hugli, Maldah, Paschim
Medinipur, Purba Medinipur, Nadia, Purulia, North Twenty Four Paraganas,
South Twenty Four Parganas, and Dakshim Dinajpur.
The omitted district is Uttar Dinajpur.
Sixty two of the 325 block dummy variables were significant (19 percent).
This indicates that they are important, indicating that there are differences in
spending levels within districts, even after we account for socio-economic
characteristics included in Table 13a. A random block was omitted for each
district.
92
Table 17
Per Capita Own Revenues for Gram Panchayats by Population Size: 2005
Population
Size
Per capita own
source revenue
(in Rs)a
Minimum Maximum Coefficient of
Variation
Gram Panchayats Less than 15,000 8.5 0.0 238.6 146.0 15,001-20,000 8.0 0.0 190.2 160.5 20,000-25,000 6.6 0.02 87.7 116.5 Greater than
25,000 5.8 0.05 37.0 101.0
Total N=2,943 7.6 0.0 238.6 146.3 a 3 GPs report values of per capita own source revenues less than 0.001 and 4 report values greater than
100.
Source: PRI-West Bengal data set, World Bank (See Annex A)
93
Table 18
OLS Estimation of the Determinants of Variations in Per Capita Own Source
Revenue of Gram Panchayats: 2005
(Log of dependent variable)
Level of Government Gram Panchayata
Constant 5.27**
(6.28)
Log Population -0.221**
(2.97)
Log Percent SC/ST 0.090*
Population (1.93)
Log Literacy Rate 0.924**
(4.86)
Log Percent of
Agricultural
-0.145**
Labor (2.56)
R2 0.55
N 2,067
a Coefficients on district and block dummy variables not reported; t-statistics in parenthesis.
* Significant at the 95% level or higher, ** Significant at the 99% level or higher
Notes:
Significant District Effects:
Positive effects: Bankura and Hugli
Negative effects: Birbhum, Darjiling, Haora, Purba Medinipur, Purulia
The omitted district is Uttar Dinajpur.
Significant Block Effects:
The block dummy variable is significant for 55 blocks—approximately 17 percent of the cases, split almost
evenly between negative and positive coefficients. A random block was omitted for each district.
94
Table 19
OLS Estimation of the Determinants of Variations in Per Capita Own Source
Revenue of Gram Panchayats: 2005
(Log of dependent variable)
Level of Government Gram Panchayata
Constant 2.091*
(1.74)
Log Population -0.080
(0.97)
Log Percent SC/ST 0.027
Population (0.54)
Log Literacy Rate 0.858**
(4.50)
Log Percent of
Agricultural
-0.164**
Labor (3.15)
Log Per Capita Grants
and Transfers
0.370*
(3.70)
R2 0.55
N 2,067
a Coefficients on district and block dummy variables not reported; t-statistics in parenthesis
* Significant at the 95% level or higher, ** Significant at the 99% level or higher
Notes:
Significant District Effects:
Positive: Bankura, Barddhaman, Hugli
Negative: Birbham, Darjiling, Purba Medinipur, Purulia
The omitted district is Uttar Dinajpur.
Significant Block Effects:
The block dummy variable is significant in 22 percent of the cases (71 blocks). A random block was
omitted for each district.
95
Table 20
Tax Effort, Gram Panchayats, 2005
GP Name Own Source
revenue per capita
Estimated Own
Source Revenue per
capita
Tax Effort
Domahana 2.77 4.47 0.62
Dutta Pulia 2.89 4.25 0.68
Patharghata 15.03 19.92 0.75
Haripur 2.12 2.79 0.76
Chuprijhara 1.53 1.83 0.83
Bansra 4.12 4.77 0.87
Lalgarh 1.30 1.46 0.89
Prusadpur 2.52 2.22 1.13
Oldabari 8.49 4.10 2.07
Kalia Chak 9.53 2.39 3.99 Source: Calculated from the PRI-West Bengal data base: The World Bank (see Annex A).
96
Table 21
OLS Estimation of the Determinants of Variations in Per Capita SGRY
of Gram Panchayats: 2005
(Log of dependent variable)
a Coefficients on district and block dummy variables not reported; t-statistics in
parenthesis
* Significant at the 95% level or higher, ** Significant at the 99% level or higher
Notes:
Significant District Effects:
Positive effects: Darjiling and Haora
Negative effects: Bankura, Barddhaman, Jalpaiguri, Koch Bihar, Paschim
Medinipur, Nadia, North 24, South 24
The omitted district is Uttar Dinajpur.
Significant Block Effects:
The block dummy variable is significant for 38 blocks—approximately 12 percent
of the cases. A random block was omitted for each district.
Level of Government Gram Panchayata
Constant 7.179**
(14.96)
Log Population -0.326**
(7.59)
Log Percent SC/ST 0.377**
Population (14.15)
Log Literacy Rate 0.264*
(2.50)
Log Percent of
Agricultural
0.029
Labor (1.31)
R2 0.64
N 2,066
97
Table 22
OLS Estimation of the Determinants of Variations in Per Capita IAY
of Gram Panchayats: 2005
(Log of Dependent Variable)
a Coefficients on district and block dummy variables not reported; t-statistics in parenthesis
* Significant at the 95% level or higher, ** Significant at the 99% level or higher
Notes:
Significant District Effects:
Positive: Darjiling, Jalpaiguru, Koch Bihar
Negative: Purulia
The omitted district is Uttar Dinajpur.
Significant Block Effects:
There are few significant block effects. Of 325 block dummies, only 10 were significant. A random block
was omitted for each district.
Level of Government Gram
Panchayat
Constant -5.85**
(10.61)
Log Population -0.250**
(5.07)
Log Percent SC/ST
Population
0.322**
(10.35)
Log Literacy Rate 0.285*
(2.05)
Log Percent
Agricultural Labor
0.132**
(5.16)
R2 0.65
N 1,984
98
Table 23
OLS Estimation of the Determinants of Variations in Per Capita Total Central
Schemes
of Gram Panchayats: 2005
(Log of Dependent Variable)
Level of Government Gram Panchayata
Constant 8.405**
(25.12)
Log Population -0.287**
(9.83)
Log Percent SC/ST
Population
0.293**
(15.87)
Log Literacy Rate 0.281**
(3.85)
Log Percent
Agricultural Labor
0.102**
(6.67)
Log Percent Marginal
Workers
-0.037*
(1.79)
R2 0.79
N 2,097
a Coefficients on district and block dummy variables not reported; t-statistics
in parenthesis
* Significant at the 95% level or higher, ** Significant at the 99% level or
higher
Notes:
Significant District Effects:
Positive effects: Darjiling
Negative effects: Bankura, Barddhaman, Birbhum, Haora, Hugli, Maldah,
alpaiguri, Koch Bihar, Paschim Medinipur, Pubra Medinipur, Nadia, North
Twenty Four Parganas, South Twenty Four Parganas, Dakshin Dinajur
The omitted district is Uttar Dinajpur.
Significant Block Effects:
The block dummy variable is significant for only 12 blocks. A random block
was omitted for each district.
99
Table 24
OLS Estimation of the Determinants of Variations in Per Capita State Transfers to
Gram Panchayats: 2005
(Log of Dependent Variable)
Level of Government Gram Panchayata
Constant 11.46**
(34.30)
Log Population -0.820**
(27.26)
Log Percent SC/ST 0.018
(0.93)
Log Literacy Rate 0.141*
(1.76)
Log Percent
Agricultural Labor
0.050**
(3.42)
R2 0.61
N 2,022
a Coefficients on district and block dummy variables not reported; t-
statistics in parenthesis
* Significant at the 95% level or higher, ** Significant at the 99% level
or higher
Notes:
Significant District Effects:
Positive effects: Barddhaman and Hugli
Negative effects: Paschim Medinipur and South Twenty Four Parganas
The omitted district is Uttar Dinajpur.
Significant Block Effects:
The block dummy variable is significant for 58 of the block dummy
variables—18 percent of all block effects. A random block was
omitted for each district.
100
Table 25
OLS Estimation of the Determinants of Variations in Per Capita Total Grants and
Transfers of Gram Panchayats: 2005
(Log of Dependent Variable)
Level of Government Gram Panchayata
Constant
9.21**
(38.31)
Log Population -0.452**
(20.90)
Log Percent SC/ST 0.195**
(14.30)
Log Literacy Rate 0.271**
(4.72)
Log Percent Agricultural
Labor 0.087**
(7.93)
R2 0.82
N 2,022 a Coefficients on district and block dummy variables not reported; t-statistics in parenthesis
* Significant at the 95% level or higher, ** Significant at the 99% level or higher
Notes:
Significant District Effects:
Positive effects: Darjiling, Jalpaiaguri, Kock Bihar, Maldah, and Purulia
Negative effects: Bankura, Birbhum, Paschim Medinipur, and Nadia
The omitted district is Uttar Dinajpur.
Significant Block Effects:
Among the block effects, we find 71 significant coefficients—about 22 percent of the total
block dummy variables. A random block was omitted for each district.
101
Table 26
Indicators of Financial Condition for Gram Panchayats: 2003-2005
2003 2004 2005
Gram Panchayats
Number of GPs with
an overall deficit (not
including opening
balance)
1,494 1,334 1,501
Number of GPs with
an overall surplus
(not including
opening balance)
1,491 1,703 1,510
Total surplus (not
including opening
balance) as a percent
of total revenue for
surplus GPs
12.0 14.0 13.0
Total Deficit (not
including opening
balance) as a percent
of total revenue for
deficit GPs
15.0 13.6 11.5
Closing balance as a
percent of total
expenditures ALL
21.8 28.3 23.5
Closing balance as a
percent of grants and
transfers ALL
19.5 26.1 21.5
Number of GPs with
closing balance equal
-5 percent to +5
percent of total
expenditures
428 242 363
N 2,985 3,038 3,012 a Eliminating one outliers, this is 72.3 for closing balance and 83.4 for deficit/total revenue
Source: PRI-West Bengal data base, World Bank (See Annex A)
102
Table 27
Gram Panchayats with Chronic Current Account
Surplus or Deficit
Number Number with Deficit Greater than
Opening Balance
Average Per Capita
Expenditures in 2005
2003 2004 2005
Deficit for
all 3 Years
199 5 3 4 129.0
Deficit for 2
Years
1,209 31 20 17 134.0
Deficit for 1
Year Only
1,313 23 17 7 141.3
Surplus for
all 3 Years
257 97.7
Surplus for
2 Years
1,313 141.3
Surplus for
1 Year Only
1,209 134.0
Source: PRI-West Bengal data base, World Bank (See Annex A).
103
Table 28
Ratio of Current Account Deficit
to Opening Balance: 2005
Number of Gram Panchayats
More than 100 percent 27
50-100 percent 608
25-50 percent 483
0-25 percent 383
Source: PRI-West Bengal data base, World Bank (See Annex A)
104
Table 29
Probit Analysis
Gram Panchayat Deficits
Variable Chronic Deficit GPs (2005)a Deficit GPs (2005)
Constant -1.41**
(13.61)
-0.118*
(1.85)
Opening Balance per capita -0.011**
(3.73)
0.016**
(11.82)
Percent SC/ST population 0.368*
(1.69)
-0.718**
(4.45)
Own Source Revenue per
capita
-0.006
(1.06)
-0.011**
(3.85)
N 2,098 2,095
Likelihood ratio 22.8 184 a Chronic deficit GPs are defined as those with deficits for each year, 2003, 2004, and
2005 (199 GPs).
Source: PRI-West Bengal data base, World Bank (see Annex A).
105
Table 30
OLS Estimation of the Determinants of Variations in Per Capita Deficitsb
of Gram Panchayat Governments: 2005
(Log Dependent Variable)
Level of Government Gram Panchayata
Constant -0.003
(0.01)
Log Population -0.171
(1.17)
Log Percent SC/ST 0.129
Population (1.43)
Log Opening balance 1.51**
(18.8)
Log Central Sponsored
Schemes Per Capita
-0.172*
(1.95)
R2 0.55
N 996
a Coefficients on district and block dummy variables not reported; t-statistics in
parenthesis b Deficits expressed as positive numbers.
* Significant at the 95% level or higher, ** Significant at the 99% level or higher
Notes:
Significant District Effects:
Positive effects: Darjiling and Koch Biharrddhaman and Hugli
Negative effects: Paschim Medinipur and South Twenty Four Parganas
The omitted district is Uttar Dinajpur.
Significant Block Effects:
The block dummy variable is significant for just 16 of the block dummy variables
out of a total of 325 block dummy variables. A random block was omitted for
each district.
106
Table 31
Per Capita Expenditures by Population Size for Districts and Blocks: 2005
Population
Size
Per capita
expendituresa
(in Rs)
Minimum Maximum Coefficient of
Variation
Districts Less than
2,500,000 176.7 103.9 225.5 29.2
2,500,001-
5,000,000 169.7 68.7 288.3 44.3
Greater than
5,000,000 103.7 79.3 116.7 16.6
Total N=17 155.8 68.7 288.3 42.0
Blocks Less than
150,000 123.1 14.7 396.3 62.7
150,001-
250,000 84.0 16.3 349.9 55.8
Greater than
250,000 71.1 20.5 177.0 55.4
Total N=288 94.8 14.7 396.3 64.2 a Unweighted mean
Source: PRI-West Bengal data set, World Bank (See Annex A).
107
Table 32
OLS Estimation of the Determinants of Variations in Per Capita Total Expenditures
of Districts and Blocks: 2005
(Log Per Capita Total Expenditures)
Level of Government Blocka District
Constant 8.07** 9.24**
(7.52) (3.96)
Log Population -0.309** -0.227
(3.59) (1.42)
Log Percent SC/ST 0.158** 0.695**
Population (2.81) (3.73)
R2 0.50 0.63
N 288 17 a Coefficients on district and block dummy variables not reported.
* Significant at the 95% level or higher
** Significant at the 99% level or higher
Notes: For the block regression, positive, significant district effects are
found for the following districts: Koch Bihar, Purulia, and Paschim
Meninipur.
The omitted district is Uttar Dinajpur.
108
Table 32a
OLS Estimation of the Determinants of Variations in Per Capita Total Expenditures
(minus salary grants)
of Blocks and Districts: 2005
(Log Per Capita Total Expenditures)
Level of Government Blocka District
Constant 7.89** 9.13**
(7.14) (4.13)
Log Population -0.295** -0.221
(3.32) (1.41)
Log Percent SC/ST 0.166** 0.722**
Population (2.88) (4.00)
R2 0.48 0.63
N 288 17 a Coefficients on district and block dummy variables not reported
* Significant at the 95% level or higher
** Significant at the 99% level or higher
Notes: For the block regression, positive, significant district effects are found
for the following districts: Koch Bihar, Purulia, and Paschim Meninipur.
The omitted district is Uttar Dinajpur.
109
Table 33
Per Capita Own Revenues for Districts and Blocks by Population Size: 2005
Population
Size
Per capita own
source revenue
(in Rs)
Minimum Maximum Coefficient of
Variation
Districts Less than
2,500,000 5.1 0.62 16.6 150.1
2,500,001-
5,000,000 4.6 1.80 8.1 53.3
Greater than
5,000,000 7.2 2.73 15.0 77.8
Total N=17 5.3 0.62 16.6 86.0
Blocks Less than 150,000 3.7 0.1 32.3 165.2 150,001-250,000 2.7 0.1 39.8 153.2 Greater than
250,000 1.5 0.02 6.1 96.3
Total N=288 2.7 0.02 39.8 160.0 Notes: 3 GPs report values of OSRPC less than 0.001 and 4 report values greater than 100.
Source: PRI-West Bengal data set, World Bank (See Annex A).
110
Table 34
OLS Estimation of the Determinants of Variations in Per Capita Own Source
Revenue of Blocks: 2005
(Log of dependent variable)
a Coefficients on district dummy variables not reported; t-statistics in parenthesis
* Significant at the 95% level or higher, ** Significant at the 99% level or higher
Notes:
Significant District Effects:
Positive effects: Barddhaman The omitted district is Uttar Dinajpur.
Level of Government Blocka
Constant 1.33
(0.40)
Log Population -0.116
(0.43)
Log Percent SC/ST -0.102
(0.61)
R2 0.12
N 277
111
Table 35
OLS Estimation of the Determinants of Variations in Per Capita Own Source
Revenue of Districts and Blocks: 2005
(Log of dependent variable)
Level of Government Blocka
District
Constant -2.13 -8.50
(0.60) (1.02)
Log Population 0.149 0.396
(0.52) (0.76)
Log Percent SC/ST -0.097 -1.097*
Population (0.54) (1.79)
Log Per Capita Grants
and Transfers
0.339** 1.419*
(1.93)
(3.61)
R2 0.14 0.11
N 277 17 a Coefficients on district dummy variables not reported; t-statistics in parenthesis
* Significant at the 95% level or higher, ** Significant at the 99% level or higher
Notes:
Significant District Effects:
Positive: Barddhaman
The omitted district is Uttar Dinajpur.
112
Table 36
OLS Estimation of the Determinants of Variations in Per Capita SGRY
of Districts and Blocks: 2005
(Log of dependent variable)
a Coefficients on district and block dummy variables not reported; t-statistics in parenthesis
* Significant at the 95% level or higher, ** Significant at the 99% level or higher
Notes:
Significant District Effects:
Positive effects: Bankura, Jalpaiguri, Koch Bihar, and Purulia
Negative effects: Murshidabad
The omitted district is Uttar Dinajpur.
Level of Government Blocka District
Constant 6.05** -48.39*
(1.92)
(17.28)
Log Population -0.218** -0.415
(6.44) (0.81)
Log Percent SC/ST 0.424** 2.37**
Population (23.76) (3.36)
Log Literacy Rate 0.195* -5.84**
(2.06) (3.07)
Log Per Capita GSP 6.14*
(2.81)
R2 0.60 0.69
N 266 15
113
Table 37
OLS Estimation of the Determinants of Variations in Per Capita IAY
Of Blocks and Districts: 2005
(Log of Dependent Variable)
a Coefficients on district and block dummy variables not reported; t-statistics in parenthesis
* Significant at the 95% level or higher, ** Significant at the 99% level or higher
Notes:
Significant District Effects:
Positive: Barddhaman
The omitted district is Uttar Dinajpur.
Level of Government Blocka
District
Constant 23.44
(1.05)
-12.00
(1.55)
Log Population -2.49
(1.55)
0.839**
(2.39)
Log Percent SC/ST
Population
0.383
(0.47)
2.07**
(4.26)
Log Literacy Rate -2.47
(0.65)
-4.63**
(3.09)
Log Per Capita GDP 3.25**
(2.74)
R2 0.40 0.59
N 229 14
114
Table 38
OLS Estimation of the Determinants of Variations in Per Capita Total Central
Schemes
of Districts and Blocks: 2005
(Log of Dependent Variable)
Level of Government Blocka
District
Constant 10.118*
(4.76
15.59**
(3.14)
Log Population -0.509**
(3.09)
0.069
(0.35)
Log Percent SC/ST
Population
0.381**
(3.55)
1.564**
(5.73)
Log Literacy Rate -0.347
(0.76)
-2.24**
(3.07)
Log Per Capita GDP 1.75*
(2.06)
R2 0.35 0.81
N 288 15
a Coefficients on district and block dummy variables not reported; t-statistics in
parenthesis
* Significant at the 95% level or higher, ** Significant at the 99% level or higher
Notes:
Significant District Effects:
Positive effects: Darjiling
The omitted district is Uttar Dinajpur.
115
Table 39
OLS Estimation of the Determinants of Variations in Per Capita State Transfers to
Districts and Blocks: 2005
(Log of Dependent Variable)
Level of Government Blocka
District
Constant 3.873
(1.33)
14.36*
(1.07)
Log Population -0.181
(1.53)
Log Percent SC/ST 0.067
(0.87)
-0.257
(0.92)
Log Literacy Rate 0.189
(0.57)
1.92*
(2.39)
Log Percent Female
Population
-2.80
(0.86)
35.96*
(2.15)
Log Per Capita GSP
-2.42*
(2.70)
R2 0.27 0.43
N 288 15
a Coefficients on district and block dummy variables not reported; t-statistics in parenthesis
* Significant at the 95% level or higher, ** Significant at the 99% level or higher
Notes:
Significant District Effects:
Positive effects: Bankura, Barddhaman, Paschin Medinipur, and Purulia
The omitted district is Uttar Dinajpur.
116
Table 40
OLS Estimation of the Determinants of Variations in Per Capita Total Grants and
Transfers of Districts and Blocks: 2005
(Log of Dependent Variable)
Level of Government Blocka District
Constant 8.15** 7.508**
(6.97) (2.97)
Log Population
-0.327**
(3.46)
-0.113
(0.65)
Log Percent SC/ST 0.177** 0.775**
Population (2.98) (3.92)
Log Literacy Rate -0.154
(.56)
R2 0.48 0.61
N 281 15 a Coefficients on district dummy variables not reported; t-statistics in parenthesis
* Significant at the 95% level or higher, ** Significant at the 99% level or higher
Notes:
Significant District Effects:
Positive effects: Koch Bihar, Paschim Medinipur, and Purulia
The omitted district is Uttar Dinajpur.
117
Table 41
Indicators of Financial Condition for Districts and Blocks: 2003-2005
2003 2004 2005
Districts
Number of districts
with an overall
deficit (not including
opening balance)
15 13 14
Number of districts
with an overall
surplus (not
including opening
balance)
2 5 3
Total surplus (not
including opening
balance) as a percent
of total revenue for
surplus districts
11.6 21.7 13.4
Total Deficit (not
including opening
balance) as a percent
of total revenue for
deficit districts
103.7a
36.7 16.2
Closing balance as a
percent of total
expenditures ALL
67.4 74.9 47.0
Closing balance as a
percent of grants and
transfers ALL
130.0a
83.9 52.0
Number of districts
with closing balance
equal -5 percent to +5
percent of total
expenditures
1 1 1
N 17 17 17
Blocks
Number of blocks
with an overall
deficit (not including
opening balance)
166 192 123
Number of blocks
with an overall
surplus (not
including opening
balance)
110 89 155
118
Total surplus (not
including opening
balance) as a percent
of total revenue for
surplus blocks
17.6 15.3 16.9
Total Deficit (not
including opening
balance) as a percent
of total revenue for
deficit blocks
35.7 37.4 22.4
Closing balance as a
percent of total
expenditures ALL
76.2 74.7 69.3
Closing balance as a
percent of grants and
transfers ALL
80.8 83.0 66.6
Number of blocks
with closing balance
equal -5 percent to +5
percent of total
expenditures
1 3 1
N 276 281 278 a Eliminating one outliers, this is 72.3 for closing balance and 83.4 for deficit/total revenue
Source: PRI-West Bengal data base, World Bank (See Annex A).
119
Table 42
OLS Estimation of the Determinants of Variations in Per Capita Deficitsa
of Districts and Blocks: 2005
(Log Dependent Variable)
Level of Government Block District
Constant -1.27 12.90
(0.32) (0.67)
Log Population 0.100 -0.801
(0.36) (0.81)
Log Percent SC/ST 0.398 1.015
Population (1.37) (0.77)
Log Opening balance 0.567** 1.672*
(2.97) (1.88)
Log Central Sponsored
Schemes Per Capita
0.120
(1.38
-0.922
(0.71)
R2 0.38 0.18
N 123 13
a Deficits expressed as positive numbers
* Significant at the 95% level or higher, ** Significant at the 99% level or higher
Notes:
Significant District Effects:
Negative effects: Jalpaiguri
The omitted district is Uttar Dinajpur.
120
Table 43
Reform Options: Alternative Expenditure Levels
(Amounts in Rs)
Three times current
per capita
expenditures
Two times current per
capita expenditures
1. Present spending levela 7,909,652,530 7,909,652,530
2. Proposed spending level 23,728,957,590 15,819,305,060
3. Increase 15,819,305,060 7,909,652,530
4. Present State Fundingb 1,451,315,982 1,451,315,982
5. Required Total State Funding 17,270,621,042 9,360,968,512
6. Required Increase 15,819,305,060 7,909,652,530
7. Required State Funding as a
Percent of State Own Revenue
10.59% 5.74%
a Calculated for all gram panchayats system using the mean per capita expenditure for gram panchayats in
the PRI-West Bengal data base, World Bank (see Annex A) times the total gram panchayat population
(57,734,690). The data base contains data for 3,074 gram panchayats for 2005. This accounts for
approximately 92 percent of gram panchayats by population base or by number of gram panchayats. b Based on the amount of state sponsored schemes to gram panchayats and state grants to gram panchayats
from the PRI-West Bengal data base, World Bank, increased by 8 percent to reflect the total of all state aid
to all gram panchayats in West Bengal.
121
Table 44
Distribution of Per Capita Receipts from State Grants
Under A Minimum Expenditure Proposals
Number of Gram Panchayats
Percent Gain in state aid
as a share of current plus
additional state aid
ThreeTimes Current
Per Capita Expenditures
Two Times Current Per
Capita Expenditures
0-75 249 385
75-80 54 221
80-90 604 1,405
90-93 850 568
93-95 726 196
Greater than 95 591 299
Total number of GPs 3,074 3,074 Source: PRI-West Bengal data base, World Bank (See Annex A).
122
Table 45
Selected Coefficients of Correlation
Variable Per Capita State Grants in
2005
Per Capita Grants Under
the Proposed Minimum
Expenditure System
Percent SC/ST population 0.065 -0.004
Literacy Rate 0.113** 0.039
Percent agricultural
workers
-0.003 -0.132**
Percent marginal workers 0.183** 0.039
Own source revenue per
capita
0.133* 0.111*
Source: PRI-West Bengal data set, World Bank (See Annex A).
* Significant at the 95% level, ** significant at the 99% level.
123
Table 46
Distribution of Per Capita Receipts from State Grants
Under A Formula Proposal
Percent Increase in
State Aid
Number of GPs
0-50 456
50-60 595
60-65 335
65-70 308
70-80 332
Greater than 80 72
Total number of GPs 2,098 Source: PRI-West Bengal data set, World Bank (See Annex A)
124
Table 47
Impacts of the General Fund and Revenue Mobilization Components
Under a Formula Proposal
GP Population Size
Class
Number of GPs Mean Per Capita
Amount Received
from General
Fund
Mean Per Capita
Amount Received
from Revenue
Mobilization
0-15,000 617 112.2 69.3
15,000-20,000 760 104.6 36.1
20,000-25,000 475 105.4 20.7
Greater than 25,000 246 105.1 13.7
Total 2,098 107.0 40.0 Source: PRI-West Bengal data set, World Bank (See Annex A)
125
Table 48
Selected Coefficients of Correlation
Variable Per Capita State Grants
(current system)
Per Capita Grants
(proposed system
including general and
revenue mobilization
components and 2X
minimum expenditure
guarantee)
Percent SC/ST population 0.065 -0.005
Literacy Rate 0.113** 0.015
Percent agricultural
workers
-0.003 -0.040*
Percent marginal workers 0.183** 0.081**
Own source revenue per
capita
0.133* 0.412**
Population -0.505** -0.263** Source: PRI-West Bengal data set, World Bank (See Annex A).
126
REFERENCES
Alok, V.N. (2006). Local Government Organization and Finance: Rural India. In A. Shah
(Ed.), Local Governance in Developing Countries. Washington: The World Bank.
Bahl, R. & Linn, J. (1992). Urban Public Finance in Developing Countries. New York:
Oxford University Press.
Bahl, R. and Martinez-Vazquez, J. (2006). Sequencing Fiscal Decentralization, Policy
Research Working Paper 3914. May 2006: The World Bank, Poverty Reduction and
Economic Management Network, Public Sector Governance Group.
Bahl, R. and Wallace, S. (2007). Intergovernmental Transfers: The Vertical Sharing
Dimension. In J. Martinez-Vazquez and B. Searle (Eds.), Fiscal Equalization:
Challenges in the Design of Intergovernmental Transfers. Springer Publishers.
Bird, Richard (1999). Rethinking Subnational Taxes: A New Look at Tax Assignment IMF
Working Paper 99/165. Washington DC: International Monetary Fund.
Brueckner, J. (1999). Fiscal Decentralization in LDCs: The Effects of Local Corruption
and Tax Evasion. Department of Economics, University of Illinois at Urbana,
mimeograph.
CMIE (2006). Annual Review of the West Bengal Economy: 2004-2005. Kolkata: Center
for Monitoring Indian Economy.
Coondoo, D., Majumder, A., Mukherjee, R. and Neogi, C. (2005). Relative Tax
Performances: Analysis for Selected States in India, Economic Research Unit, Calcutta:
Indian Statistical Institute.
Fisman, R. & Gatti, R. (1999). Decentralization and Corruption: Cross-country and
Cross-State Evidence. World Bank, mimeograph.
Fox, W. F. and Gurley, T. (2006). Will Consolidation Improve Sub-National
Governments? In Policy Research Working Paper 3913, Poverty Reduction and
Economic Management. The World Bank, Public Sector Governance Group.
Government of India (2006). Annual Report of the Ministry of Rural Development, 2005-
2006.
127
Government of India (March 2005). Concurrent Evaluation of Sampoorna Grameen
Rozgar Yojana. Center for Management Development: Thiruvananthapuram.
Government of India (2004). Report of the Twelfth Finance Commission, (2005-2010).
Government of West Bengal (2005). Assignment of Responsibilities on Three Tier
Panchayat Raj Institutions and Mapping of Activities of P.R. Bodies. Kolkata:
Department of Panchayats and Rural Development.
Government of West Bengal (2006). Budget Speech.
Guruswamy, M., Sharma, K. & Mohanty, J.P. (2005). Economic Growth and
Development in West Bengal: Reality versus Perception. In Economic and Political
Weekly, Perspectives, May 21.
Institute of Social Sciences (2005). Impact of Assessment Study of Rural Develoment
Programmes in Bhavnagar District, Gujarat. New Delhi: Ministry of Rural
Development.
Jha, R., Mohanty, M.S., Chatterjee, S., & Chitkara, P. (1999). Tax Efficiency in Selected
Indian States, In Empirical Economics. Volume 24,(pp. 641-654).
Martinez-Vazquez, J., Timofeev, A., & Boex, J. (2006). Reforming Regional-Local
Finance in Rusia. Washington, DC: The World Bank Institute.
Mathur, On Prakash (2001). Approach to State-Municipal Fiscal Relations: Options and
Perspectives. New Delhi: National Institute of Public Finance.
Oommen, M.A. (2006). Fiscal Decentralisation to the Sub-State Level Governments. In
Economic and Political Weekly, Vol. LXI, No 10, March 11.
Oommen, M.A. (2004). Fiscal Decentralization in Kerala. In G. Sethi (Ed.), Fiscal
Decentralization to Rural Governments in India. New Delhi: Oxford University Press.
Pal, M. & Adak, S. (Undated). Assessment and Collection of Land and Building Tax: A
Case Study in Bankura District of West Bengal, mimeographed.
Purfield, C. (2004). The Decentralization Dilemma in India. Working Paper WP/04/02,
Washington: International Monetary Fund.
Rajaraman, I. (2001) Growth Accelerating fiscal Devolution to the Third Tier.
Conference Paper, “India: fiscal Policies to Accelerate Economic Growth” (NIFPF: New
Delhi.), May.
Rajmal (2006). State Finances and Effectiveness of Policy Measures: An Analysis of
Indian States. Reserve Bank of India Occasional Papers.
128
Rao, M. G., H.K. Amar Nath, and B.P. Vani (2004). Fiscal Decentralization in
Karnataka. In G. Sethi (Ed.) Fiscal Decentralization to Rural Governments in India, New
Delhi: Oxford University Press.
Rao, M. G. and N. Singh (2001). Federalism in India: Political Economy and Reform.
Presented at, India: Ten Years of Economic Reform. Danielson Institute, University of
Michigan, September.
Singh, N. and T.N. Srnivasan (2006). Indian Federalism, Economic Reform and
Globalization. In J. Wallack and T.N. Srnivasan (Eds.) Federalism and Economic
Reform: International Perspectives (pp. 301-363). New York: Cambridge University
Press.
SRC (undated). Resource Mobilization by the Panchayati Raj Institutions: A Study of Six
Districts. Post Design Studies for SRD, Government of West Bengal.
State Finance Commission (2002). Recommendations of the Second State Finance
Commission: West Bengal (unpublished).
Subrahmanyam, K. S. (2004). Performance of SFCs, Impact of their Recommendations,
Suggested Framework for the TFS. Hyderabad: National Institute of Rural Development.
Subrahmanyam, K. S. & Annamalai, V. (2004). Innovative Features in Panchayati Raj
in Select States & Scope for Their Replication: A Quick Study. Hyderabad: National
Institute of Rural Development.
World Bank (2006). Reforming Public Services in India: Drawing Lessons from Success
Report No. 35041- IN, Poverty Reduction and Economic Management Sector Unit, South
Asia Department.
World Bank (2005a). State Fiscal Reforms in India: Progress or Prospects. Delhi:
MacMillian.
World Bank (2005). India: Rural Governments and Service Delivery, Agriculture and
Rural Development Unit, South Asia Region, June 14.
World Bank (2004). Urban Finance and Governance Review Report No. 32253-
IN,Energy and Infrastructure Unit, South Asia Region, December
