180

availability and absorption issues that affect everyone. It concludes with a workable alternative, which, 170 although short of universal, could be a way forward.

160

Why Universalisation? 150

This paper begins with the prior that a universal PDS

140

is the only option consistent fully with a rights-based

130

approach, and argues that feasible alternatives that

are more universal and less targeted are more likely to be effective in ensuring food security for the poor.4 Since a legal right must apply to all citizens with any exclusion defined precisely, targeting the “poor” or “priority” will involve definition of these terms and possible litigation. Targeting was also considered when the UPA government was enacting earlier landmark legislations. But finally, both the National Rural Employment Guarantee Act and the Right to Education Act, which govern schemes providing basic rights to citizens, are universal.

Moreover, since targeting was introduced in June 1997 after a long experience with universal PDS and since PDS is still universal in Tamil Nadu, there is evidence on this that can and should inform NFSA design. The shift to a targeted PDS was based on results of various studies in the early 1990s (Parikh 1994; World Bank 1996) showing that the then universal PDS was inefficient and did not reach the poor, mainly because poorer states in the north received less subsidies than richer states in the south. But there is now also a large body of evidence showing that targeting did no better on this and in fact increased inefficiency and leakage.

Notably, a High Level Committee (HLC) on Long-term Grain Policy, appointed by the government at a time when food inflation, grain stocks and hunger were all increasing, had in 2002 recommended a return to a universal PDS after finding that targeting had “served to blunt the efficacy of the PDS in meeting its original goal of price stabilisation, while not delivering fully in terms of the new concern to focus subsidies to the poor” (Ministry of Food and Consumer Affairs 2002).5 Subsequent data reinforce this finding. For example, the coefficient of variation of real cereal prices has increased significantly after targeting (Figure 1), mainly because public grain management after 1997 has been much less effective in reducing variability of per capita availability below variability in per capita production (Figure 2). The HLC analysis of 1997-2001 applies almost ditto to the present situation of high food inflation amidst rising food stocks. Moreover, analysis by the Planning Commission (PC) shows leakages from the PDS doubled after targeting.6 National Sample Survey (NSS) consumption expenditure surveys routinely collect information on the quantity of foodgrains purchased from the PDS.7 Reported PDS purchase in the NSS should ideally Figure 1: Cereals WPI Relative to All Commodities WPI

Economic Political Weekly

EPW

1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2009

equal PDS sales reported by the Ministry of Food and Consumer Affairs (MFCA), and the gap between these two measures the extent of leakage. The PDS was universal in 1993-94 but targeted in 2004-05, and NSS data for these years show what happened to leakages. In 1993-94, leakage in rice was 19% which increased to 40% in 2004-05; in wheat, it went up from 41% to 73% and for rice and wheat together, it increased from 28% in 1993-94 to 54% in 2004-05. Per capita per month consumption of PDS rice and wheat remained unchanged (0.99 kg in 1993-94 and 1.01 kg in 2004-05) although PDS offtake doubled and subsidy increased even more. If feasibility is judged on the ability to stabilise food prices and in terms of physical and financial leakage, as the RC does, the targeted PDS scores very poorly against the universal PDS by past official assessments. This is not to suggest that Table 1: Percentage of Households in Each Quintile Purchasing Rice/Wheat from PDS

there was no improvement in

MPCE Quintiles 1993-94 2004-05

targeting. For example, Table 1

Bottom 20% 29.2 35.5

(which gives the percentage of

Next 20% 27.8 28.1 households in each monthly Next 20% 27.1 23.2 per capita expenditure (MPCE) Next 20% 25.0 18.5 quintile that purchased any PDS

Top 20% 21.7 11.5

Total 26.2 23.0

cereal in 1993-94 and 2004-05) shows that targeting did halve PDS access by the richest 20% and did lead to some improvement in access by the poorest 20%. However, this improvement disappears as we move to the next poorest quintile. For the bottom 50% that the NAC views as the Priority group, improvement in access was only from 28% in 1993-94 to 30% in 2004-05. Since its main purpose was to improve access by the poor, this negligible improvement means that the TPDS must be judged a colossal failure, especially given the costs of massive increase in leakage.

But why was improvement in access for the poor so negligible? This was because the solution was worse than the problem. When universal, PDS prices were slightly above MSP making PDS attractive to only those who faced relatively high market prices. After targeting, while prices were reduced for those with BPL cards, the poor were denied access by this very same instrument since most did not get BPL cards. Table 2 (p 40) gives by quintile group the percentage of households with BPL/Antyodaya Anna Yojana (AAY) cards in 2004-05 and also the percentage of households which purchased from the PDS, separately for those with and without BPL/AAY cards. It may be seen that although the poor are more likely to purchase from PDS than the rich, what really determines household purchases from PDS is whether or not the family has a BPL card. Irrespective of quintile class, only around 2008-09 2010-11 10% of households without such cards made any PDS

39

In Lakh Tonnes In Lakh Tonnes Ratio of Official Offtake

self-selection among those who had BPL/AAY cards: less than 65%

Rajasthan 0.0 5.2 5.2 0.0 11.8 11.8 44.2 44.2

of such households actually purchased any cereal from the PDS.

Tamil Nadu 25.4 0.7 26.1 26.6 0.6 27.2 95.5 116.7 96.0

Both the above observations, that there are huge exclusion

Uttar Pradesh 2.6 3.7 6.3 16.7 22.4 39.1 15.6 16.5 16.1

errors in distribution of BPL cards and significant self-targeting,

West Bengal 2.3 1.3 3.6 7.4 17.0 24.4 31.1 7.6 14.8

are well known and are strong arguments for universalisation.

All-India 98.0 34.7 132.7 164.6 128.9 293.5 59.5 26.9 45.2

But, since the original argument for targeting was the north-south

caps. But even in the latter year in only four major states (Andhra

Jharkhand 1.0 0.5 1.5 5.4 2.9 8.3 18.5 16.0 17.6

Pradesh, Himachal Pradesh, Kerala and Tamil Nadu) did more

Karnataka 9.8 1.9 11.7 16.3 2.7 19.1 60.0 69.4 61.4

than 50% of households report any PDS purchase. These states

Kerala 8.4 1.2 9.6 8.6 2.9 11.5 97.1 43.2 83.6

which did better before targeting also had the lowest leakage in

Madhya Pradesh 3.3 8.1 11.4 4.3 13.2 17.5 75.5 61.5 64.9

both 2004-05 and 2007-08. At the other extreme, the poor

Maharashtra 7.2 7.8 14.9 11.2 12.7 24.0 63.7 60.9 62.2

northern states of Assam, Bihar, Jharkhand and Uttar Pradesh,

Orissa 8.6 0.0 8.6 14.6 1.3 15.9 58.8 2.7 54.1

which performed badly before targeting, continued with very low

Rajasthan 0.4 4.7 5.1 1.6 9.9 11.4 25.0 47.7 44.6

2004-05 2007-08 Rural Urban Total Rural Urban Total

Tamil Nadu 32.6 1.2 33.8 36.2 0.9 37.1 90.1 133.3 91.1

Gujarat 32.6 7.9 23.0 26.7 6.9 18.2

numbers reporting any PDS purchase and also had the highest

Haryana 4.1 5.3 4.5 12.1 5.0 9.7

leakage. Thus, instead of improving geographical utilisation of

Himachal Pradesh 52.0 15.1 46.6 69.1 41.6 65.4

the food subsidy, targeting led to much higher leakages where

Jammu and Kashmir 33.6 47.3 37.5 37.5 50.2 41.0

access was originally low.

has by far the largest percentage of population accessing PDS and

Orissa 21.6 6.7 19.1 36.1 16.7 32.5

also almost no leakage. While this is mainly due to better PDS

Punjab 0.3 0.7 0.5 9.3 4.3 7.3

governance in the state, universalisation contributes.10 An even

Rajasthan 12.8 2.0 10.0 11.4 8.9 10.8

more interesting state is Chhattisgarh which has recently moved

which had similar PDS access and leakage ratios as the all-India

EPW

average in 2004-05, had by 2007-08 reduced leakage to nil while doubling access. Among other states with low leakages, Andhra Pradesh also has near universal PDS access with almost 80% of the households with BPL cards, and Kerala is historically a low leakage state. None of these low leakage states have adopted the identification methodology of the BPL Census 2002 or the BPL cut-offs of the PC. On the other hand, all high leakage states follow the central norms and end up with low PDS access.

Clearly, universal or near universal PDS is far more efficient in enabling access and preventing leakages than poverty-count based targeting. But do we have enough grains for a universal system? Is it true that universalisation is not feasible?

How Much Grain Is Required for a Universal PDS?

With population projected at 119 crore at the end of 2010-11, a universal PDS with per capita entitlement of 7 kg would commit the government to supply 100 million tonnes of grain if the entire population were to lift 100% of its quota.12 It is numbers such as this which scare. But assuming a 100% offtake is highly unrealistic. There are bound to be sections of the population who will not take their foodgrain quota because they are well off and can do without PDS grain which is generally of inferior quality. People most likely to exclude themselves because of this reason are the better off in urban areas and surplus farmers in rural areas who consume their own produce.13 There will also be others who are dissuaded by access problems such as the distance from PDS shops and their infrequent opening.

While it is obvious that there will not be 100% offtake of foodgrains, it is not easy to get a concrete idea of what percentage of population will voluntarily opt out of the PDS. This will be a function of how cheap PDS grain is compared to the market, the economic status of the households and their tastes and preferences, and also reliability of PDS supply and ease of access. In principle, it should be possible to get an idea on this from NSS consumption expenditure surveys (CES) which report how many households purchase grain from the PDS, and how much. But, since PDS became targeted in 1997, the latest NSS data available on universal PDS is for 1993-94 when only 26% of households actually made any PDS purchase. This is not good evidence because 1993-94 was quite some time ago when issue prices were higher than MSP and the coverage was low. More recent NSS data also give the percentage of households actually purchasing from the PDS, but these relate to the targeted PDS and may therefore not be a good proxy for future projection with universal entitlements. For the record, only 23% in the country as a whole reported purchasing cereals from PDS in 2004-05. The most recent estimate available is for 2007-08, which was the year of the international food crisis when India had to import wheat and cereal price inflation was high. PDS usage was higher that year than in 2004-05, with 30% of households reporting any purchase of cereals from the PDS. But, since this too relates to all households including those with APL or even no ration card, it may underestimate numbers likely to access universal PDS.

However, there is one direct way to get an idea of the likely offtake with universal PDS, and this is to look at Tamil Nadu which is the only state that follows a universal PDS without

Economic Political Weekly

EPW

distinguishing between APL and BPL. As seen earlier, this is the state with the best functioning PDS that has the highest PDS participation and negligible leakages despite having the lowest prices. A further interesting aspect of PDS offtake in Tamil Nadu is clear evidence of self-selection by income deciles. From a near 100% offtake among the bottom deciles there is a gradual decline in the percentage of households purchasing from the PDS, with this falling to below 40% in the top rural decile and to almost nil in the top urban decile. Overall, the NSS estimates that 65% of all Tamil Nadu households (80% of rural and 50% of urban) purchased any cereal from the PDS in 2004-05. This increased to 76% in 2007-08 after Tamil Nadu reduced its PDS rice prices to Rs 2 per kg (two-thirds of what NAC is proposing) amidst increasing market prices.14 The Tamil Nadu PDS participation rates are an indication of what might happen with universalisation, and also show the limitation of using a targeted approach to food security. Only 30% of Tamil Nadu households would have Priority entitlement with the proposed NFSA as against 76% which actually accessed a universal PDS in 2007-08. Caps and exclusions would exclude many genuinely food insecure from the food security net.

However, the Tamil Nadu proportion of households choosing to access a universal PDS is likely to be well above the upper limit of what will happen nationally, if PDS was made universal in India as a whole. One reason for this is that no one is proposing to reduce all-India PDS prices to Tamil Nadu levels. Another is that Tamil Nadu is a food deficit state and there is a clear pattern of higher PDS demand in deficit states. The states that top in terms of households purchasing foodgrains from PDS in 2007-08 – Tamil Nadu (76%), Andhra Pradesh (70%), Himachal Pradesh (65%) and Kerala (54%) – are all cereal deficit. While better functioning of a PDS in these states owes a great deal to better governance, one could also argue that the food deficit nature of these states is why people demand more from the PDS and that it is this which compels governments to make PDS more effective. On the other hand, the surplus northern states that contribute most significantly to national procurement all show very low offtake: the percentage of households purchasing any cereals from PDS in 2007-08 was only 10% in Haryana, 7% in Punjab, 11% in Rajasthan and 15% in Uttar Pradesh.

Thus, it is quite safe to assume that the all-India proportion of households who would choose to buy cereals from a universal PDS will be less than in Tamil Nadu during 2007-08. Besides the fact that the NAC proposes a higher PDS price than in Tamil Nadu and that the country as a whole is much less food deficit, it is very unlikely that the outreach and reliability of Tamil Nadu’s PDS delivery system can be immediately replicated nationally even with best intentions on PDS reform. Therefore, at least till comparably good PDS systems are put in place across all the states, a more reasonable upper bound on the proportion of households buying PDS grain if a universal PDS is introduced would be around 70%, i e, between actual levels in Tamil Nadu of 76% in 2007-08 and 65% in 2004-05 when PDS rice price in the state (at Rs 3.50 per kg) was close to what NAC is proposing now for the country as a whole.

An alternative and independent estimate is possible at the all-India level from data used in Table 2 which shows that about 65% of households with BPL/AAY cards actually made any PDS purchase in 2004-05. These were all households with an entitlement to highly subsidised grain and yet 35% of them did not avail of this at all. Among quintile groups, only 55% of BPL/ AAY cardholders in the richest quintile made any PDS purchase. Among states, the percentage of households with BPL/AAY cards which accessed PDS for cereals in 2004-05 was less than a third in Bihar, Jharkhand, Haryana, Punjab and Uttar Pradesh, almost certainly because these had the worst functioning PDS. From this data it is possible to calculate what would happen if all households had been given BPL cards in 2004-05 and if state-wise, quintile-wise the behaviour of new recipients was the same as of those already having such cards. It turns out that only 56% of households would have made any PDS purchase if the PDS had been universal in 2004-05 with every household entitled to BPL prices and quantities.15 However, since market prices have risen after 2004-05 and BPL/AAY prices remain unchanged, a reasonable lower bound on the proportion of households likely to purchase grain if PDS were universal today should be placed at about 60%, although this is likely to be on the higher side.16

Table 5: Requirement of Total Foodgrains Table 5 gives estimates of for Universal PDS (million tonnes)

grain requirement for a uni-

Percentage of Average Quantity Purchased Households Per Capita Per Month by

versal PDS based on this range

Accessing PDS Households Accessing PDS

of 60-70% of households likely

7 kg 6 kg 5 kg

to purchase from PDS and with

100% 100.1 85.9 71.6

alternative assumptions regard

70% 70.1 60.1 50.0

60% 60.1 51.5 42.9 ing per capita grain purchase Calculations assume a population of 1,192 million. by those who do access PDS. The upper bound is 70 million tonnes if 70% of households do access PDS and all of them purchase 100% of their NAC stipulated quota of 7 kg per person per month. The actual requirement will be lower. The requirement will be 60 million tonnes if 60% of households access the full 7 kg. Moreover, not all households purchase their full quota. For example, according to the NSS, the average PDS purchase by BPL/AAY cardholders who did purchase any PDS grain in 2004-05 was only 21.5 kg per household (about 4.5 kg/person) per month. The grain requirement for a universal PDS based on the actual behaviour of BPL/AAY cardholders in 2004-05 is therefore lower than the lowest estimate of 43 million tonnes in Table 5 (60% of households purchasing an average of 5 kg/person/month). These alternative estimates of requirement with universal PDS may be compared with the actual TPDS offtake of 42.4 million tonnes in 2009-10 and the RC’s projected offtake of 41.9 million tonnes if PDS entitlements are restricted to only 40% of the population.

It is clear from these estimates that a universal PDS, at prices and quantities being suggested by the NAC for priority households, will certainly require more grain than the current actual offtake from TPDS. However, it is also clear that these requirements are not

Table 6: Allocation and Off-Take of Rice and Wheat in TPDS

impossible. Not only is the estimated requirement only marginally higher than current offtake at the lower bound, even the upper bound of 70 million tonnes is less than what was actually being allocated (although not lifted) as late as 2005-06 (Table 6). There could well be other reasons why commitment to universal PDS is not desirable, but to conclude “it will not be possible” is more a statement of ambition than about feasibility.

Real Issues Regarding Feasibility of a Universal PDS

However, two issues need clarification, since these do relate to feasibility. First, with TPDS allocations currently at 48 million tonnes, how can our calculations suggest that this may, albeit at the lower bound, suffice for a universal PDS while the RC argue that this will only allow 40% of population to receive NAS priority entitlements? Second, will it really be feasible to meet the requirement for universal PDS if it reaches the upper bound of 70 million tonnes?

As regards the first issue, this is mainly a matter of leakages. All estimates in Table 5 relate to likely household demand under a universal PDS and do not include leakage. On the other hand, it is known that out of the actual offtake in Table 6 about half currently leaks out from the system (Table 4), which, if plugged, could double the amount reaching intended households. The earlier discussion makes explicit that moves towards a universal PDS should be accompanied by PDS reforms that minimise leakage as has actually been done in Tamil Nadu and Chhattisgarh and is therefore demonstrably feasible.

In contrast, the RC criticises NAC estimates of requirement based on offtake/allocation ratio of 90% and argues that this ratio should be 100%. This means that the RC not only accepts as inevitable the leakage contained in the offtake figures of Table 6, but also assumes that the leakage ratio will increase if access to cheap grain is extended to more households.17 This is a common and plausible implicit assumption made by those critical of PDS because it does leak. However, as the discussion above has tried to show, this assumption is refuted by all the available evidence: the correlation between access to PDS and extent of leakage is strongly negative, both across states and over time.18 In other words, the more universal the PDSsystem is, the less likely it is to suffer from leakages.

Why this is so may need further analysis. But it is clear that leakages are much lower where a majority demands its PDS entitlements than where this attracts only a few. Higher participation almost certainly makes it more difficult to divert supplies from PDS shops and also puts pressure on state governments to carry out governance reforms. Nonetheless, since profits from diversion are high if PDS grain is priced very cheap, clear steps are necessary to contain leakage. But restricting numbers entitled to PDS is no solution, since this reduces access, not the leakage ratio. It is

more efficient to allow wide access, monitor actual tonnes,20 i e, over 40% of production of rice and wheat. This is important, since the RC’s main argument against the NAC proposal is that its requirement is not feasible because “it may be imprudent to assume an average procurement level of more than 30 per cent”.

EPW

Procurement is a function of production of foodgrains, the overall supply-demand balance and the MSP at which the government offers to buy. Despite the setback during 1997-2005 and continued weather-induced fluctuations, foodgrain production has recovered to per capita levels before the setback. Moreover, projections of foodgrain demand suggest that this will be adequately met by a production increase at least in the short run.21 The year 2008-09

saw the highest foodgrain pro-

Table 7: Production and Procurement of Foodgrains (million tonnes)duction at 235 million tonnes,

Production Procurement

although a severe drought in

2000-01 196.8 42.6

the following year (2009-10)

2001-02 212.9 41.5

caused production to decline to

2002-03 174.8 32.3 2003-04 213.2 40.3 218 million tonnes. Along with 2004-05 198.4 40.3 increased production, procure2005-06 208.6 38.0 ment has also increased, partic2006-07 217.3 36.2

ularly after 2006-07 (Table 7).

2007-08 230.8 51.6

Procurement of rice, wheat and

2008-09 234.5 60.4

millets was over 60 million

2009-10 218.0 54.9 Procurement against 2009-10 refers to marketing tonnes in 2008-09 and nearly years for 2009-10 output, i e, 2009-10 kharif

55 million tonnes even in the

marketing year and 2010-11 rabi marketing year. Source: Ministry of Food and Consumer Affairs. drought year of 2009-10.

However, the upper end of likely requirement for a universal PDS is much higher than recent procurement, which in turn is higher than the RC’s “prudent” level. Does this mean that extraordinary measures (such as movement restrictions or even raiding farmers’ houses as in the 1970s) will be needed to procure the extra grain if demand reaches this upper end? This is not so. Take last year when, despite drought reducing rice production by over 10 million tonnes,

30.5 million tonnes of paddy was procured. Total market arrival during procurement period in those markets where procurement was conducted was 51.2 million tonnes, i e, only 60% was procured. Similarly, 22.5 million tonnes of wheat were procured from markets where total arrival was 25.9 million tonnes during the procurement period.22 Thus, even in this drought year, total rice and wheat arrival during the procurement period in those markets where procurement was conducted was more than 77 million tonnes. This does not include sales by farmers in other markets or through private retail, and the total marketable surplus was far higher – 107 million tonnes according to the RC, quoting the Ministry of Agriculture. On this evidence, procuring 78 million tonnes in a normal year will of course require more effort but it is certainly not impossible, especially since procurement of coarse cereals, currently negligible, could add to the total pool of grains needed for a universal PDS.23

Nonetheless, it is reasonable to expect that such a large increase in procurement will only occur if MSPs offered are more attractive relative to market prices. And, it is on this count that the RC is pessimistic regarding higher procurement while noting that “one method of doing this is to increase the MSP which will ensure higher procurement but the limitation here is the potential inflationary impact on food prices”. However, one member of the NAC has on logical grounds questioned this RC view that higher procurement would “lead to a lower availability of foodgrain

Economic Political Weekly

EPW

for the open market, pushing up prices”, opening up a debate on the issue of grain management which although somewhat technical is very important (Dreze 2011). Before turning to this, it is necessary to reiterate (i) that plugging PDS leakage is essential and that this is possible is demonstrated amply in states where ambition exists for more universalisation; and (ii) that it is not impossible to physically procure grain required for universalisation even at the upper end of consequent likely household demand.

Procurement, Stocks and Grain Management

The weakest aspect of the NAC’s approach to the NFSA has been its obsession with access – i e, how many should be entitled to how much cheap grain and at what price – to the relative neglect of the other two pillars of food security: availability and absorption. To some extent, absorption has been incorporated in its suggestions on strengthening the Integrated Child Development Services (ICDS). But on availability it has not addressed the key issues of adequate incentive for farmers and adequate protection to consumers against unanticipated inflation. These issues affect everyone, and should have been the starting point of the NFSA, particularly in the present context of ongoing climate change and volatile world food markets. Not having done this, the NAC has tied itself into knots trying to cut the access coat to fit procurement. Without much thought within, the RC’s observations on procurement, MSP and inflation have thus come as a rude shock.

To rectify this, it is necessary to ask the following questions:

(i) Are current levels of food stocks and their spatial distribution sufficient to provide adequate protection against unforeseen price spikes due to output fluctuations, world price variations and other such contingencies? (ii) Do current levels of MSP give farmers adequate incentives to produce? and (iii) If answers to the above are yes, what is the likely average procurement at this level of MSP such that a steady flow of grain equal to this average can be maintained from the PDS to ensure NFSA entitlements, while actual procurement and stock levels vary around their averages with variations in output and/or demand? The size of this steady flow provides an idea of how much is available for the NFSA without this risking inflation.

On the first of these three questions, i e, on stocks, the answer is that while there are many considerations that can go into determining what is an optimal level of stocks, the present stock levels are higher than almost any definition. These are twice current buffer norms, are adequate to stabilise grain consumption at its trend through three consecutive years of short production of the magnitude of 2009-10, and also enough to provide for the contingency that PDS demand may in situations of price spikes rise sharply from normal levels to the full NFSA legal obligation.24 There is therefore no need to plan for higher stocks, although running these down may also not be prudent given that the future may be more uncertain than the past. Normal offtake under NFSA should, therefore, equal whatever is normal procurement. But, NFSA design should also prevent the dysfunctional behaviour seen in the recent past, e g, building stocks up during the drought of 2009-10 and therefore adding to inflation when the very purpose of holding buffer stocks is to offload these in shortage situations (Himanshu 2010a; Basu 2011).

In particular, this means that the NFSA should not be so designed to completely shut out the non-poor who may not access PDS normally but could turn to this in periods of high inflation. To shut the non-poor out even in such situations is bad not only for the non-poor but also for the poor since entry of the non-poor into PDS in such situations helps reduce public stocks and, therefore, inflation. Once again, a dynamic response to self-selection is not only better for economic outcomes but also more suited to a rights approach than BPL targeting. However, inflation like cancer begins in limited locations and then spreads unless nipped in the bud. Dynamic response in this context means to increase allocations for the non-poor whenever and wherever food inflation crosses some limit. But for this to be possible there must be adequate stocks in every location, which is not always the case at present.

It would therefore be appropriate that even before defining entitlements and access, NFSA commits that there will at all time and in all locations be enough stocks of foodgrains to provide everyone with their full requirement. Such a universal statement on availability need not state at which price this will be available but it does require enough storage space in all locations (say, blocks) and always enough stocks (say, two month’s full consumption of 10 kg per individual), with continuous replenishment. This is easily doable since capital investment in decentralised storage space required for this is long overdue and affordable, and the stocks that need to be repositioned and replenished more evenly across space are only about half the stocks already with the government.

On the second question, regarding MSP of rice and wheat, the answer again is that while there may be differences on the matter, it is probably best to continue at present levels with the usual adjustment by the Commission on Agricultural Costs and Prices (CACP) to reflect changes in the cost of production. MSPs were increased very substantially during crop years 2007-08 and 2008-09, almost bringing these in line with the recommendation of the Farmers’ Commission to fix these 50% higher than the C2 cost of production. But subsequent increases have not kept up with costs which currently are rising faster than overall inflation, and will rise even more with a revision in NREGA wage rates and if fertiliser and fuel subsidies are cut. Although it is valid to argue that the present high stock levels suggest that MSPs may be too high, farmers strongly disagree and the failure to build in actual cost increases may dampen incentives.

This brings us to the third and critical question regarding normal levels of procurement and, therefore, of offtake. It may appear that maintaining present MSPs will also maintain the present size of procurement at current levels of output. But this is not so. Procurement has exceeded offtake by an annual average of over 12 million tonnes since 2007, leading to the currently high level of stocks. This means that market availability of rice and wheat has been significantly less than production, and market prices of cereals have been higher than they would have been without stock increase. Consequently there will be a correction as soon as offtake is increased to equal procurement: market prices will fall and selling to government will become more attractive to farmers, increasing procurement even at current MSPs.

The extent to which procurement will increase can be assessed on the basis of available estimates of relevant price elasticities. On this the basis, we estimate that since availability will increase by at least 5% when government stops increasing stocks, market prices are likely to decline by at least 10% due to increased supply.25 Also, that this extent of fall in market prices is likely to increase procurement from around 55 million tonnes presently to 60-65 million tonnes.26 The NFSA can therefore be planned around an offtake of 60-65 million tonnes while stocks are on average maintained at their present high levels, but allowed to fluctuate with variations of production from trend. This involves no additional risk of inflation since MSPs need not increase at more than the rate of increase of the cost of production, and market prices will in fact be lower as a percentage of MSP than at present.

This conclusion differs significantly from the RC’s observations that “In view of the cycles in agricultural procurement, it may be imprudent to assume an average procurement level of more than 30 per cent”, leading them to set a ceiling on NFSA offtake that is less than current procurement. It is, however, important to note that policy decisions contribute to “cycles in agricultural procurement” and that what RC is suggesting may in fact perpetuate this.

The problem with the RC view is that unless MSPs are reduced in nominal terms immediately, procurement will be higher than the RC ceiling on offtake. Stocks will continue to increase unnecessarily, along with unnecessarily high market prices for cereals. Although it may be possible to bring procurement down over time, by holding nominal MSPs constant as market prices increase with inflation, a balance between procurement and ceiling offtake will only be achieved with higher stock levels and lower levels of real MSP than at present. If the experience of 1997-2005 is any guide, the final outcome could be that pressures build up to allow export in order to reduce stocks while production incentives are reduced unduly.

During 1997-2001, a combination of high MSP and high issue prices for APL caused huge stock accumulation despite stagnant per capita production, and this in turn led to high food inflation. Subsequently, in view of the large stocks, MSP increases were kept below general inflation during 2001-06, thus reducing procurement while exports were also allowed to deplete stocks. However, per capita cereals production in the triennium ending 2005 fell to its lowest in 20 years and stock levels also fell below buffer norms and remained below the norm throughout 2006 and 2007. It is this tight domestic supply situation, amidst high world food inflation, that was corrected by the large MSP increases since 2007.

We are now roughly at the 2001 situation in that previous cycle. The choice is between stabilising offtake at around the 60-65 million tonnes procurement levels consistent with the present MSP levels or attempting once again to reduce procurement. The RC clearly favours the latter option, where procurement is brought down to around present offtake levels through reduction of real MSPs, since this would obviously reduce the fiscal outgo. However, it ends up arguing quite incorrectly that NAC proposals on NFSA are not feasible without MSP hikes that risk more food inflation. Not only is the NAC proposal feasible without risking further inflation, so is universalisation under most likely scenarios.

An Alternative Proposal for PDS in NFSA, Short of Universal PDS

However, despite being feasible in most scenarios of Table 5, a universal PDS is currently not on anyone’s agenda and hence unlikely to be accepted, particularly after the RC report and

march 19, 2011 vol xlvi no 12

EPW

because the upper end of likely demand with a universal PDS is, at 70 million tonnes, higher than the 60-65 million tonnes consistent with present MSP levels. We, therefore, suggest an alternative which takes elements of the NAC proposal and merges them with elements of a proposal that the PC had made earlier (Planning Commission 2010). This is done to be consistent with a rights based approach, avoid the main problems of targeting, preserve the major benefits of universalisation and also to reduce virtually to nil the possibility of demand rising so high that the government is forced to renege on its NFSA obligations. The proposal is as follows:

Unlike the NAC proposal, but like the earlier PC proposal, this involves a universal entitlement which is MSP-linked. Like the NAC proposal and unlike the earlier PC proposal, there are three groups selected on exclusion and inclusion criteria. However, based on results from the recent BPL Census pilot27 we recommend that selection of these groups be delinked from poverty estimates. While exclusion criteria can be similar to those proposed by NAC and exclude roughly 25% of population, our proposed inclusion criteria are more stringent than for NAC’s priority group. This will include the specific categories that NAC has already identified but in addition include only those who suffer from at least three identifiable deprivations. As a result, the priority group is likely to cover only about 20% of the population that is either disabled or destitute or identifiably deprived on multiple dimensions. This is unlike both the NAC and PC proposals, which stretch the full discount to Tendulkar poverty numbers. However, again unlike both the NAC and the PC, about 75% of the population (i e, all except the excluded group) will have entitlement to some very low priced grain and therefore a strong incentive for PDS participation.

Compared to the present AAY/BPL/APL division, this will double the number of beneficiaries with present AAY entitlements from 10 to 20%, and entitle another 55% of population to not only

3.5 kg/month at present AAY prices but also a full quota of 7 kg/ month which, if fully drawn, will be at an average price less than proposed by NAC for a lower quota of 4 kg/month for the general category.28 However, there are only two prices in this proposal, not three as today, and the higher MSP-linked price will be much higher than the current APL price.

What are the advantages over the NAC proposal? There are three:

(a) First, as far as targeting is concerned, our proposal limits this only to identification of the very disadvantaged and those undoubtedly non-poor, based on verifiable inclusion and exclusion criteria. Results of the recent BPL pilot show that while this can be achieved with reasonable probability of containing errors of

Economic Political Weekly

EPW

inclusion and exclusion within acceptable limits, use of score-based rankings to reach caps based on Tendulkar poverty numbers will inevitably exclude a very large number of families that are very similar to other families which are included. Such caps and scoring are therefore likely to perpetuate the rampant rent-seeking in BPL selection that many now consider to be the biggest stumbling block to effective delivery of public services. Moreover, verifiable exclusion and inclusion criteria are necessary in a rightsbased approach since this must provide a transparent legal definition of who is entitled to what, avoiding the almost impossible problem of defining who is poor and who is not.

What is the grain requirement for this proposal? If the full entitlement at AAY prices is drawn by everyone eligible, this will require 47.6 million tonnes but, as discussed above, actual purchase is likely to be much less. However, even assuming that leakages cannot be immediately reduced and the offtake/allocation ratio remains at 95% currently observed for the AAY, the maximum demand will be 45.2 million tonnes. In addition, entitlements at MSP-linked prices work out to 52.5 million tonnes if everyone draws full quota, but the actual demand is unlikely to exceed 9.5 million tonnes even with leakage.29 The likely PDS requirement in our proposal will thus not exceed 54.7 million tonnes, which is less than the revised TPDS allocation of 57.7 million tonnes for 2010-1130 (up from 47 million tonnes through extra ad hoc allocations). Even adding requirements that NAC has projected for other welfare schemes, the total grain requirement is within the 60-65 million tonnes of likely procurement. Moreover, as pointed out earlier, stocks will be sufficient to meet statutory obligations in the very unlikely event that demand shoots up to full entitlements.

And, finally, what will be the subsidy outgo? This works out to Rs 79,400 crore, calculated as excess of economic cost over issue prices multiplied by the likely maximum offtake.31 On the same subsidy calculation basis, the NAC places the outgo at Rs 79,930 crore for its proposal in its final phase and RC projects Rs 68,540 crore for its watered down version in its first phase. As against these, actual outgo on this basis is placed at Rs 60,600 crore in 2010-11(RE), but could end up at Rs 68,950 crore.32 All these figures exclude cost of maintaining buffer stocks and also underestimate subsidy whenever procurement is larger than offtake and therefore involves a procurement cost without corresponding sales revenue. Since our proposal assumes a balance between procurement and offtake at present stock levels, while both the present and the RC projection involve rising stocks with procurement higher than offtake, the difference in total subsidy between our proposal and either of these will be less than corresponding differences in figures above. Moreover, it should be noted that in their 2010-11 budgets, eight state governments had together made a provision of Rs 11,280 crore for food subsidies over and above the centre’s outgo.33 Since these state subsidies are mainly used to bring down the issue price of grain obtained at higher prices from the centre, much of this is likely to get subsumed in the proposal above. If so, our proposal may not involve much additional fiscal cost (centre plus states) of PDS, even before factoring in what we consider to be its main cost reducing feature: that leakages are likely to reduce since it is more universal.

Conclusion: Feasibility, Reform and Ambition

The proposal above should be seen as a serious alternative to the NAC view on NFSA not only because of the three advantages listed specifically in the previous section, but also because it addresses all the concerns raised by the RC. In fact, the proposal should be seen as a way in which feasibility concerns raised by RC may be addressed while going beyond NAC on the matter of universalisation. However, it should also be noted that while RC has judged infeasible the NAC proposal on grounds of grain availability, this involves a PDS requirement of 57.4 million tonnes in the first phase which is less than the 57.7 million tonnes to which the official allocation for 2010-11 has already been raised by the Ministry of Food and Consumer Affairs. Obviously, notions of feasibility apply unevenly to normal decisions within the government and when brought to bear on major policy reform and legislation.

For this reason, it is important on matters of policy and legislation to discuss upfront the points of unevenness that go beyond details to the idea of reform itself. For example, it is interesting but irrelevant that our proposal, on proper costing, may not impose an additional fiscal burden. Our focus has been universalisation, benefits of which have been argued but not incorporated in the costing of the alternative proposal. There are also other suggestions on reform of PDS that could further reduce costs but have not been discussed here: e g, decentralised procurement that can reduce the gap between economic cost and MSP, and, as suggested in the Planning Commission proposal, use of smart cards to avoid corruption due to dual pricing by delivering directly to PDS shops the difference between the MSP-linked and the lower AAY-based prices. These can easily be incorporated into this proposal, although like universalisation exact benefits are difficult to quantify till these are put in place.

However, the main messages of this paper are two: First, that targeting has not worked and that universalisation is likely to reduce PDS leakage, not increase it. Second, that food security cannot be attained without addressing issues of physical availability, distribution and stock management – it is not simply a matter of access that can be dealt with through transfers, either directly in cash or through coupons or by differential pricing.

Both these messages have detractors which is probably why the NFSA has not rolled out yet, although over the last three years we have seen considerable ambition in state after state to make PDS more universal and to address its governance details. The proposal above should be seen in this context. It addresses concerns of the detractors but once rolled out, and if successful in reducing leakage while increasing access, could be converted easily to full universal PDS. All that is needed to achieve this is to set the percentage of MSP in the MSP-linked price so that this is the same as an acceptable price for the priority group. Since our analysis supports very strongly the case for a fully universal PDS with below MSP pricing, this is a correct ambition but clearly not at present at AAY prices which are far too low. Of course, the antyodaya (i e, the old, disabled, orphaned, widowed or abandoned) would then lose and need to be compensated. This is where direct cash transfers are relevant through better pensions and disability payments, not as replacements for PDS.

Notes

1 NAC has accepted as priority the below poverty line population of 42% rural and 26% urban for 2004-05, following acceptance of the Tendulkar Committee by the Planning Commission. As is the standard practice, 10% of the existing poverty estimates are added to account for the transient poor making it 46% (42+4) in rural and 28% (26+2) in urban. This has already been accepted by the food ministry for new BPL entitlements.

2 The NAC suggests a phased implementation of the Act with the responsibility of the present government only limited to 85% of rural population and 40% of urban population.

3 Accepted in the 14 July 2010 meeting of the NAC. See http://www.nac.nic.in/press_releases/14_july_ 2010.pdf

9 Although the quantity of foodgrains received is different based on family size and a small percentage of the poorest households get higher allocation, the price paid by all is the same.

10 All PDS shops in the state are run by cooperatives/government. Ninety-nine per cent of taluks have their own godowns. The quality of foodgrain delivered is good and along with cereals, it also gives other food items such as pulses and edible oil. Tamil Nadu uses technology, e g, an online monitoring system, GPS and so on to track foodgrain movement. Details on working of the PDS in Tamil Nadu are available at http://www.tncsc. tn.gov.in. The two main contributions of universalisation are that unlike elsewhere there is no tyranny of the BPL list and the single PDS price reduces arbitrage possibilities between different clients of a PDS shop for the same commodity.

march 19, 2011 vol xlvi no 12

EPW

11 Chhattisgarh began PDS reforms in 2005 and in 2007 redefined eligibility criteria to include 75% of population. On the reforms undertaken, see Mishra (2010) and Patnaik (2010).

12 This is based on the individual, rather than household, entitlements proposed by the NAC. The number of households is not known for the current population, and depends on the definition of a household (nuclear or common kitchen). However, this corresponds to 35 kg of foodgrain for a family of five.

13 There is strong empirical evidence of this from the 1993-94 NSS consumption expenditure survey, which had asked the question of whether the household is engaged in cultivation of foodgrains and if yes, then which foodgrains. Comparison of PDS purchase by cereal growers versus those who are not shows that those who grow cereals are three times less likely to take foodgrains from PDS. This increases to five times less likely for those who belong to the medium and large farmer category.

14 The PDS rice price in Tamil Nadu was further reduced to only Re 1 per kg in 2008.

15 This is less than the 65% of actual BPL cardholders who report some PDS purchases because the percentage of households who report not having BPL cards was much higher in states such as Bihar, Haryana, Punjab and UP where actual BPL cardholders report particularly low PDS purchase. Unfortunately, a similar calculation is not possible for 2007-08 since the 64th round did not collect data on type of PDS card.

16 For example, take Chhattisgarh where, despite expansion of eligibility to include 75% of population, lowering of prices to Rs 2/kg and PDS reforms, only 40% of households (i e, 53% of eligible households) purchased any foodgrains from PDS in 2007-08.

17 Both NAC and RC start with the full requirement to provide 100% of priority households 7 kg per month, but while NAC assumes only 90% of this will be lifted (based on actual offtake/allocation ratios), the RC argues for full 100%. In fact, demand by BPL/AAY households was only 40% of entitlement in 2004-05 and this is unlikely to cross 60%. Therefore, the leakage ratio assumed implicitly by RC is 40-60%, equal or more than the current actual. It should also be noted that all percentages here are with respect to full requirement and not allocation. This is because the latter is an administrative decision. Despite population growth, allocations for BPL/AAY were kept constant and for APL sharply reduced after 2004-05. The recent rise in offtake/allocation ratios in Table 6 is more due to this than to the increase in demand.

18 The correlation across major states between the percentage of households accessing PDS and percentage of leakage from PDS was -0.9 in 2004-05 and -0.7 in 2007-08. Over time, access declined and leakage doubled between 1993-94 (pre-targeting) and 2004-05 (after targeting); and then access increased and leakages decreased between 2004-05 and 2007-08 when many states introduced their own subsidies to expand BPL in the face of higher inflation.

19 Indeed a variant of this was followed in universal PDS before 1997 and is still followed in Tamil Nadu. And new technology (CCTVs/swipe cards) now makes real-time monitoring of actual household demand much easier.

20 Other welfare schemes, including the ICDS and MDM, require an additional 3 to 4 million tonnes at present levels but may reach 8 million tonnes if NAC recommendations on other components of NFSA are accepted.

21 The total demand of foodgrains in 2011-12 (including direct and indirect demand taking into account feed, seed and wastage is projected at

212.6 million tonnes (Mittal 2008), 235 million tonnes (Chand 2007), 223.7 million tonnes (Kumar 1998). IFPRI projects the demand of foodgrains at

237.3 million tonnes in 2020. As against these, production was already 235 million tonnes in 2008-09 and may be even higher in 2010-11.

22 Department of Food and Public Distribution; presentation made to NAC on 24 September 2010.

30 Including special ad hoc allocations up to 6.1.11, these are 10.2 mt AAY, 22.5 mt BPL and 25 mt APL.

31 Calculated using economic cost of Rs 20.43/kg for rice and Rs 15.44/kg assuming a 60:40 rice-wheat ratio, and offtakes of 45.2 and 9.5 million tonnes at AAY prices and 75% MSP respectively.

32 With economic costs as above, and actual issue prices, applied to likely offtake in 2010-11. Likely offtake is obtained by applying actual offtake/ allocation ratios for April-November 2010 to the full 2010-11 allocation as in footnote 31 above. Excluding special ad hoc allocations, likely subsidy is Rs 57,000 crore, about the same as 2010-11(BE).

33 Food subsidies in State Budgets for 2010-11 were Rs 4,000 crore in Tamil Nadu, Rs 3,000 crore in Andhra Pradesh, Rs 1,250 crore in Karnataka, Rs 1,000 crore in Chattisgarh, Rs 910 crore in Orissa, Rs 700 crore in West Bengal, Rs 290 crore in Madhya Pradesh and Rs 130 crore in Kerala.

References

Basu, Kaushik (2011): “India’s Foodgrain Policy: An Economic Theory Perspective”, Economic Political Weekly, 29 January.

Chand, Ramesh (2007): “Demand for Foodgrains”, Economic Political Weekly, 29 December.

Dreze, Jean (2011): “A Notional Advisory Council?”, Hindu, 10 January.

Economic Advisory Council (2011): “Report of the Expert Committee on National Food Security Bill”, available at http://eac.gov.in/reports/rep_NFSB.pdf

Himanshu (2010a): “Of Food Inflation, Farmer Interest and Policy Failure”, MINT, 1 April.

– (2010b): “Food Entitlements Should be Universal”, MINT, 28 April.

Kumar, Praduman (1998): “Food Demand and Supply Projections for India”, Agricultural Economics Policy Series 98-01, Indian Agricultural Research Institute, New Delhi.

Kundu, Amitabh and Ravi Srivastava (2007): “Meeting the Food Security Challenge in India: Mediumterm Goals and Strategies”, Indian Journal of Human Development, Vol 1, No 2.

Ministry of Food and Consumer Affairs (2002): “Report of the High Level Committee on Long Term Grain Policy”, available at www.fcamin.nic.in

Ministry of Rural Development (2011): “Identification of Poor: Preliminary Results from the Pilot Survey”, 7 February, processed.

Mishra, Udit (2010): “How the PDS Is Changing in Chhattisgarh”, Forbes, 10 December.

Mittal, Surabhi (2008): “Demand-Supply Trends and Projections of Food in India”, ICRIER Working paper number 209, ICREIR, New Delhi.

National Advisory Council (2011): “Note on the Draft National Food Security Bill”, available at http:// www.nac.nic.in/foodsecurity/nfsb.pdf

Parikh, Kirit (1994): “Who Gets How Much from PDS: How Effectively Does It Reach the Poor?”, Sarvekshana, Vol XVII, January-March.

Patnaik, Biraj (2010): “Chhattisgarh’s Food for Thought”, MINT, 21 April.

Planning Commission (2008): “Nutrition and Social Safety Net” in Eleventh Five-Year Plan 2007-12: Volume II, Oxford University Press.

– (2010): “Towards a Workable Food Security Bill”, July, Processed.

Swaminathan, Madhura (2000): “Consumer Food Subsidies: What Needs to be Done in India”, ARC Working paper No 2, Asia Research Centre, London School of Economics and Political Science.

World Bank (1996): “India’s Public Distribution System: A National and International Perspective”, Poverty and Social Policy Department.

Economic Political Weekly

EPW