Efficiency and Equity of Food Market Interventions

Author(s): Bharat Ramaswami

Source: Economic and Political Weekly, Vol. 37, No. 12 (Mar. 23-29, 2002), pp. 1129-1135
Published by: Economic and Political Weekly
Stable URL: http://www.jstor.org/stable/4411900
Accessed: 08-11-2017 08:03 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

Economic and Political Weekly is collaborating with JSTOR to digitize, preserve and extend
access to Economic and Political Weekly

This content downloaded from 14.139.240.201 on Wed, 08 Nov 2017 08:03:47 UTC
All use subject to http://about.jstor.org/terms
 I Special articles
Efficiency and Equity
of Food Market Interventions
This paper reviews the economic rationale of food market interventions in India, the problems
that arise in designing these policies and their performance. The evaluation points to certain
directions of reform which require that the framework of a centralised public distribution
system be set aside in favour of a regionally differentiated safety net of food subsidies sensitive
to local consumption patterns, needs and circumstances.

BHARAT RAMASWAMI

I Even with complete markets, the market between procurement and distribution)
Introduction solution in poor countries produces out-should on average be zero. The variance
comes that involve unacceptably low foodor swings in stock levels would depend on
T his paper examines the efficiency consumption for the poor. The equity the extent of price stabilization. Greater is
and equity of food market inter- objective of food market intervention is to the extent of price stabilisation, larger is
ventions in India. The paper re-augment the food consumption of such the market intervention (in terms of pur-
views the economic rationale of such in- target groups. This is accomplished by chases and sales). In an open economy, the
terventions, the problems that arise in offering subsidies on such consumption.government would coordinate its interven-
designing these policies and their perfor- A mechanism for achieving both thesetion with international trade. Empirical
mance in the Indian context. The evalu- goals is the procurement (purchase), stor-studies usually find that international trade
age and distribution (sales) operation.reduces the size of market intervention
ation points to certain directions of reform.
Suppose the subsidy scheme requires anrequired to achieve a given degree of price
It will be argued that these reforms require
that the framework of a centralised public annual distribution of 15 million tonnes stabilisation.
of grain. Annual procurement would
distribution system be set aside in favour
of a regionally differentiated safety net however
of vary depending on the size of Ill
harvest and the available stocks with the
food subsidies sensitive to local consump- Designing Sustainable and
tion patterns, needs and circumstances. government. It could be greater than 15 Successful Interventions
Dissolving the existing arrangements million tonnes in times of abundant sup-
would, however, face formidable but not plies. In such times, the government There are several problems in operating
insurmountable political economy con- builds up stocks. It could be less than 15 and designing a sustainable intervention.
straints. The scale of food subsidies and million tonnes in times of shortfall. In suchAs the discussion above suggests, the idea
its incidence in particular states would, times,
of the public distribution draws down is to stabilise prices around their mean.
stocks. With luck, such a scheme can
course, be subject to political initiative and However, the mean price (and average
interest. smooth out the inter-temporal variability supply) does not remain unchanged.
in crop supplies. Stabilisation itself changes it. Typically,
II Only in exceptional circumstances, such in a developing economy, average supply
Ideal Market Interventions as a sequence of bumper harvests or a tends to increase over time. As a result,
sequence of disastrous crop failures, would interventions that try to stabilise with
The principal efficiency argument forsuch schemes fail. Except for the limita- reference to historical supply levels and do
food market intervention arises from the tion in such rare events, successful price not adjust will tend to carry too much stock.
absence of risk markets. Private storage stabilisation should mean that over crop First, there are the exogenous factors:
of foodgrains across years is typically and weather cycles (such as 8-10 years), technological progress and Engel's law.
unprofitable. However, such storage might the average size of government purchases Supply grows faster than demand and the
be socially profitable because poor risk- would be equal to the annual distribution relative price of food declines. Second,
averse food consumers cannot obtain of 15 million tonnes. If this condition were there are the endogenous factors. Price
not met, price stabilisation would not stabilisation of food crops favours them
credit or insurance against crop failures.
be sustainable. Stocks would either be
Price stabilisation is therefore a substi- over the crops that are not supported. The
continuously augmented or depleted.
tute for missing risk markets. The reduc- supply response to price supports must not
Sustainability therefore requires that
tion in risk is beneficial to producers be underestimated. In markets where pro-
as well. the change in stocks (i e, the difference
ducers hold rational expectations of future

Economic and Political Weekly March 23, 2002 1129

This content downloaded from 14.139.240.201 on Wed, 08 Nov 2017 08:03:47 UTC
All use subject to http://about.jstor.org/terms
prices and supply, price stabilisation IV 1990s. This reversed the trend of declin-
decreases price variability and producers Stabilisation ing prices observed during the period
increase supply as a response to reduction 1970-90 (Table 2).
in risk. These effects are even greater in In this section, I consider the stabilisation
The situation is unsustainable and present
developing country markets, where with- outcome of the food market intervention. policies can continue only if the next two
out futures markets and information dis- In particular, I show the contrast between crops fail. In the next two sections, the
semination services, producers have little the decade of economic reforms and pre- paper considers the responses of producers
means of forming rational price expecta- vious decades. and consumers to food market interven-
tions. In addition to decreasing price(a) 1991-92 was the last year that public tion that exacerbate the problem of excess
variability, price stabilisation reduces ir- distribution offtake (of rice and wheat) stocks.
rational forecasting errors and substan-exceeded procurement (of rice and wheat).
tially reduces risk relative to the non-Since then procurement in every year has V
supported crops. exceeded public distribution offtake.Procurement and Speculation
If intervention does not adjust to these (b) In the period before 1991-92, procure-
long-term supply factors (by letting aver-ment and public distribution were more Once stocks are on the unsustainable
age prices fall), it will tend to accumulateevenly matched. During the 20 years path, they can build up very quickly be-
stocks. In addition, market interventions between 1972-73 and 1991-92, procure- cause of speculative expectations.1 The
develop political interests. As price ment exceeded public distribution offtake government announces a procurement price
stabilisation is eroded by the goal of in 11 years and was smaller than public
that is fixed at the same level throughout
supporting producer prices, it furtherdistribution system (PDS) offtake inthe
the
season. For a grain seller, the oppor-
distorts the allocation of resources to the
other nine years. tunity cost of sale to the government is the
supported crops deepening the problem of (c) Define AS = (Procurement - PDS market price of grain but at a later point
sustainability. Offtake). It is a measure of excess stock. in time. Thus price expectations determine
On the distribution side, a successfulThis can fluctuate widely from year tothe sales to the government. Price expec-
year. It does not include imports/exports ortations are in turn dependent on future
intervention is one where the target groups
receive the entire subsidy. Here the prac-open market sales. Nonetheless, it is a use-government actions. When government
tical issue is that while the poor can be ful measure in our contextbecause these are stocks are large, it is natural to expect
the quantities that are largely determined byfuture sales from these stocks. Indeed,
counted, it is not easy to identify them. The
targeting criteria cannot be those that canthe government's choice of administeredopen market sales are the principal means
be claimed or mimicked by the non-poorprices: the procurement and issue price. by which the government has brought down
in order to receive subsidies. Associated Net exports and open market sales are dis- stocks from unwanted levels. They reduce
with any targeting scheme are errorscretionary
of policies that are put in place only
private storage in two ways. First, because
exclusion (due to exclusion of some when stock levels are not at desired levels. future supplies are expected, smaller quan-
members of the target group) and errors (d) In an economy, where the governmenttities of private storage are sufficient
of inclusion (due to inclusion of some stabilises annual supplies over time, the(to arbitrage the seasonal price differences).
members of the non-target group). While average value of AS over a long enough Second, there is no systematic way in
an ideal targeting scheme keeps both these period ought to be zero. The average valuewhich the open market sales occur. It is
errors to the minimum (perhaps, even zero),of AS over the period 1972-73 - 1991-92very difficult to predict the timing, quan-
nobody has discovered such a scheme that is .43 million tonnes. When the period is tity and price of grain offered in open market
extended to 1999-2000, the average value
works everywhere. Usually, there is a trade-
of AS rises to 2.3 million tonnes. In the Table 1: Change in Real Prices of Rice
off between the two kinds of targeting and Wheat
errors. Liberal or loose targeting schemes,period of eight years from 1992-93 to (Per cent)
of which universal access is a polar case,1999-2000, the average excess stock was
1970-71-1989-90 1990-91-1999-2000
minimise exclusion errors but maximise 7 million tonnes.
(e) The real surprise is that at the end of Rice
inclusion errors. Conservative or tight tar- 3.23. +23

geting schemes minimise inclusion errors1999-2000, we had only about 45 millionWheat -40 +28
but increase the chances of leaving out tonnes of grain stocks. The excess stocks Source: Author's
some of the target group. The trade-off during 1992-93 - 1999-2000 sum to 56 various issues o
implies that efficient targeting design million
is tonnes. Since the stocks at the The wholesale p
wheat are deflat
the outcome of minimising a weighted
beginning of 1992-93 were about 11 million index of all commodities.
sum of exclusion and inclusion errors. For tonnes, we should have had 67 million
instance, developed countries, concernedtonnes. The discrepancy is accounted by Table 2: Changes in Average
as they are about inclusion errors, have open market sales (about 20 million tonnes Procurement and Offtake
especially in recent years, opted for con-of principally wheat) and exports (about (Million tonnes)
servative targeting schemes. Subsidies are 2 million tonnes of mainly rice) during this Rice Rice Wheat Wheat
period.
time-bound and are subject to a work require- Procure- Offtake Procure- Offtake
ment. On the other hand, in India there is (f) The primary reason for the mismatch ment ment

between procurement and PDS in the 1990s

a strong case for weighting exclusion errors 1981-82 to
more than inclusion errors. In rural India,was a jump in procurement rather than a 1991-92 9 8.4 8.6 7.8
fall in offtake (although that has been
as much as 75 per cent of the population 1992-93 t
earns an income that lies within twice pronounced in wheat) (Table 1). 1999-2000 12.6 9.8 11 6.6

the poverty line [Gangopadhyay and (g) Prices of rice and wheat, measured in Source: Author's calculations using data from
Wadhwa 2000]. constant rupees, increased during the various issues of the Economic Survey.

1130 Economic and Political Weekly March 23, 2002

This content downloaded from 14.139.240.201 on Wed, 08 Nov 2017 08:03:47 UTC
All use subject to http://about.jstor.org/terms
sales. In such situations, private storage is discounts on PDS grain falls, which leadsprices are award
also very risky. some consumers to exit the PDS and switchmarket prices .(i
Forboth these reasons, large public stocks demands to the open market. An empirical
can lead to even larger stocks in the future. analysis of these demand switches is VII
Indeed stocks can be so large that private contained in Balakrishnan and Ramaswami Food Subsidies:
stocks might not be carried at all. This is (2001). As shown in Balakrishnan and
Scale and Efficiency
what happened with wheat procurement in Ramaswami (1997), these demand switches
2001 when the government procured 19 are commonly observed in the seasonal Consider now the equity objective of
million tonnes, which is reckoned to be patterns of PDS wheat offtake. market intervention schemes. With annual
almost the entire harvest that came to the We therefore observe the following sales between 15 and 20 million tonnes of
market in Punjab and Haryana. Private sequence: higher procurement price, highergrain, the public distribution system (PDS)
trade picked up only 10-15 per cent of procurement, higher issue price, lower accounts for 15 per cent of the total avail-
supplies procured from other regions where offtake, and larger stocks. This describes ability of rice and wheat and for about 40
the minimum support price was not opera- government operations till 1997. The in-per cent of the grain that arrives in markets.
tive. As a result, stocks at the end of thetroduction of targeting which created dual The central government spends over
2001 wheat procurement season amounted prices for above poverty line (APL) house-Rs 9,000 crore on the food subsidy
to a huge hoard of 65 million tonnes, whichholds and below poverty line (BPL) house-programme. If expenditures by the state
was the quantity of rice and wheat output holds led to two opposing effects. The governments are taken into account, the
in 1971. APL households, which had to pay an total expenditures on food subsidies is in
issue price equal to the economic cost ofthe neighbourhood of Rs 12,000 crore.2
VI the FCI, moved out of the PDS reducingThe debate about the PDS does not ques-
offtake. The BPL households received a
Offtake, Quality and Price tion the need for a safety net for the poor
substantial discount on the economic cost
Transmission but whether it is an adequate and efficient
and had incentives to utilise the PDS. The mechanism for augmenting the food con-
Now consider consumer behaviour that states with a well-functioning PDS adaptedsumption of the target groups.
can cause the government to carry excess quickly to the targeted scheme and made There are two major issues with the
stocks. According to case studies and full use of the BPL allocations. These are PDS. The first issue is whether the target
arecdotal accounts, consumers perceivepredominantly the rice consuming south- groups receive significant subsidies from
the grain from the PDS to be of lower ern states of AP, TN and Kerala and they the PDS. The second issue is whether these
account for the bulk of the offtake. The
quality than the grain from private retail. subsidies are provided efficiently. Roughly
It has not been realised that such consumer principally wheat consuming states of speaking, the first issue concerns the scale
behaviour has immediate implications for Rajasthan, Maharashtra and UP which have of the PDS while the second issue con-
government stocks and food subsidy ex- weak public distribution systems were slow
cerns the efficiency of PDS expenditures.
penditures. Before we turn to the impli- to identify BPL beneficiaries. As a result,
While, for convenience, I will examine
cations, note that as government and pri- wheat offtake fell drastically under the these issues separately, they are intimately
vate trade procure grain at similar prices targeted PDS while rice offtake has re- connected. A higher efficiency of PDS
(especially wheat), the lower quality at retail mained at its earlier levels. expenditures, scale remaining unchanged
level marks the inefficiency of government The effect of lower quality on govern- is equivalent to a higher scale, efficiency
operations. Newspaper accounts speak of ment stocks, however, remains. For a given
remaining unchanged.
quality deterioration at different points in subsidy rate, the government will carryIn recent years, evidence has steadily
the distribution chain. At the point of larger stocks than what it would if theremounted that the poor receive meagre
purchase itself, the government often were no quality differences. There is a
buys lower quality and damaged grain second implication as well. By causing Table 4: Cost of Providing One Rupee
some households to switch demands to the of Subsidy
(often because of political pressure).
(Rs)
Quality loss is severe in the commonly open market, an increase in issue prices
used methods of storage (covered and plinth also increases market prices [Balakrishnan State/Target Entire Bottom Bottom Bottom
storage). Furthermore, the FCI does not and Ramaswami 2001]. The complete Group Popu- 40 30 20 lation Per Cent Per Cent Per Cent
have the systems to implement the first- sequence is therefore: higher procurement
Andhra
in, first-out principle of inventory manage- price, higher procurement, higher issue
Pradesh 1.71 3.14 4.05 5.81
ment. As a result, their stocks often tend price, lower offtake, larger stocks, and
Maharashtra 1.82 4.02 5.72 9.05
to be old. higher market price. The sequence be-
comes a cycle when higher procurement Source: Dutta and Ramaswami (20
Quality preference means that unless the
PDS grain is sufficiently discounted, some
Table 3: Decomposition of the Cost of Food Subsidies
consumers would prefer to buy market
grain even when it is more expensive. Total Transfer to Excessive Leakages/ Transfer to
Expenditures Non-Target Costs Fraud Target Group
Consumers, however, differ in their valu-
Group
ation of quality and hence in the price
Andhra Pradesh 7778 2059 2058 1161 2477
discounts that they require to buy PDS
(26.5) (26.5) (15) (32)
grain. Suppose the government increasesMaharashtra 1883 568 295 529 468
the issue price in order to keep the food (31) (16) (28) (25)
subsidy within manageable limits. This isNote: Figures in brackets are perc
typically the response (usually lagged) to
Source: Dutta and Ramaswami (2001
hikes in the procurement price. The price population ranked by expenditure.

Economic and Political Weekly March 23, 2002 1131

This content downloaded from 14.139.240.201 on Wed, 08 Nov 2017 08:03:47 UTC
All use subject to http://about.jstor.org/terms
benefits from the PDS [Dev and plier [Dev and Suryanarayana 1991; Parikh To get an idea of their quantitative
Suryanarayana 1991; Parikh 1994; World 1994 and World Bank 2001] presumably magnitude, we draw upon examples from
Bank 2001]. In the southern states, espe- because ration quotas are limited and Andhra
not Pradesh and Maharashtra [Dutta
cially Andhra Pradesh, Kerala and Tamil available for purchase continuously. and This Ramaswami 2001a]. Suppose the
Nadu, the poor seem to make reasonable means that consumer benefits from the objective of food subsidies is to subsidise
use of food subsidies.3 This is not so PDSindepend not just on the scale of sub- the food consumption of the bottom 40 per
sidies (which are meager for poor house-
the rest of the country. Indeed, the majority cent ranked by income. Table 3 shows how
of the poorest 20 per cent of households much of the expenditures on food subsi-
holds in most parts of the country) but also
in the northern and eastern Indian states on how the subsidies impact the market dies go to the target group (the fourth
column) after accounting for targeting
does not purchase any foodgrains from theprice of grain. If, as the evidence suggests,
PDS [Parikh 1994]. errors (the first column), excessive costs
the PDS increases the market price of food,
(the second column) and leakages or ille-
The poor might be excluded or might notthen these effects may well dwarf the direct
participate for a number of reasons. First, benefits of food subsidies. gal diversions to the open market (the third
the geographical coverage of the PDS is It can therefore be concluded that, with column). These computations are based on
limited especially in the northern Indianthe exceptions of a few states, the effect food subsidy costs and estimates of in-
states. For instance, in Maharashtra, 30 perof PDS on the well-being of the poor has come transfers (from NSS consumption
cent of the poor do not use the PDS becausebeen negligible, if not perverse. To register expenditure data) in 1993-94. These fig-
of lack of access [Dutta and Ramaswamilarger benefits, the scale of the programme ures imply (dividing the income transfer
2001a]. Second, to obtain access, house-would have to expand to be a majorby the subsidy cost) that it costs Rs 3.14
holds must show proof of residence. Thissupplier of grain to the poor. Swaminathan and Rs 4 to transfer a rupee to the target
is difficult for migrants. Third, ration(2000) argues that fiscal problems should group (of bottom 40 per cent) in AP and
entitlements can be accessed only oncenot constrain expansion as food subsi- Maharashtra respectively. Table 4 presents
every fortnight. Often, poor households dodies are less than 1 per cent of GDP - the a same kind of data where we consider
not have incomes that permit savings forratio lower than that of many other different choices of the target group - the
this duration.4 Fourth, given the costs ofcountries. As a proportion of central entire population, the bottom 40 per cent,
utilising the PDS, because of factors suchgovernment expenditure, the central food the bottom 30 per cent and the bottom 20
as queues, uncertain supplies, inferior subsidy has fluctuated between 2.5 per per cent.
quality, and inconvenient location, thecent and 3.5 per cent [Swaminathan 2000]. These examples illustrate that it costs
slender subsidy (because of limited quotasThese ratios would have to increase sig- much more than one rupee to transfer a
and subsidy rate) offered in most statesnificantly, perhaps even doubled, for the rupee of subsidy to a reasonably defined
might not justify the participation of poorPDS to make a major difference to the target group. It is sometimes thought that
households. In 1993-94, the average per livelihoods of the poor. Such a recommen- targeting errors are responsible for these
capita subsidy received from purchases of dation runs counter to the prevailing con- departures from efficiency. However,
rice, wheat, sugar and kerosene was Rs 3.5 cerns about fiscal control. But surely the Table 3 makes it clear that the costs of
in rural areas and about Rs 6 in urban areas. issue is one of productivity of government delivering food subsidies to the target group
The median value of the ratio of total expenditure rather than of its scale. In are high because of targeting errors as well
subsidy to household expenditure wasthe lessmacro-perspective, the scale constraintas lapses in implementation. To see the
to food subsidies stems from the un-
than 1 per cent in both urban and rural areas relative contribution of these factors,
[Tarozzi 2000]. consider the cost-effectiveness of transfer-
willingness and inability of the govern-
ment
Another way to judge the importance of to undertake reforms of itself so that
ring a rupee of subsidy to the target group
subsidy to households would be to look
resources could be released for safetyofnet
bottom 40 per cent. Suppose targeting
expenditures;
at the subsidy in relation to the maximum were perfect in the sense of zero subsidies
income that can be transferred to targetThis brings us to the issue of efficiencyto the non-target groups and nothing else
groups by a food subsidy system. For of PDS expenditures. If there are potentialchanges. The change in cost-effectiveness
example, suppose we take 4 kgs of grain
efficiency gains, reforms within thedepends PDS on whether the savings are trans-
could expand the scale of food subsidies.
per capita per month as the desired level ferred to the target group or to the general
There are three principal reasons whybudget.
of supply from the PDS. If the market price the In the first scenario (call it case
PDS does not deliver food subsidies effi-
of grain is p, food subsidies can increase A), costs remain the same but effective-
ciently. These have to do with targeting
per capita incomes by at most Rs 4p, which ness increases while in the latter scenario,
happens when supplies are free. Com- errors, i e, income transfers to non-targetcosts fall but the target group receives the
puted this way, Dutta and Ramaswami groups, excessive costs of procurement, same amount of subsidies (case B). The
(2001a) find that the bottom 40 per storage
cent and distribution (relative to thefirst row of Table 5 reports the cost-effec-
in Andhra Pradesh receive, through private
the sector) and leakages or fraud, i e,tiveness of perfect targeting. Notice that
PDS, about a quarter of the maximum illegal diversions of subsidised grain to thesubsidy transfer still costs more than one
income transfer while in Maharashtra, open
food market. rupee because of excessive costs and
subsidies transfer to the bottom 40 per cent
Table 5: Cost of Providing One Rupee of Subsidy to the Bottom 40 Per Cent under
only about 5 per cent of the maximum Perfect Targeting or Organisational Reform
(Rs)
possible levels.
Even where publicly subsidised grain Andhra Pradesh: Andhra Pradesh: Maharashtra: Maharashtra:
Case A Case B Case A Case B
reaches the poor, the market is just as
important a supplier. Most households
Perfect targeting 1.71 2.3 1.82 2.81
Perfect organisational reform 1.37 1.84 1.46 2.2
depend on a mix of the two. In the typical
pattern, the market is the dominant sup-
Source: Author's calculations using Table 3. Case A and case B

1132 Economic and Political Weekly March 23, 2002

This content downloaded from 14.139.240.201 on Wed, 08 Nov 2017 08:03:47 UTC
All use subject to http://about.jstor.org/terms
illegal grain diversions, which together can vided grain at FCI's economic cost, BPL Using NSS household consumption
be called implementation failure.5 households receive grain priced at 50 per survey data, Dutta and Ramaswami (2001 b)
The contribution of the costs of imple- cent of FCI's economic cost (which covers evaluate this policy for Andhra Pradesh
mentation failure can be considered by a the cost of procurement, taxes, transport and Maharashtra. The results support the
corresponding hypothetical experiment. and distribution).7 Thus, the subsidies are case for subsidising coarse cereals in
Suppose these costs were reduced to zero, restricted to the BPL population.8 Maharashtra but do not endorse it for
everything else remaining unchanged. The In principle, TPDS ought to make food Andhra Pradesh. The results are divergent
second row of Table 5 reports the cost- subsidies cost-effective. However, it would even though (a) the food subsidy systems
effectiveness of organisational reform be naive to expect targeting errors to vanish. are comparable between the two states in
under case A and case B. Although neither Indeed, exclusion errors are built into the their coverage of the non-poor and (b) coarse
case is completely realistic, it can be seen implementation process. Identification of cereals are an inferior commodity group
that implementation efficiency offers at the poor is the responsibility of the state in both states. This happens because the
least as much if not greater gain as im- government, which in turn is expected to welfare gains also depend on the shares
provements in targeting. use local bodies and village panchayats for of the subsidised commodities and of
this purpose. Even though the poverty line coarse cereals in the budgets of the poor.
VIII is an expenditure-based norm, it is not In Maharashtra, poor households consume
Targeting feasible to elicit expenditures for identi- significant amount of coarse cereals and
fication. Identification would then depend correspondingly smaller quantities of su-
For public spending in general, the lit- on household characteristics such as oc- perior qualities of foodgrains. Hence, they
erature makes a distinction between broadly cupation, dwelling type and size and so on.benefit from the relative price change in
targeted and narrowly targeted programmes Even when done honestly, there is no reasonfavour of coarse cereals. In contrast, poor
[van de Walle 1998]. The idea of broad to expect that the total of such beneficia-households in Andhra Pradesh lose from
targeting is to subsidise basic consumption ries will match the BPL population in the such 'local' changes because of their con-
goods and services for all households state because (a) targeting indicators are siderable consumption of subsidised rice.
because such goods and services matter imperfectly correlated with poverty andThe general point is that self-targeting
more to the poor. As the poor spend more (b) poverty is itself measured with error. schemes cannot work on an all-India basis
on food, in proportionate terms, than the If there is an excess of beneficiaries, there
because it uses food preference patterns
non-poor, a universal food subsidy such is a problem because their BPL subsidy that are necessarily state-specific. Coarse
as the erstwhile PDS is an instance of will not be funded by the central govern-cereals are in particular not suited to a
ment. At least some of the state govern-
broad targeting. But, as we have seen, such centralised procurement, storage and dis-
programmes also benefit the non-poor ments
and might be expected to trim the num- tribution because of their limited shelf life.
are therefore expensive. Narrow targeting ber of beneficiaries by whatever means However,
to subsidising coarse cereals could
is of two types. The first is indicatormatch
tar- the BPL population. So exclusion work as a component of a regional food
geting which identifies an easily observ- errors can be expected even when the subsidy programme.
able characteristic of households that is process is faithful to its intentions.9 More
realistically, we, can expect errors also
highly correlated with low income. The IX
because of lack of interest and capture by
indicator is then used as a proxy for income Food Stamps
non-target groups.10 In addition, the dif-
to identify and target poor households. The
ference between APL and BPL prices pro-A
second approach is self-targeting. Here the feature of India's food subsidy
vides strong incentives for illegal diver-
design of subsidy is such that it is much programme is the deep involvement of the
sions to the market. For these reasons,government
more costly for the non-target groups than it and its agencies in physically
for the target group to participate in is
thenot clear that BPL targeting is the best
handling the grain. The government buys
subsidy scheme. route for target groups to access subsidies.
the grain, stores it in its warehouses, trans-
The difficulties of indicator targeting
Clearly, the success of indicator target- ports it to different depots in the country
ing depends on the correlation of the make self-targeted programmes, in which and distributes it to authorised retail out-
the relatively rich voluntarily opt out lets.
indicator with poverty. In India, indicator of Agencies of the central and state
targeting at the central level began with
the programme, particularly valuable. Self-
governments carry out this operation. An
the revamped PDS in 1992 where certain targeting in food subsidies can work by alternative to such an arrangement is the
backward regions received higher subsi- subsidising commodities consumed pri- system of food stamps. In this scheme, the
dies.6 While there is no published evalu-
marily by the poor. Rice and wheat are thepurchase, storage, movement and distribu-
main commodities that are subsidised undertion of grain is performed by the private
ation of this programme, Jha and Srinivasan
(2001) demonstrate the potential of geo-the PDS. On the other hand, coarse cerealssector. Could this be superior to the ex-
graphical targeting especially when itcomprising
is sorghum, pearl millet and maize
isting PDS? As there is no published work
are known to receive higher shares in the
done at the district level where disparities that has examined this question, consider
household budgets of the poor in several
are marked. There is therefore a strong case the factors that will matter in the cost-
benefit calculus.
regions of the country. But these com-
for treating regions differently. However,
geographical targeting was given upmodities
in are unsubsidised. Would it be A food stamp is a cash voucher, which
1997, when it was replaced countrywide
welfare improving to transfer one rupee ofbe exchanged by the recipient for only
can
by the targeted PDS (TPDS). In the new food. It is usual to restrict the list of foods
subsidy from rice and wheat to the coarse
cereals? Since the importance of coarse
programme, the PDS makes a distinction by excluding alcoholic beverages, snack
between below poverty line households foods and other processed food. Here
cereals varies by state and residence, the
answers to this question must be state-
(BPL) and above poverty line (APL) house- consider food stamps that can be only used
holds. While APL households are pro- specific as well. to purchase foodgrains. In order to

Economic and Political Weekly March 23, 2002 1133

This content downloaded from 14.139.240.201 on Wed, 08 Nov 2017 08:03:47 UTC
All use subject to http://about.jstor.org/terms
preserve comparability with the PDS, as- stamps to reduce food subsidy expendi- central government at all. For the central
sume a food stamp programme that offers tures and to target them to the poor. Thegovernment, the major consequence of
recipient the same level of subsidy as the PDS. reduction in subsidy levels (often accom-TPDS is that it ties the central government
We noted earlier how targeting errors, plished by not indexing stamp value) and subsidy to the BPL population within a
excess costs and illegal diversions erode state thereby providing a formula for the
initiation of targeting meant that the food
the cost-effectiveness of the PDS. How stamp programme was not comparabletransfer to of food subsidy funds to the states.
could food stamps be better or worse? Like earlier subsidy programmes. The small The second step that the central govern-
the PDS, a food stamp programme also literature that discusses the (adverse) ment is pushing for is decentralised pro-
requires the identification and registration consequences of food stamps in develop- curement. If this happens, the implemen-
of beneficiaries. Food stamps are therefore ing countries does not distinguish the tation of the PDS will be the sole respon-
on the same footing as the PDS with respect impacts of changes in subsidy levels (andsibility of state governments and the cen-
to targeting errors.1l Excess costs, which the new targeting schemes) from the tral ef- government will restrict itself to the
constitute a significant part of government ficiency gains in transferring food distri-
financing. The advantage of these arrange-
subsidies, cannot arise in a food stamp bution to the private sector. This work ments for the central government is that
programme since the grain is transacted therefore does not illuminate the efficiency
(a) as funding under TPDS is tied to poverty
through private markets. This is the major gains from food stamp programmes when estimates, it is bounded and will decline
gain from a switch to food stamps from subsidy level and the targeting scheme with
are falling poverty and (b) decentralised
PDS. A leakage in terms of the diversion held constant. procurement will undermine the power of
of grain is clearly not an issue with food procurement lobbies, and will thereby
stamps. However, other kinds of fraud X reduce procurement price, stocks, the
economic cost of FCI and hence reduce
might still occur. In particular, food stamps Policy Response
meant for identified beneficiaries might be foodgrain subsidies as well.
hijacked by intermediaries and sold to non- India's food market intervention is in Decentralised procurement has not found
target groups (or to target groups them- crisis. Unable to resist procurement lob-favour with the states. While opposition
selves resulting in loss of subsidy). As bies, public money in the last decade hasfrom the grain surplus states is expected,
food stamps are like currency, diversion been used to build grain stocks, subtractother states too have not welcomed it either
of food stamps might be easier in terms supplies and increase foodgrain prices. because they do not wish to confront a
of transactions costs than the diversion of Neither is there much compensation procurement
in lobby in their backyard or
because they are unsure about their capa-
grain. With respect to fraud, it is not clear, terms of an effective distribution system.
a priori, whether food stamps are moreWith the exception of the southern states, bilities in undertaking this operation. It is
advantageous than PDS. However, as food too few of the poor use the PDS and the also not clear how decentralised procure-
stamps can be numerically tagged, it might bulk of the subsidy is wastefully spent ment on will mesh with minimum support
be cheaper to inspect and investigate food transfers to non-target groups, illegal di- prices. If these are obligatory for state
stamp fraud than illegal grain diversions. versions and inefficiencies in distribution. governments as well, predominantly
Finally, all of this discussion assumes the While the PDS is the joint responsibility grain-deficit states might not save much
existence of a private retail grain-market- of the central government and the state by undertaking procurement themselves.
ing network. Food stamps might not work governments, their roles are unequal. The With these uncertainties, it is reasonable
in remote regions with poor transport links central government procures, stocks, trans- for the states to resist decentralised
although it must be acknowledged that ports and supplies grain to the state govern- procurement.
these factors work against the PDS as well. ments and absorbs the costs of these Although the evolution to a federal
This review suggests that the cost-effec- operations. Once the grain is allocated relationship
to seems, ironically, to suit the
centre more than the states, there are
tiveness of food stamps depends on its the states, it is the job of the state govern-
administrative costs (relative to the PDS) ments to 'lift' the grain and distribute it
opportunities especially for the states that
and the efficiency of the private sector. to retail PDS outlets within the state. are committed to the PDS to gain control
This is an empirical issue and can be Decisions about the major policy para- on food subsidies and to restructure them
efficiently.
established without too much difficulty. It meters (procurement price, issue price, ra- It is well known that PDS
is likely that food stamps are more cost- tion quotas) are vested with the centralperformance differs across states which
effective in some regions (perhaps urban government. Some state governments, suggests
most that local factors matter and should
therefore be taken into account in food
areas) than in others. Yet the government notably Andhra Pradesh, have participated
(and many researchers) has resisted such in policy-making (with, of course, influ-subsidy policy [Ahluwalia 1993; Dev and
institutional mechanisms. While an aver- Suryanarayana
ence restricted to their domain) by offering 1991; Dutta and
sion to markets might be one reason, a
subsidies Ramaswami 2001 a; Mooij 2001 and Parikh
in excess of the central govern-
superficial evaluation of food stamp ment ex- subsidy. But, by and large, the1994].
role Indeed, while there is enormous
periences in developing countries might of state governments is to support thescopeFCI for improving efficiency by reforms
be another. Most of the developing coun- in procurement and distribution with suchlittleas geographic targeting, self-target-
tries instituted food stamp programmesparticipation
in in policy-making, except ingby and food stamps, their design and
response to a fiscal crisis. Typically, these
way of lobbying for special interests. effectiveness are specific to local prefer-
countries (e g, Jamaica, Sri Lanka, Mexico) It is unlikely, however, that this ences,
will knowledge, infrastructure and cir-
underwent major macroeconomic reform remain unchanged. A more federal struc- cumstances. For instance, a state could
that led to a reduction of general food price subsidise coarse cereals, use food stamps
ture seems to be in the offing. The first
subsidies. In order to protect the poor, stepa in the evolution was the targeted in
PDS.urban areas, allow universal access in
food stamp programme was introduced. The principal innovation, i e, the imple-
backward areas, and temporarily increase
The objective was therefore to use food mentation of targeting does not involve the subsidy rate in regions that are
the

1134 Economic and Political Weekly March 23, 2002

This content downloaded from 14.139.240.201 on Wed, 08 Nov 2017 08:03:47 UTC
All use subject to http://about.jstor.org/terms
adversely affected by floods, drought and Notes were identified in Bihar.

other natural disasters. In a decentralised 11 If anything, food stamps might lead more of
the poor to use food subsidies because they
framework, the central government would[An earlier version of this paper was prepared for no longer have to save up to buy the twice
primarily be a funding agency and its rolea conference titled Perspectives on the Indian monthly rations from the PDS.
Economy and Evaluation of Economic Reforms
in operations would be limited to the storage
in India, organised by the Planning and Policy
of emergency reserves. Through the Plan-Research Unit of the Indian Statistical Institute, References
ning Commission, the central government Delhi Centre. For comments, I am grateful to the
could monitor the targeting performanceconference participants and in particular to J V Ahluwalia, Deepak (1993): 'Public Distribution of
Meenakshi and Arvind Virmani. Much of this Food in India', Food Policy, February 33-54.
of individual states and reward the states
paper draws on joint work with Pulapre Bala-Balakrishnan Pulapre and Bharat Ramaswami
that do well with greater subsidy allocationkrishnan and with Bhaskar Dutta. Responsibility (1995): 'Public Intervention and Private
irrespective of their poverty status for itfor errors is entirely mine.] Speculation: The Case of Wheat Procurement',
can be wasteful to allocate subsidies to Journal of Quantitative Economics, 11, No
1 Speculative expectations are not taken into2, 59-83.
poor states without regard to targeting
- (1997): 'The Quality of the Public Distribution
account in reduced form regressions that relate
performance [Mooij 1999]. States withprocurement quantity to procurement priceSystem: Why it Matters', Economic and
little interest in food subsidies will lan-
and output. For studies of speculative ex- Political Weekly, Vol XXXII, No 4, 162-65.
- (2001) 'Food Prices and the Efficiency of Public
guish with ineffective distribution. But pectations, see Balakrishnan and Ramaswami
this is an outcome that has to do more with (1995) and Ramaswami (2000). Intervention: Modelling Quality Differences
2 State expenditures in 1998-99 amounted to between Public and Private Grain Supply in
state politics than with central policies. Rs 2,840 crore [World Bank 2001)]. India', mimeo, Indian Statistical Institute, Delhi
Such a move towards a federal relation- Centre.
3 These were the only states where the average
ship in food subsidies is essential if the purchase of PDS grains by a household in the Dev, S Mahendra and M H Suryanarayana (1991):
bottom 20 per cent exceeded 9 kgs per month 'Is PDS Urban Biased and Pro-Rich?',
food subsidy system is to be flexible and
in 1993-94 (Table A3. a, World Bank (2001)). Economic and Political Weekly, October 12
contingent on local circumstances and 26 (4), 2357-66.
By contrast, these figures were negligible for
needs. It need not result in the central
Bihar, Haryana, Punjab, Rajasthan, Uttar Dutta Bhaskar, and Bharat Ramaswami (2001
government abandoning its responsibili-
Pradesh and West Bengal. In Assam, Gujarat, 'Targeting and Efficiency in the Public
Karnataka, Madhya Pradesh, Maharashtra and
ties, as some critics fear, provided the state Distribution System: Case of Andhra Prades
Orissa, the figures were between 1 to 4 kgs. and Maharashtra', May 5-11, Economic an
governments negotiate with the central Political Weekly, Vol XXXVI, No 18,
4 Conditional on purchase, top deciles purchase
government to ensure the scale of financ-
more grain per capita from the PDS than the 1524-32.
ing is commensurate with the needs of abottom deciles. See Dutta and Ramaswami - (2001b): 'Reforming Food Subsidies in India:
(2001a) as to why the usual income effect
secure safety net. Indeed, this ought to be Estimating the Gains from Self-Selection',
alone is not responsible for the regressive mimeo, Indian Statistical Institute, Delhi
the major agenda of the states. Centre.
nature of PDS purchases. Rather the finding
The major hurdle to a federal structureseems more consistent with the view thatGangopadhyay,
poor Shubhashis and Wilima Wadhwa
is the tricky issue of procurement. The (2000): 'Poverty Policy: Old Wine in an Old
households are more liquidity-constrained than
Bottle' in Surjit S Bhalla (ed), New Economic
practical difficulties in decentralising pro-rich households. Of course, the liquidity effect
depends on the income level. Policies for a New India, Indian Council of
curement are much exaggerated. After all, Social Science Research.
5 Targeting errors can also be due to lax
private trade does move grain from the Government of Andhra Pradesh (2001): 'Public
administration. However, targeting errors arise
surplus to the deficit states. It is immedi-
fundamentally because of imperfect knowledge Distribution System', www.nic.in/appds
ately feasible for the states to constitute
about household characteristics. It is therefore Jha, Shikha and P V Srinivasan (2001): 'Taking
useful to keep the costs of targeting failures to the PDS to the Poor: Directions for Further
a clearing house (which invites bids for
distinct from the costs of organisation failure. Reform', September 29, Economic and
purchase and sale from FCI as well as Political Weekly, 3779-86.
6 Indicator targeting at the state level precedes
private trade) that matches the demand central policy. Andhra Pradesh used (and Ramaswami, Bharat (2000): 'Forecasting Errors
with supply. The more durable difficulties continues to use) land ownership to target rice in the Absence of a Futures Market: Assessing
are political. There is a proposal of income subsidies in rural areas. the Speculative Efficiency of the Indian Wheat
To reduce stocks, the government temporarilyMarket', Review of Development Economics,
supports to replace procurement. But 7this 4 (2), 184-93.
cut APL prices by 30 per cent in 2001.
would only exchange the present troubles 8 If FCI's economic cost is below the market Mooij, Jos (1999): 'Dilemmas in Food Policy:
for future problems of an even greater price, the APL users would also receive About Institutional Contradictions and Vested
magnitude. The case for using price sup-subsidies. Usually, however, the market price Interests', December 25, Economic and
is below the FCI's economic cost. Political Weekly, Al 14-A120.
ports to subsidise foodgrain producers is
9 States such as Andhra Pradesh and Karnataka - (2001): 'Food and Power in Bihar and Jharkhand:
no longer as compelling as it once was. had some form of indicator targeting even PDS and its Functioning', August 25,
Stabilisation is defensible provided the before the advent of TPDS. The target groups Economic and Political Weekly, 3289-99.
operation is limited. Fortunately, there is
so identified are numerically much greater Parikh, Kirit (1994): 'Who Gets How Much From
a sustainable way of holding up foodgrain than the estimates ofBPL population. However, PDS - How Effectively Does it Reach the
the states have chosen not to bring their Poor'?', Sarvekshana, January-March.
prices above what they would otherwise Senauer, Ben (1993): 'America's Second Currency:
estimates of target groups in line with the BPL
be. After all, food subsidies have favourable
population perhaps because of the political Focus on Food Stamps', The Region, Vol 7,
effects on food demand. In the US, one difficulties of such an exercise or because of 1, 4-11, published by the Federal Reserve
concern
$ of food subsidy is estimated to increase over the exclusion errors that would Bank of Minneapolis.
arise or a combination of both. In Karnataka, Swaminathan, Madhura (2000): Weakening
food demand by 20-35 cents [Senauer the numbers of BPL beneficiaries are six million Welfare: The Public Distribution of Food in
1993]. Jha and Srinivasan (2001) find households while BPL households are India, LeftWord Books, New Delhi.
similar sorts of effects from a simulation estimated as 3.9 million [Mooij Tarozzi,
1999]. Alessandro
In (2000): 'A Bird's Eye View
model of Indian foodgrains markets. If the of PDS
Andhra Pradesh, the corresponding figures are Utilisation using the 50th Round of
11 million and 3.8 million [Government of
the Indian National Sample Survey (July 1993-
farm lobby can perceive its self-interest in 1994)', mimeo.
Andhra Pradesh 2001]. The subsidy for the
an efficient and targeted safety net, as it World
excess beneficiaries is borne by the Bank (2001): India: Improving Household
state
does for instance in the US, then it opens governments. Food and Nutrition Security, Rural
possibilities for meaningful reforms. mi10 Mooij (2001) describes the way beneficiaries
Development Sector Unit, South Asia Region

Economic and Political Weekly March 23, 2002 1135

This content downloaded from 14.139.240.201 on Wed, 08 Nov 2017 08:03:47 UTC
All use subject to http://about.jstor.org/terms