You are on page 1of 16


Volume and Composition of Government

Subsidies in India, 1987-88
Sudipto Mundle
M Govinda Rao
This paper attempts to measure the volume and composition of subsidies provided by the Central government
and major State governments and concludes that a substantial proportion of GNP, much larger than the explicit
subsidy as revealed in the budgets or as computed even by the broader National Accounts definition, is being
distributed in the form of subsidies through the Central and State budgets, much of it invisible, and that it is
not at all clear that these subsidies are flowing to the intended beneficiaries.

Introduction Government has been defined in this exer- with associated tax prices, could lead to fair-
cise to include only those departments which ly efficient outcomes. However, in the
ECONOMISTS are interested in analysing directly come under the Central government absence of such voting mechanisms, the op-
government subsidies for a number of dif- or the governments of fourteen major States. timal level of public provision of these ser-
ferent reasons. A macro economist dealing In particular, non-departmental public enter- vices remains indeterminate and their costs
with India's fiscal imbalance, i e, the grow- prises or co-operatives have been treated as have to be met out of the general budget
ing revenue deficit in evidence since the early lying outside the government proper. This since they cannot be easily recovered.4
eighties, would be interested in better tar- is admittedly a narrow definition. However, Under these conditions it would be inappro-
geting of subsidies and pruning of unintend- it is necessary in order to frame the inter- priate to apply the concept of a subsidy to
ed subsidies as part of a stabilisation pro- face between the government budget and the expenditure on pure public goods. Could
gramme which attempts to reduce the public enterprises. The difference between we say, for instance, that defence expenditure
revenue deficit, A price theorist would be in- financial assistance extended to such enter- is a subsidy?
terested in the allocative effects of subsidies prises and the returns which government There is clearly a case for excluding pure
while a welfare economist might be in- receives from them is included in our public goods from our computation of sub-
terested in their overall welfare effects. measure of the volume of subsidies flowing sidies. But empirically where does one draw
Political economists would want to interpret through the government budget and this the line between pure public goods and merit
the allocation of subsidies in terms of their component of government subsidy is dis- goods or private goods? There are obvious
perception of the distributive coalitions cussed further below. public good candidates like defence and
which control the state. This paper does not (b) Public Goods: The wide range of police. But then there are less clear cases
belong to any of these particular perspec- general, social and economic services offered where the benefits are not immediately
tives. Instead, it undertakes an exercise which by the government at the Centre and in the tangible, such as agricultural extension, or
is a necessary first step for addressing any States can, for analytical purposes, be the beneficiaries not exclusively identifiable,
or all of these questions operationally. It at- classified into three broad groups. At one as in a literacy programme. The conservative
tempts to measure the volume and composi- end of the spectrum there are pure public rule of thumb followed in this exercise is to
tion of subsidies provided by the Central goods (services) in the Samuelson sense, treat the general administrative services in
government and fourteen major State characterised by non-rivalry and non- the functional classification of government
governments in India, as observed in the year excludability in consumption. 2 At the other expenditure as pure public services, along
4987-88, the last financial year for which end there would be pure private goods with relief on account of natural calamities,
complete accounts are so far available. The characterised by rivalry, excludability and no the general secretariat expenses of social and
first part of the paper deals with concepts externality. Then there would be the vast economic services and the compensation
and method. Part two presents estimates of majority of services in the middle category, and assignment to Local Bodies and Pan-
the volume and composition of subsidies at characterised by rivalry and excludability but chayati Raj institutions. The expenditure in-
the national level. Part three analyses inter- also varying degrees of externalities. We may curred on these items has been excluded
State variations and the main conclusions stretch Musgrave's notion to describe this from the computation of subsidies.
are summarised in part four. class of services as 'merit goods'.3 Of these, It is possible to take the view that a
the concept of subsidy is properly applicable number of other items, particularly certain
I only to the last two. social services, are also pure public services.
Subsidies and Theory of In the case of pure public goods we know To the extent that these have not been
Public Expenditure from the theory of public expenditure that eliminated, the estimated value of subsidies
the well known Samuelson pricing rules can- would be larger than the actual value.
Government subsidies may be defined as not in fact be applied because of the free Readers are welcome to apply their own
the difference between the cost of delivering rider problem. Given the characteristic of judgment on which additional social or'
various publicly provided goods or services non-excludability, consumers will not reveal economic service ought to be treated as a
(henceforth, services) and the recoveries aris- their preferences for such goods and the de- public good and use our disaggregated sub-
ing from such deliveries.1 However, a mand information necessary for calculating sidy estimates to make the appropriate ad-
number of qualifications and adjustments Samuelson prices will not be available. justments and arrive at their preferred
must be introduced before this concept can Wicksell had anticipated this problem before measure of the total volume of subsidies.
be applied to measure subsidies from the Samuelson and he, followed by Lindhal and However, it must be noted that there could
available data on government expenditure more recently Musgrave, argued that a voting be an element of hidden producer subsidies
and receipts. These are as follows. mechanism of near unanimity, choosing bet- even in pure public goods, whether they be
(a) Government and Public Sector. ween alternative expenditure proposals along supplied by government departments or

Economic and Political Weekly May 4, 1991 1157

firms, if these are not supplied cost
(c) Transfer Payments and Tax Expen-
diture: The public expenditure incurred on
transfer payments have been excluded from
the computation of subsidies since these can-
not be treated as costs incurred in the public
provision of a service which could be pric-
ed in principle. For the same reason tax ex-
penditures, i e, revenue losses incurred in tax
incentives, have also been excluded from the
computation of subsidies though these are
usually treated as subsidies in the literature.
(d) The Different Elements of Subsidy:
The concept of subsidy adopted in this ex-
ercise actually combines three different
elements of subsidy as demonstrated in the
diagram. Let OY be the quantity of some
service which is publicly provided, YB the
actual cost per unit, YD the efficient cost
per unit and EF the curve of per unit
recoveries. XX is the demand curve for the
The rectangle ABHG measures the total
volume of subsidy actually required in order
to ensure that the market absorbs OY quan-
tity of this publicly provided service if the
market clearing quantity OY' is considered
socially inadequate. However, ABHG has
two components, i e, a necessary element
CDHG which is a genuine allocative subsidy i is an imputed interest rate representing as total expenditure of the jth sector and C
and an additional element ABDC paid to the opportunity cost of money for is not a measure of the total volume of
suppliers to cover their inefficiency. Final- government; public expenditure.
ly, there is a subsidy element GHFE which d is the depreciation rate; The imputed interest rate or the average
need not have been paid to support con- yj is revenue receipts by the sector; cost of money to the government, calculated
sumption level OY, given the state of de- rj. is income by way of interest or dividend as the ratio of domestic interest payments
mand. We may therefore describe this as a on loans and equity; and by government to the stock of domestic
purely distributive subsidy. Thus, our tj is a transfer payment from the sector to public debt, works out to 6.04 per cent. The
measure of subsidy which conceptually cor- individual agents. depreciation rate has been set at 2 per cent
responds to the rectangle ABFE, in fact The total volume of subsidies on all in real terms, assuming an average life of fif-
combines three distinct elements, i e, a pro- serives is given by ty years for capital stock in government ac-
ducers subsidy, the allocative subsidy and a tivities as on March 31, 1987.5 Allowing for
distributive subsidy. However, it is not possi- an inflation rate of 7.4 per cent this works
ble to disentangle these different elements out to 10.4 per cent depreciation in nominal
of the subsidy without detailed estimates of terms.
cost and demand functions for all the dif- Similarly the cost of any service j The data used for the exercise has been
ferent subsidies. drawn primarily from the Finance Accounts
(e) Method of Computation: The exercise of the Union and State governments publish-
covers the provision of public services by the ed by the Office of the Comptroller and
Central government and fourteen major while the total cost of all services, including Auditor General. This has been supple-
State governments for the year 1987-88. In transfer payments and pure public services mented by additional information drawn
all, there are 123 major categories of public is given by from budget documents and the Indian
services or sectors of government activity Economic Statistics: Public Finance publish-
identifiable from the budget classification, ed by the ministry of finance
of which 37 sectors in general administrative The concept of subsidy employed in mis
services, etc, are treated as pure public ser- study should be distinguished from the con-
vices. For each of the remaining 86 social Notice that in calculating the cost of a ser- cepts used in the budget and National Ac-
and economic services subsidy has been vice we have added the variable cost or counts. The concept of 'subsidy' used in the
computed as revenue expenditure (net of transfer pay- budgets simply applies to the explicit
ments) of the sector with the imputed in- payments made to producers to alter their
terest cost of cumulative capital expenditure price or output decisions. The best examples
by the sector and the depreciation on capital are the food and fertiliser subsidies. The Na-
where j = 3 8 . . . 123 indexes the services. accumulated within the sector. It is this in- tional Accounting concept is broader as it
For the jth sector; terest cost and the depreciation rate which includes, in addition to these explicit
Sj is the subsidy; together constitute the element of fixed cost payments, the implicit subsidies arising from
Vj is the variable cost or revenue expen- associated with the current level of a service the tosses of departmental enterprises. The
diture on the service; and not the capital expenditure of the sector concept of subsidy employed in this study
K j is the capital stock in the sector; in the current period. That will form a com- is still broader because, in addition to the
L j is the stock of investments outside ponent of the cumulative capital expenditure National Accounts concept of subsidies, it
government by the sector in the form of which supports future deliveries of services includes subsidies to households implicit in
loans or equity; from the sector. Therefore c. is not the same the provision of social and economic services

1158 Economic and Political Weekly May 4, 1991

below cost as well as the unrecovered cost the rest through State governments, even ed in the inter-State analysis of subsidies in
of loans given and investments made in non- though the aggregate cost of social and Section 3. It is worth noting that user charge
departmental enterprises and co-operatives. economic services is more or less evenly recoveries from secondary education and,
(f) Potential Sources of Bias: It has shared between the Centre and the State especially, university or technical education
already been noted above that the volume governments. This is because the States ac- such as medicine and engineering would
of subsidy measured in this exercise may be count for the bulk of social services, which make it possible to almost double the volume
an under estimate because it excludes tax are more heavily subsidised as a matter of of subsidies in primary education even
expenditures. Another possible source of policy, while the Central government is without any increase in the total volume of
under estimation could be some services, predominant in the provision of economic subsidies. Of course, this would require
e g, higher technical education such as services. These details are discussed further associated action, such as means test
medicine or engineering, where the market below. It is this difference in the composi- scholarships and special bank loan schemes,
clearing price may be higher than the actual tion of publicly provided services which also to ensure that higher levels of education re-
cost of supply, viz, a state of demand il- accounts for a lower overall recovery rate of main accessible to the disadvantaged. These
lustrated by curve X' X' in the diagram. On 16 per cent in the States as compared to 48 issues are not pursued further in this paper.
the other hand, there are also some sources per cent at the Central level. Poverty group targeting in the allocation
of upward bias in our estimate. The possi- It should be clarified here that in of subsidies in other social services appears
bility of some pure public services not be- calculating the recovery rate of the States, to be equally weak. In health services, for
ing excluded from the computation has been receipts in the form of transfers from the instance, out of total subsidies of the order
noted earlier. In addition we must remember Centre have not been counted and these have of Rs 2,925 crore in 1987-88, less than
that if the existing level of subsidies, and also been excluded from the expenditure side Rs 600 crore flowed to the rural sector.
therefore the aggregate level of public expen- of Central government accounts. These Similarly in the case of water supply, sanita-
diture were to be reduced, then ceteris receipts and expenditures cancel out when tion and housing, out of a total subsidy of
paribus this would also reduce the level of the accounts of the two levels of government Rs 2,363 crore, only Rs 823 crore flowed to
aggregate output and the volume of revenue. are combined for a consolidated picture of the rural sector. Such an allocation of sub-
Thus, in principle, subsidies should be government finance. Even if the accounts at sidies does not even appear to be equitable,
calculated net of the revenues which they in- different levels of government are analysed let alone progressive, given that about 76 per
directly generate. Keeping in view these separately it would be odd to treat such cent of the total population and the vast ma-
possible sources of bias it must be emphasis- transfers at the Central level as expenditure jority of those below the poverty line live in
ed that the estimates presented in this paper on services which it has not delivered and rural areas.
should be regarded as nothing more than a at the State level treat them as if they were Clearly, there would be much room for
first approximation. recoveries from recipients of publicly pro- substantially increasing the volume of
vided services at the State level. carefully targeted subsidisation of social ser-
II (b) Social Services: Social services ac- vices to genuinely deserving sections of the
counted for 40 per cent of the total volume population, even without any increase in the
Level and Composition of
of subsidies or about Rs 16,760 crore in total volume of subsidies, if a serious at-
Subsidies: All India, 1987-88 1987-88. This works out to almost 6 per cent tempt could be made to prune subsidies
(a)The Volume of Subsidies: Going by this ot the GDP in that year. As noted above, flowing to unintended beneficiaries. Whether
user charge method of costing public ser- the major component of these subsidies on or not such expenditure switching is com-
vices, the total cost of all services plus social services, amounting to Rs 14,460 patible with the political economy of fiscal
transfer payments for the year 1987-88 work- crore, flowed through the budgets of the policy in India is, of course, another matter.
ed out to Rs 91,276 crore, of which Rs 48,599 State governments. These social services have (c) Economic Services: Subsidies in
crore was accounted for by the Centre and been provided virtually free to the recipients economic services amounted to Rs 25,564
the balance of Rs 42,677 crore was attri- as a deliberate matter of policy, with less crore or about 60 per cent of the total
butable to the States. Compared to the ac- than 4 per cent of the cost of these services volume of subsidies. A little over half of this
counts figures of total government expen- being recovered (Table 2). Such a policy flowed through the Central budget. Costs
diture in that year of the order of Rs 1,01,754 could be seen as an effective redistributive were not fully recovered in any economic ser-
there is a difference of about Rs 10,000 crore. measure if the subsidies were targeted to vice and the average recovery rate was less
This difference arises primarily because in reach intended beneficiaries. Experience has than 44 per cent. However, there was con-
this exercise the imputed interest cost and shown that progressive tax structures by siderable variation around this average with
depreciation on the cumulative capital ex- themselves are usually not very effective recovery rates varying from as little as 20 per
penditure shown in the accounts has been redistributive instruments, whereas the ex- cent to over 75 per cent (Table 4).
taken as the fixed cost element instead of the penditure on social services covers all the The highest recovery rates of 75.7 per cent
actual capital expenditure in (987-88. non-food basic needs items which are known and 70 per cent were recorded in items like
Transfer payments, including the alloca- to be highly correlated with welfare in- transport and communications. Disaggre-
tion for employment programmes, amounted dicators in the physical quality of life gated data show that these rates were in fact
to Rs 3,836 crore in 1987-88 and the cost of index.6 higher for some items, e g, 95 per cent in
pure public services (general services) ac- However, the data presented here does not the case of rail transportation. However, the
counted for another Rs 25,000 crore. The indicate that the subsidisation of social ser- sector averages were brought down by very
balance Rs 62,440 crore would have been the vices is being effectively targeted towards low recovery rates of around 3 per cent in
total user charge on social and economic ser- disadvantaged groups. Take for instance edu- other items like roads and bridges. Given the
vices provided by the Central and State cation—the single largest item of subsidies critical role of infrastructure like transport
governments, if these services were not sub- which alone accounted for Rs 9,576 crore or and communications and their relatively im-
sidised. In fact only 32 per cent of the cost 23 per cent of all subsidies (Table 2). Much pressive performance in cost recovery, it is
of these services was recovered, thus leaving less than half of this was spent on primary unfortunate that the share of these sectors
a subsidy element amounting to Rs 2,324 education. The major component of Rs 5,460 in aggregate public expenditure has tended
crore or almost 70 per cent of the cost of crore was spent on secondary and higher or to diminish7 in recent years.
these services. As a proportion of GDP this technical education, sports, art and culture We next come to power and energy which
works out to about 15 per cent (Table 1). (Table 3). In our view, this reflects rather is generally believed to be a major area of
A little over a third of this total bill of sub- weak targeting of the disadvantaged in a hidden subsidies. Admittedly, the recovery
sidies, adding up to about Rs 16,065 crore, situation where 64 per cent of the popula- rate of around 35 per cent in this sector is
flowed through the Central government and tion is illiterate. This issue is further discuss- only about half of that observed in transport

Economic and Political Weekly May 4, 1991 1159

and communications. However, it is much culture and irrigation. themselves from the government. These are
higher than the recovery rates recorded in The single largest item in the bill of sub- shown separately for departmental enter-
services relating to agriculture or industry sidies to agriculture is the food subsidy which prises, non-departmental enterprises and
and the subsidy of Rs 3,221 crore to power amounted to Rs 2.512 crore in 1987-88, Here, co-operatives in Table 5.
and energy, in fact, accounted for less than a question arises as to whether the difference Subsidies to public enterprises added up
8 per cent of the total volume of subsidies between the cost of grains to government, to Rs 15,080 crore or a little over a third of
in 1987-88. Much of this covered the losses calculated as a mark up on the procurement the total volume of government subsidies in
of State Electricity Boards. price, and the issue price of grain in the 1987-88. Of this, Rs 9,213 crore went to Cen-
The average recovery rate in services public distribution system should really be tral public enterprises whereas the State level
related to industry was only about 25 per treated as a subsidy to crop production or enterprises received Rs 5,866 crore worth of
cent and the volume of subsidies close to to consumers. This will make no difference subsidies. The average recovery rate was only
Rs 5,000 crore. However, of this over to the total volume of subsidies, but it will 55 per cent for the public enterprises sector
Rs 2,000 crore flowed as subsidies to the fer- effect our assessment of the incidence of as a whole, while the average rate for State
tiliser industry alone and it is debatable subsidies. This question is discussed further level enterprises was still lower at 41 per cent.
whether this element should be treated as a below. In other words, far from contributing a
subsidy to the industry or to the activity of Apart from food, the other important net surplus to the revenues of the govern-
crop production. This issue is taken up fur- items of subsidy to agriculture include ment, the public enterprises have remained
ther below. Apart from fertilisers the other various types of rural development and a major source of resource drain from the
industries which absorbed substantial sub- special area programmes (Rs 1,397 crore), government. In the present fiscal crisis this
sidies in 1987-88 include village and smalt crop husbandry (Rs 1,105 crore), animal calls for a major policy reform vis-a-vis the
industries (Rs 640 crore), engineering and husbandry (Rs 472 crore) and agricultural public sector. Ways must be found of
telecommunication equipment (Rs 490 research, etc, (Rs 384 crore). The subsidy in hardening their budget constraint and en-
crore), consumer industries (Rs 490 crore) major and medium irrigation, minor irriga- suring some improvements in their financial
and atomic energy (Rs 342 crore). tion and flood control worked out to performance so that they at leas cease to
Finally, we come to agriculture and co- Rs 2,679 crore, Rs 1,362 crore and Rs 327 drain financial resources from the govern-
operation. The cost of these services, taken crore respectively. ment, even if they are not able to immediate-
along with irrigation and flood control, was (d) Subsidy to Public Enterprises: We turn ly contribute a net surplus to the revenues
close to Rs 15,000 crore Only about 20 per now to the interface between government of the government.
cent of this cost was recovered, leaving a sub- and the public enterprises. It was explained It is interesting to note in this context that
sidy element of around Rs 11,554 crore. This in Section I that the subsidies estimated in there is considerable variation between the
works out to a little under half the total this paper are only the subsidies flowing recovery rates from different types of public
volume of subsidies in economic services. from government proper. Subsidies extend- enterprises. The recovery rate from co-
The bulk of this subsidy in services related ed by public sector enterprises to the rest of operatives is the lowest at 20 per cent.
to agriculture obviously flowed through the the economy are not estimated. However, we However, since the total cost incurred on this
State budgets since they account for an over- do estimate the extent of net budgetary sup- category of enterprises is quite small, sub-
whelming proportion of the outlay on agri- port or subsidy to the public enterprises sidies to co-operatives account for less than

1160 Economic and Political Weekly May 4, 1991

2 per cent of total subsidies. The more im- tion though, in fact, it is likely to be lower. cidence of direct and indirect taxation, which
portant contrast is between non-departm- is likely to be lower for the rural sector.
The other two items are food and fertiliser
ental enterprises and departmental enter-
which respectively account for about 6 per
prises which account for 16 per cent and 18
cent and 4 per cent of all services. In the
per cent of total subsidies respectively. The
recovery rate from the former is only about
basic classification the food subsidy is shown Inter-State Analysis of Budgetary
under agriculture which is included in the Subsidies
30 per cent as compared to an average
rural sector. However, it is arguable that
recovery rate of 67 per cent realised from the
much of this subsidy flows to the urban sec- The analysis of subsidies at the all-India
latter. Thus, the rate of resource drain is
tor since a major portion of the actual off- level presented above cannot address a
much higher in the case of non-departmental
take of subsidised foodgrains from the number of subsidy related issues which come
enterprises as compared to the departmen-
public distribution system actually goes to into focus only when the data are analysed
tal enterprises. This is despite the fact that
consumers in urban areas. In the case of the at the level of the States. For example, the
the former includes all the oil companies
fertiliser subsidy, on the other hand, though problem of resource inadequacy is parti-
which have been enjoying windfall gains
in the basic classification it appears as a sub- cularly severe at the State level8 and this
because of the oil shocks. If these were ex-
sidy to the fertiliser industry, it can be argued underlines the urgency of targeting subsidies
cluded, the recovery rate from non-depart-
that the beneficiaries of this subsidy are real- for the intended groups and making ade-
mental enterprises would be even lower.
ly the farmers belonging to the rural sector. quate cost recoveries from those with higher
(e) The Rural Share of Subsidies: purchasing power so that the prevailing levels
Measurement of fiscal incidence or the in- Estimate 11 in Table 6 gives an upper
of social and economic services which are
cidence of taxes and subsidies remains one bound estimate of the rural share in sub-
abysmally low can be expanded to satisfac-
of the most intractable problems in public sidies by including both the food and ferti-
tory levels and equitably distributed.
finance and certainly no firm measure of the liser subsidies along with the rural share of
the education subsidy. Estimate 111 gives a (a) Inter-State Analysis of Subsidy: As
incidence of subsidies can be culled out of
lower bound estimate which includes the indicated in section 2, the total cost of pro-
the present data. Nevertheless, some very
broad contours of the pattern of subsidy in- rural share of the education subsidy but not viding public services and transfers in the
cidence have been indicated such as the share the food and fertiliser subsidies. Estimate IV States in 1987-88 amounted to Rs 42,677
of social and economic services, the share is our preferred estimate which includes the crore. The cost of general and administrative
of public enterprises and so on. We now pre- rural share of the education subsidy and the services was Rs 8,070 crore and transfer
sent an estimate of the share of the rural fertiliser subsidy, but not the food subsidy. payments amounted to Rs 3,186 crore Of
population in total subsidies. With these assumptions it turns out that the the total cost of social and economic services
Each item of subsidy has been classified rural share lies between 41 per cent and 53 of Rs 31,422 crore, cost recoveries amounted
as rural or non-rural based on the evidence per cent of the total volume of subsidies. to Rs 5,162 crore, leaving the subsidy
available in the budget documents about the Our preferred estimate places it at about 46.5 amount of Rs 26,259 crore or 7.9 per cent
identity of the beneficiaries. However, there per cent. It may appear that some rural: of GDP. The subsidy amount formed over
are three major items where such an unam- urban inequity is implied here since the rural 62 per cent of the total cost of public ser-
biguous classification was difficult. The share is less than in proportion to its share vices and transfers.
largest item is education which accounted of population and per capita incomes are The most notable feature of inter-State
for 23 per cent of all subsidies as indicated also lower in the rural sector. However, any distribution of subsidies presented in
earlier. It has been assumed here that the such inference about fiscal incidence would Table 3.1 is its inequitable spread. It is clearly
flow of education subsidy to the rural sec- be premature without taking into account seen that more than a proportionate share
tor is in proportion to its share of popula- the flow of transfer payments and the in- of subsidies accrued to the high and middle

Economic and Political Weekly May 4, 1991 1161

income States. The four high income States attributed to the difference in recovery rates Rs 14,460 crore, forming about 55 per cent
with only 20 per cent share of population since they do not seem to follow any syste- of the total subsidy flowing through State
claimed almost 26 per cent of the subsidies, matic pattern (see, Table 7) consistent with governments. Among the social services,
whereas the share of the five low income differences in subsidy levels. In fact, recovery subsidy in education alone constituted over
States with over 46 per cent of population rate in the middle income States was only 32 per cent of the total subsidy, while the
was only about 38 per cent. In fact, ail the 12 per cent whereas, in the low income States subsidy to protective and preventive health
high and middle income States with the sole it was 17 per cent. Nevertheless, in the States care (medical, public health, water supply
exception of West Bengal claimed a share of of Gujarat, Kerala, Punjab and Tamil Nadu, and housing) constituted another 18 per
subsidies higher than their population share. the higher subsidy levels have to be partly cent.
Similarly, in each of the low income States attributed to their lower recovery rates. In The estimates presented in Table 8 show
with the exception of Rajasthan, per capita contrast Haryana, Karnataka and Maha- that in each of the 14 major States, social
subsidies were lower than the all-States rashtra present cases where subsidy levels, services claimed a predominant share of sub-
average. While, for the high income States were higher despite relatively high recovery sidies ranging from 47 per cent in Haryana
taken together per capita subsidies amounted rates and among the low income States, to about 68 per cent in Kerala. The broad
to Rs 466, the corresponding figure for the Bihar and Madhya Pradesh present cases of similarity in the relative shares of various
low income States was just about Rs 299. low subsidy levels with high recovery rates. sub-sectors of social services among the
This was lower than the all States average The observed pattern of higher per capita States is also notable. In every State, the
by 17 per cent. Per capita subsidies in the subsidies in more developed States clearly highest share of subsidy was in education.
middle income States amounted to Rs 384 shows that subsidy levels were higher in A large share of subsidy was also claimed
which was higher than all-States average by States with higher capacity to raise revenues. by protective and preventive health care (in-
6 per cent. In other words, the federal transfer policy cluding medical, public health, water supp-
Per capita subsidies in high and middle has failed to achieve its major objective, ly, sanitation and housing) in all the States.
income States were larger because either the namely, offsetting the lower revenue raising Considering that social services accounted
per capita expenditures in these States were capacities of fiscally disadvantaged States. for almost 50 per cent of subsidies in many
higher or their recovery rates were lower. So In other words, Central transfers have failed of the States, it would be instructive to
far as recovery rates are concerned our to enable the fiscally disadvantaged States analyse this in greater detail.
analysis shows that recoveries as a ratio of to provide a standard 9 level of public ser- The most striking feature that emerges
the cost of social and economic services vices at a uniform tax-effort. Consequently, from the analysis of subsidies in education
were, by and large, very low with an average the residents in fiscally disadvantaged States and health is that, generally, per capita sub-
of 16 per cent for the States taken together. have had to be satisfied with lower levels of sidies were higher in the States where the
In eight States, it was less than 15 per cent, services as well as lower subsidy levels than levels of educational and health services were
the lowest being about 6 per cent in West their counterparts in the better off States. also higher and vice versa. In the case of
Bengal Only in four States, it was higher (b) Subsidy in Social Services: Subsidies education, for example, per capita subsidies
than 20 per cent. However, inter-State dif- in the provision of Social Services in all the were higher in States where the literacy rates
ferences in subsidy levels cannot be largely major States taken together amounted to were higher. In Kerala, both the literacy rates

1162 Economic and Political Weekly May 4, 1991

and per capita subsidies were the highest. of subsidies in preventive health care services of population who have better access to
Similarly, in the States of Gujarat, Karna- also. social services get them virtually free, and
taka, Maharashtra, Punjab and Tamil Nadu Thus it is seen that per capita subsidies hence, appropriate large consumer surpluses,
where literacy rates were higher than the all- in social services were larger in more while the vast majority do not even have ac-
States average, the per capita subsidies wer developed States. What is more, even within cess to these services, let alone, availing the
also substantially higher. Subsidy levels were the States the benefit of subsidies is concen- subsidies involved in their delivery. Ensur-
the lowest in Bihar, Madhya Pradesh, Orissa trated to a small proportion of the popula- ing greater accessibility to larger proportion
and Uttar Pradesh all of which had very low tion. Even in less developed States, although of population involves both better targeting
literacy rates. per capita subsidies were lower, it is probable and massive expansion in the levels of these
A similar positive association between that the benefit of subsidies accrues mainly services. Given the severity of the resource
levels of the service and per capita subsidy to a smaller proportion of population which constraint with the States, expansion in the
is also noticed in the case of preventive and is literate. Therefore, per capita subsidy levels of services can come about only by
protective health care (medical, public received by the benefiting group may not be charging higher user charges on those con-
health, water supply, sanitation and hous- very much lower even in less developed sumers having higher purchasing power. In
ing). In the States of Haryana, Karnataka, States. The more literate who also have fact, in the case of higher education and
Kerala, Maharashtra, Punjab and West greater purchasing power seem to have bet- technical education, there is no reason why
Bengal the infant mortality rate were very ter access to social services and, therefore, greater recoveries cannot be made from
low indicating substantially higher than the it would be reasonable to infer that the economically better off consumers. At the
average availability of health care services. benefit of subsidy in social services accrues same time, it is necessary that the benefits
These were also the States with higher per mainly to this small and relatively privileged of these services should be made accessible
capita subsidies in protective (medical and proportion of population. at subsidised rates to those who are econo-
public health) health care services. In Kerala, data presented in Table 9 also show mically disadvantaged.
which had the lowest infant mortality rate that the recovery rates in social services were We may now look at some equity aspects
(27 per cent per 1,000 births), per capita sub- extremely low in all the States, only 5 per of the subsidy to the education sector. As
sidy in protective health care was higher than cent or less. The recovery rates were very low mentioned above, education accounts for
the average by 33 per cent. In Punjab where both in education and health sectors. Clear- almost a third of total subsidies at the State
per capita subsidies were higher than the ly, the low recovery rates reflect a deliberate level. The composition of subsidies in
average by 54 per cent, the infant mortality policy of providing these services free or at various sub-sectors within the education sec-
rate was 29 per cent lower than the average. very low prices. However, the consequence tor presented in Table 9 points to a number
Similar pattern can be observed in the case is that small and relatively privileged section of important inferences. First, in spite of the

Economic and Political Weekly May 4, 1991

fact that almost 65 per cent of the people amounted to almost Rs 1,500 crore.10 To sidy. What is implied is the need to proper-
in the States are illiterate, the allocation to this has to be added an additional amount ly target the subsidies on higher educational
primary education was just about 48 per of Rs 210 crore on account of agricultural levels. These statistics sharply underline the
cent. Thus, more than a half of the subsidies education and Rs 190 crore due to medical inequitable allocation of subsidies not mere-
in education is allocated to higher levels. The education. Thus, the total subsidy bill in- ly in terms of the regional spread but also
pattern was broadly similar in all the States, volved in higher levels of education amounts in terms of the distribution between the bet-
the share of primary education ranging from to a staggering Rs 1,900 crore. It may be ter off and the worse off within the regions.
39 per cent in Haryana and West Bengal to noted that complete cost recoveries at higher Reduction in the subsidy to the privileged
57 per cent in Bihar, Madhya Pradesh and education levels can augment the primary groups can be achieved only by enhancing
Orissa. outlay on education almost by 50 per cent. recoveries on higher education. It is in-
The bill of subsidies on higher, technical Of course, this is not to imply that teresting to note that recovery rates on higher
and other education which accrues largely economically weaker sections availing higher education for the States averaged only 1.7
to the literate sections of population educational facilities shoud not receive sub- per cent, which was lower than even the

Notes: Estimate I: Unadjusted Estimate: Includes food subsidy but not fertiliser subsidy or any share of education subsidy.
Estimate II: Maximum Estimate: Includes food and fertiliser subsidy plus share of education subsidy.
Estimate III: Minimum Estimate: Excludes food and fertiliser subsidy but includes share of education subsidy.
Estimate IV: Preferred Estimate: Excludes food subsidy but includes fertiliser subsidy and share of education subsidy.

1164 Economic and Political Weekly May 4, 1991

Economic and Political Weekly May 4, 1991 1165
1166 Economic and Political Weekly May 4, 1991
Economic and Political Weekly May 4, 1991
1168 Economic and Political Weekly May 4, 1991
Economic and Political Weekly May 4, 1991 1169
1170 Economic and Political Weekly May 4, 1991
recovery rates on secondary education. 11 sidy are distributed equitably. Equally wor- Nadu whereas the share to non-depart-
Except in Gujarat and Kerala where the rates rying consequence of improperly designed mental enterprises was larger. This is main-
were a little over 5 per cent and 7 per cent subsidy schemes is the possibility of over use ly due to the large subsidies accruing to the
respectively, all the States had recovery rates of water resources and undesirable changes electricity boards in the States. The share of
lower than 3 per cent. In as many as five in the cropping pattern induced by subsidis- co-operatives in total State subsidy averag-
States, it was even less than 1 per cent. In ed irrigation. Of course, this is not to argue ed to about 1 per cent in the States taken
technical education too, the recovery rate that subsidising irrigation per se is together and it was generally low in all the
was only 5.7 per cent on the average and 5 undesirable. What is implied, however, is the States.
per cent in eight States including the need to ensure that the objectives of such In no State was the recovery rate high
economically more advanced States of subsidisation should be clear and it should enough to meet the entire cost of providing
Gujarat (3 per cent), Haryana (1.8 per cent) not result in unintended resource misalloca- the services, in the case of either departmen-
and Punjab (3 per cent), the rates were lower tion. With regard to other subsidies in tal or non-departmental enterprises or co-
than 5 per cent. agriculture and allied activities, the shares operatives. The average recovery rates in
Apart from the stated equity considera- of agriculturally advanced States of Gujarat, departmental enterprises (46 per cent) was
tion highlighted above, the low recovery rates Haryana, Maharashtra, Punjab and Tamil higher than in non-departmental enterprises
in social services also have an unfortunate Nadu were much higher than their popula- (32 per cent) and co-operatives (29 per cent).
dynamic implication. It has been noted else- tion shares. In the power sector also a very This pattern however was not uniform across
where that the expenditure on social services high percentage of subsidies went to agricul- the States. In fact, the recovery rates showed
has been growing faster than both general turally advanced States like Haryana, Punjab very wide inter-State variations with respect
and economic services. If the recovery rates and Tamil Nadu, largely on account of the to departmental and non-departmental
continue to remain at such low levels, it abysmally low rates of power tariff levied on enterprises as well as co-operatives. In the
follows that both inter-regional and inter- electricity consumed for irrigation purposes. case of departmental enterprises, the rate
personal inequity in the allocation of sub- Irrigation and power, along with road varied from 19 per cent in Punjab to about
sidies will increase over time. Better targeting transport, constitute three important econo- 92 per cent in Bihar. Similarly, the variation
of subsidies in social services, perhaps mic services accounting for about 29 per in non-departmental enterprises ranged from
through differential pricing should, there- cent of total State subsidies mainly because less than 3 per cent in Rajasthan, Tamil
fore, constitute an important item on the of low recovery rates. In irrigation in all the Nadu and West Bengal to more than 90 per
agenda on fiscal reform. States cxcept Maharashtra (41 per cent) the cent in Andhra Pradesh and Karnataka.
recoveries were less than a third of the cost (e) Subsidy to the Rural Sector. As ex-
(c) Subsidy in Economic Services: The and lower than 10 per cent in Bihar, Kerala, plained in section II, we have constructed
quantum of subsidies in economic services Madhya Pradesh and Orissa. The average three different estimates of the share of sub-
amounted to Rs 11,800 crore, forming about recovery in the States taken together as just sidy accruing to the rural sector, according
45 per cent of the total bill in the 14 major about 21 per cent. In the power sector, in all to three alternative definitions. The estimates
States taken together. The largest component the States except Karnataka and Kerala, the are presented in Table 12. In the aggregate,
of this amounting to Rs 4,465 crore was ab- volume of subsidies on account of recoveries the subsidy accruing to the rural sector
sorbed in irrigation and another Rs 4,010 was substantial. The all-States average re- amounted to Rs 15,713 crore according to
crore was in agriculture and allied activities. covery rate was about 28 per cent. However, the narrowest definition and Rs 16,363 crore
Other important sectors involving significant in Bihar, Gujarat, Rajasthan, Tamil Nadu according to the broadest definition. This
subsidies include irrigation, power and and Uttar Pradesh there was virtually no formed about 60 to 62 per cent of the total
transport and communication sectors. These recovery and in Punjab and West Bengal the bill of subsidies in the States. Substantial
together accounted for almost Rs 7,600 crore recovery rate was as low as 6 per cent. In inter-State differences were also seen in sub-
of subsidies. fact, in Punjab per capita subsidy in the sidies per rural person ranging from about
The inter-State variation of subsidies in power sector at Rs 123 was about 6 times the Rs 213 in Uttar Pradesh in alt alternatives
economic services presented in Table 10 average. In the transport sector the average to over Rs 525 or Rs 533 in Haryana, depen-
again points towards a large concentration recovery rate was only 13 per cent and in as ding upon which definition was considered.
of subsidies in the more developed States. many as 9 States including the more advanc- The point to note, however, is that even
In Punjab, per capita subsidy on economic ed States of Gujarat and Kerala, recovery if we take the broadest definition, the share
services amounted to Rs 326 which was more rates were less than 5 per cent of the cost. of subsidies accruing to rural areas was
than 3.3 times the amount in Bihar, the least (d) Budgetary Subsidy to Public Enter-
developed State and about two times the prises: The flow of subsidies to public sec-
average. In Gujarat and Haryana, the sub- tor enterprises as a whole have been analysed
sidies were higher than the average by 57 per in section II. We now take up the flow of
cent and 65 per cent respectively. On the subsidies to these enterprises at the State
other hand, as mentioned above, per capita level. Table II presents the subsidies given
subsidy in Bihar amounted to only Rs 98 to departmental and non-departmental
and in Uttar Pradesh and Madhya Pradesh enterprises as well as to co-operatives in 14
at Rs 126 and 156, it was lower than the major States. In the aggregate, the total sub-
average by 18 per cent and 4 per cent sidy accruing to public enterprises and co-
respectively. operatives amounted to Rs 5,866 crore, for-
The inter-State distribution of subsidies ming 22 per cent of the total subsidy given
in some important economic services also at the State level. However, this share show-
points towards the inequitable pattern ed wide variation across States, ranging from
observed above. In irrigation, significantly 7.5 per cent in Bihar to over 38 per cent in
larger than the average per capita subsidies Punjab.
accrued to the residents of better off States In the aggregate, the largest share of State
like Gujarat, Haryana and Punjab. However, subsidy, constituting almost 14 per cent, ac-
higher levels of subsidy were also seen in crued to departmental enterprises, while the
some of the poorer States like Madhya subsidy to non-departmental enterprises
Pradesh, Orissa and Rajasthan. At the same constituted about 7 per cent. Departmental
time, within the States there is no evidence enterprises claimed a larger share of subsidy
to show that the benefits of irrigation sub- in all the States except Punjab and Tamil

E c o n o m i c a n d Political Weekly M a y 4 , 1991

much lower than the share of rural popula- much lower, and the rest of the economy. In 2 See P Samuelson, 'The Pure Theory of
tion in every State except Haryana where the all these cases the disadvantaged seem to be Public Expenditure', Review of Economics
two shares are more or less equivalent. In getting less than their proportionate share and Statistics, Vol 36, 1954, and 'Diagram-
the aggregate, whereas the share of rural of subsidies. matic Exposition of a Theory of Public Ex-
population was over 76 per cent, the share Admittedly, this in itself cannot be taken penditure', Review of Economics and
of subsidies accruing to the rural sector was as conclusive evidence that the overall fiscal Statistics, Vol 37, 1955.
just about 62 per cent. The maximum dif- system is regressive. For that the incidence 3 See R A Musgrave, "On Merit Goods',
ference was in Punjab where the subsidy of taxation and expenditure and the other Public Finance in a Democratic Society,
share was lower than the population share indirect effects of fiscal policy must also be Wheatsheaf Books, 1986, Vol 1, Chapter 3.
by over 25 percentage points. Very large dif- taken into account. But clearly, where sub- 4 See R A Musgrave and A Peacock (ed).
Classics in the Theory of Public Finance,
ference in the shares was seen also in Orissa, sidies are concerned, it is necessary to reform
London, Macmillan, 1958; R A Musgrave,
Tamil them in a more egalitarian direction. Our
Nadu, Kerala, Uttar Pradesh, Maha-
Samuelson on Public Goods in E C Brown
rashtra and West Bengal. It may be noted estimates show that with greater transparen-
and R Salow (ed), Paul Samuelson and
that these only indicate a broad pattern of cy and better targeting it should be possible
Modern Economic Theory, McGrow Hill,
subsidies accruing to the rural sector. In to significantly increase the flow of services New York, 1983 and K Arrow, Social Choice
order to arrive at firm conclusions, however, as well as subsidies to disadvantaged groups and Individual Values, Wiley and Co, New
a detailed analysis o f the incidence of sub- without any increase, perhaps even with a York, 1951.
sidy would have to be undertaken, which is reduction, in the total bill of subsidies. 5 A rough estimate of the average life of dif-
not attempted in this study. This can be done provided the leakage to ferent types of fixed assets attempted by the
unintended beneficiaries is plugged. This CSO shows that these range from as little as
IV particular implication is of immediate 10 to 30 years in the case of machinery and
relevance in the context of the fiscal im- transport equipment to 80 or 100 years in the
Concluding Remarks balance and negotiations with the Interna- case of construction such as buildings, dams
In this paper an attempt has been made tional Monetary Fund which are likely to and other construction works. See Estimates
to estimate the total volume and composi- lead to a major fiscal squeeze from the next of Capital Stock of Indian Economy; Cen-
financial year. Careful advance planning is tral Statistical Organisation, Ministry of
tion of government subsidies in India in the
necessary if we are to protect those who are Planning, Government of India, New Delhi,
year 1987-88, after costing government ser-
already vulnerable from bearing the further 1988. Checks indicated that the subsidy
vices on a user charge basis. The exercise
costs of adjustment. estimates are not very sensitive to the assum-
shows that the actual volume of subsidies
ed life of capital assets.
was huge, amounting to Rs 42,324 crore or We must also reconsider in this context the
6 See Sudipto, Mundle 'The Human Element
almost 15 per cent of the GDP. issue of budgetary support to public enter-
in India's Economic Development'. Paper
Pure transfer payments are transparent prises. Our estimates show that over 35 per presented at the North South Round Table
and their beneficiaries are explicitly targeted. cent of government subsidies have been at Istanbul, September, 1985 and reprinted
Unfortunately, such direct transfer payments flowing to these enterprises. Given exter- in K Haq and U Kirdar (ed), Human
are stilt relatively small in India. By com- nalities and missing markets, there is no Development: The Neglected Dimension,
parison the total volume of subsidies in question that public enterprises must play Islamabad, 1986. On the limits of tax policy
1987-88 was more than ten times as large and a major role in any programme of industria- as a redistributive instrument in developing
it turns out that the bulk of this subsidy was lisation. However it is worth asking whether, countries, See Richard Goode, Government
not visible. The explicit subsidy, as revealed even after 40 years of protected domination Finance in Developing Countries, Brockings
in budgets for 1987-88, amounted to only of the commanding heights of the economy, Institution, 1984.
Rs 5,982 crore Even by the broader National these enterprises should still remain depen- 7 See Sudipto Mundle 'Pattern of Public Ex-
Accounts definition, the volume of visible dent on budgetary support. Oven if they are penditure in India: A Financial Perspective
subsidy worked out to only Rs 11,795 crore not immediately able to pay back to govern- of the Developmental State Paper presented
or about 28 per cent of the actual volume ment an adequate return on its investments, at a conference on The State and Interna-
of subsidies. surely they should at least pay their own way, tional Linkages', The Hague, October, 1988.
especially when the opportunity cost of 8 During the Seventh Plan, for example, the
There can, of course, be differences in
budgetary support to these enterprises may States' actual plan expenditure (Rs 74,519
judgment about whether or not a part of this
have to be measured in terms of forgone crore) was lower than the planned outlay by
includes expenditure on pure public services, about 8 per cent. This shortfall was largely
on what should be the correct interest rate wages for unemployed agricultural labourers
due to below target budgetary contributions.
or the appropriate depreciation rate and so in government employment programmes.
It may also be noted that during the eighties
forth. But none of this can detract from the the severe resource constraint has caused a
essential fact that a substantial proportion substantial deceleration in the growth of
of the G D P is being distributed in the form Notes capital expenditures at the State level. On this
of subsidies through the budget, much of it see, Rao, M G, and Tulasidhar, V B, Public
invisible, and that it is not at all clear that [The paper has been prepared at the instance
Expenditure in India—Emerging Trends,
of the Planning Commission. Research
these subsidies are flowing to intended NIPFP (Mimeo), 1991.
Assistance from V Geetha, S Gopalakrishnan,
beneficiaries. 9 In some federations, 'average' level is taken
T S Rangamannar, G P Sahni and Dipchand
We have attempted to make this pheno- as the 'standard' level. For the shortcomings
Maity, who undertook all the computational
menon transparent by quantifying the flow in the design of general purpose transfers in
work, is gratefully acknowledged. We have also
of these subsidies, even if only as a first ap- India, see, Rao, M G and Aggarwal, V, 'Cen-
benefited from discussions with Amaresh
proximation. The resulting estimates show tral Transfers to Offset Fiscal Disadvantages
Bagchi, Raja Chelliah, Arindam Das-Gupta,
not only that the total volume of subsidies of States: Measurement of Cost Disabilities
Biswanath Goldar, Mihir Rakshit, Uma Roy
and Expenditures Needs', Indian Economic
is very large but also that it is inequitably Choudhury, V B Tulasidhar, A Vaidyanathan
Review (forthcoming).
distributed. This is the picture which comes and other participants of the Workshop on
10 This excludes a small element (about Rs 100
through if we look either at the distribution Subsidies held at the NIPFP on November 10,
crore) of subsidy on account of adult educa-
of social services between socially disadvan- 1990. However, the usual disclaimers apply
tion which really qualifies as primary since
taged groups and others or the inter-regional too. Thanks are also due to R Periannan and
this item largely consists of a basic literacy
allocation of subsidies per capita between R S Tyagi for excellent secretarial assistance.)
programme for illiterate adults.
high and low income States or the inter- 1 We ignore; for the moment, the issue of dif- 11 This is partly due to the higher recovery
sectoral allocation of subsidies between the ference between actual cost and efficiency from the activity of selling text books at
rural sector, where per capita incomes are cost of publicly provided goods or services. secondary education level in some States.

1172 Economic and Political Weekly May 4, 1991