SPECIAL ARTICLE

Casting the Net

India’s Public Distribution System after the Food Security Act

Jean Drèze, Prankur Gupta, Reetika Khera, Isabel Pimenta

I
A broad-brush assessment of the public distribution ndia’s National Food Security Act (NFSA) (2013) is one of
system is presented in six of India’s poorest states— the largest social security initiatives in global history. It
covers more than 800 million people through the public
Bihar, Chhattisgarh, Jharkhand, Madhya Pradesh, Odisha,
distribution system (PDS) alone, aside from mandating nutri-
and West Bengal—soon after the National Food tious midday meals for children and maternity benefits for
Security Act, 2013 came into force. Important gains have pregnant women. Oddly, however, the rollout of the act has
been made, including broader coverage, lower targeting received little attention from the research community and
mainstream media.
errors, accelerated PDS reforms, and a greater political
This is a serious blind spot, considering that the provisions
commitment to food security. In four of the six reference of the act are of great importance for the poor. The PDS, in par-
states, the PDS seems to be doing reasonably well, but ticular, is a significant source of economic security for many.
Bihar and Jharkhand still have a long way to go. Even in The recent starvation deaths in Jharkhand highlight the dan-
gers of disrupting this critical lifeline of the rural poor (Drèze
the leading states, much remains to be done to achieve
2017b; Dutta 2018).
the purpose of the NFSA: ending food insecurity. Monitoring the implementation of the NFSA is also impor-
tant from the perspective of future planning. The successful
implementation of the act is likely to require regular course
correction. For instance, if the selection of eligible households
is to improve over time, then lessons must be learned from ear-
lier attempts in order to avoid inclusion and exclusion errors.
This paper presents a broad-brush assessment of early expe-
riences with the NFSA in six of India’s poorest states, with spe-
cific reference to the PDS.1 Much of it draws on a field survey
conducted by student volunteers in June 2016. This was not a
large-scale survey, but we shall combine it with insights from a
series of other surveys conducted in the same states.2
The authors are grateful to Sachin Jain, Rajkishor Mishra, Nandini
Nayak, Sulakshana Nandi, Gangabhai Paikra, and Kumar Rana for their
The National Food Security Act
help in planning the field survey in the different states. They thank
Swati Chaurasia, Anshika Jain, Souparna Maji, and Anand Prakash The NFSA, enacted in mid-2013, requires the government to
for their assistance with data verification. They are also grateful to the extend foodgrain subsidies under the PDS to 75% of the popu-
food department officials in the reference states for providing us helpful lation in rural areas and 50% in urban areas. The recipients fall
clarifications, especially Ashok Barnwal, Subhra Chakrabarti, Vinay
into two categories, jointly known as “eligible households”—pri-
Kumar Choubey, Rajeev Jaiswal, Prakash Kumar, Madhusudan Padhi,
Pratibha Sinha, and A K Somasekhar. They thank Mihika Chatterjee, ority households and Antyodaya Anna Yojana (AAY) house-
Aaditya Dar, Aarushi Kalra, Chinmaya Kumar, Siddhartha Mitra, holds. Priority households are entitled to 5 kg of foodgrains per
Raghav Puri, Anmol Somanchi, and the anonymous EPW referee for person per month at the nominal price of `3, `2, and `1 per kg
their helpful comments. Finally, they thank all the volunteers who for rice, wheat, and millets, respectively. Antyodaya households,
participated in the field survey for their spirited cooperation.
the poorest of the poor, are entitled to 35 kg of foodgrains per
Jean Drèze ([email protected]) is a visiting professor at the month at the same prices, irrespective of the size of the family.
Department of Economics, Ranchi University. Prankur Gupta The proportion of eligible households varies by state, as
([email protected]) is a research scholar at the Charles H
Dyson school of Applied Economics in Management, Cornell University.
well as between rural and urban areas; thus, the minimum
Reetika Khera ([email protected]) is an associate professor of coverage of the PDS is higher in the poorer states—for example,
economics at the Indian Institute of Technology Delhi. Isabel Pimenta it is 86% in rural Jharkhand.3 The selection of eligible
([email protected]) is an economist, currently working at households is the responsibility of the state governments. The
the European Commission in the Directorate General for International act also provides for a range of PDS reforms, aimed at
Cooperation and Development, Brussels.
ensuring transparency and avoiding leakages.
36 FEBRUARY 9, 2019 vol lIV no 6 EPW Economic & Political Weekly
 SPECIAL ARTICLE

Chhattisgarh enacted its own food security law, the Chhattis- excluded to ensure that a full door-to-door survey was feasible
garh Food Security Act (CFSA), in December 2012. The CFSA is within one day, and tiny villages were avoided to ensure that
similar to the NFSA, with some notable differences. Unlike the each survey day was well used. The exclusion of large villages
NFSA, the CFSA explicitly defines eligibility criteria; for instance, means that the sample was biased, if at all, towards relatively
all households headed by single women are entitled to an poor and less well-connected areas and households.
Antyodaya card, unless they meet pre-specified exclusion criteria.
Also, eligible households receive more entitlements under the Integrity of the NFSA Lists
CFSA than under the NFSA. Initially, under the CFSA, both prior- The NFSA Survey 2016 was mainly an exhaustive door-to-door
ity and Antyodaya households had similar entitlements (with survey in 36 sample villages. The survey teams, however, were
minor differences in issue prices), defined in per household— also in possession of the official list of ration cards for each
rather than per capita—terms: 35 kg of foodgrains and 2 kg of village. The households on these lists were divided into three
pulses or black gram per month.4 In April 2015, however, Chhat- groups—sample households, households that could be traced
tisgarh amended Schedule 1 of the CFSA to offer per capita enti- (they were known to live in the village) but were not inter-
tlements for priority households. Unlike the NFSA, the CFSA does viewed (for example, due to temporary outmigration), and
not pre-specify a minimum coverage for the PDS—the coverage “untraceable households” (those that no one in the village had
derives from the eligibility criteria. Because the eligibility criteria heard of). The survey teams were trained to scrutinise the
are fairly broad, the coverage is close to universal in rural areas. official list after the door-to-door survey and make a shortlist
This paper focuses mainly on the benefits rather than the of the untraceable households in each village.6
costs of the PDS. The costs are not easy to estimate. The food Unfortunately, these shortlists were lost after the survey. How-
subsidy accounted for a little below 1% of India’s GDP (gross ever, from surviving summaries and the debriefing workshop pro-
domestic product) in 2016–17, but this is best considered an up- ceedings, we can say something about the extent of the problem.
per bound since it includes not only the consumer subsidy, but In West Bengal, the entire exercise proved unmanageably
also the producer subsidy and the cost of buffer-stock opera- tedious because the state has an odd system of issuing individ-
tions aimed at price stabilisation. ual (instead of household) ration cards. In Chhattisgarh, MP,
and Odisha, the proportion of untraceable households was
The Survey negligible. In Bihar and Jharkhand, a significant proportion of
The survey mentioned earlier (hereafter referred to as NFSA the households on the official list were untraceable—perhaps
Survey 2016) was conducted between 1 and 10 June 2016 by 5%–10%. An untraceable household, however, does not neces-
student volunteers in Bihar, Chhattisgarh, Jharkhand, Madhya sarily mean that it is a bogus or ghost household. It was clear,
Pradesh (MP), Odisha, and West Bengal (hereafter the “refer- for instance, that the lists in Bihar and Jharkhand had some
ence states”). The reference states are all situated in eastern location errors—sometimes a ration card was tagged to the
India and are contiguous with each other. The survey covered wrong village or ration shop, perhaps due to data-entry glitches.
three randomly selected villages in each of the 12 sample In Jharkhand, and to a lesser extent in Bihar, the lists were still
blocks (two blocks per state, in different districts). In each of in the process of being finalised at the time of the survey.7
the 36 sample villages, the teams carried out an exhaustive This is not to deny that bogus cards may exist, but their pro-
door-to-door survey, verifying ration cards and collecting portion appears to be relatively small. For Jharkhand, this
basic data on PDS purchases for all households; a small propor- finding is consistent with a recent, large-scale study (Muralid-
tion of households, about 2.5%, could not be interviewed haran et al 2018a/2018b) which estimates that the proportion
because no adult was at home on the day of the survey. Close of bogus ration cards in Jharkhand is “at most 2.6%.” Judging
to 3,800 households were surveyed. The survey teams also from the state of the ration card lists in the sample villages, the
made unannounced visits to PDS shops in the sample villages proportion is likely to be even smaller in other reference states,
and analysed their official lists of ration cards. with the possible exception of Bihar.
Because the survey was designed to be a low-budget exercise, This is an achievement of sorts, considering the notorious
we selected one district in each state based on accessibility from unreliability of the earlier ration card lists, known as “BPL lists.”
Ranchi (the survey headquarters, so to speak)—Gaya (Bihar), It reflects the effectiveness of the massive clean-up of ration
Balrampur (Chhattisgarh), Anuppur (MP), Sundargarh (Odisha), card lists that occurred during the rollout of the NFSA, when all
and Purulia (West Bengal). These districts are typically near the the lists were redone based on simpler criteria and more relia-
state border (with Jharkhand) and relatively far from the state ble data. It is important to note that this cleanup preceded the
capital. To balance this out, the second district was selected closer imposition of mandatory Aadhaar linking to avail PDS benefits,
to the state capital, bearing in mind the need for reasonably discussed later in this paper. Perhaps, Aadhaar integration
good connectivity with the first district as well as with Ranchi.5 helped take it further, for example, by eliminating the odd dupli-
In Jharkhand, we selected Gumla as the district close to the cate, but the bulk of the hard work had been done earlier.
state capital and Dumka as the more remote district.
Within the districts, block selection was random. Within Eligibility Criteria
blocks, sample villages were selected at random among those As mentioned earlier, the selection of eligible households (pri-
with a population of 75 to 125 households. Large villages were ority and Antyodaya) for the NFSA is the responsibility of state
Economic & Political Weekly EPW FEBRUARY 9, 2019 vol lIV no 6 37
SPECIAL ARTICLE

governments. When the act was drafted, the Socio-Economic approach, or some variant of it, and possibly some inclusion
and Caste Census (SECC) of 2011 was widely considered a pos- criteria as well (Table 1).
sible basis for the selection exercise. The release of SECC data
was delayed for several years, also delaying the rollout of the Identification of Eligible Households
NFSA in many states. Eventually, some states—including three Even when based on a relatively simple method such as
of the reference states—used the SECC data to identify eligible the exclusion approach, identifying eligible households is a
households, but others did not. mammoth exercise (for a taste of its complexity, see Satpathy
In contrast with the convoluted “scoring system” of the ear- 2017). The reference states were just concluding this mam-
lier below poverty line (BPL) surveys, many states defined moth task at the time of the survey.
NFSA eligibility in terms of simple inclusion and exclusion cri- Chhattisgarh, where PDS coverage was near-universal before
teria.8 The first step was to specify inclusion and exclusion cri- the enactment of the state law, was a little ahead of the other
teria, and also whether the latter override the former or vice reference states, but even here, the task of identifying eligible
versa. Table 1 presents a summary of the main eligibility criteria households was far from easy. Chhattisgarh notified inclusion
used in the reference states. and exclusion criteria and adopted a “self-declaration app-
The simplest way of identifying eligible households is what roach.” Application cum self-declaration forms were to be sub-
might be called the “exclusion approach.” In this approach, mitted to the gram panchayats. Households were expected to
the government notifies a list of simple and transparent ex- honestly self-select the correct ration card category.
clusion criteria, and every rural household is eligible by de- As mentioned earlier, the CFSA initially endorsed the old
fault unless it meets some of these criteria. To illustrate, in system of household entitlements rather than per capita enti-
rural Jharkhand, the main exclusion criteria are as follows— tlements. This created a problem at the identification stage,
regular government employment, ownership of a four-wheeler, because some households began to split up—or pretended to
ownership of more than five acres of irrigated land (or 10 split up—in the hope of getting several ration cards instead of
acres of any kind of land), and possession of a pacca (con- just one. As a result, the number of single-member households
crete) house with more than three rooms.9 This approach is in Chhattisgarh shot up. In a knee-jerk reaction, the state gov-
relatively straightforward and limits the risk of exclusion er- ernment called for the cancellation of all single-member cards,
rors—an important concern in the context of an act that seeks at the risk of causing temporary hardship to those who actual-
to ensure that vulnerable households get food rations as a ly lived on their own (2.25 lakh single-member cards were can-
matter of right. However, as we shall see, this requires reliable celled). Eventually, the state government amended the CFSA—
household listings. Most reference states used the exclusion a new category was created for single-member households,
Table 1: Eligibility Criteria in Rural Areas with an entitlement to 10 kg of free rice per month,
State Eligibility Main Exclusion/Inclusion Criteria and the foodgrain entitlement of priority house-
Bihar All households that Exclusions: Government job; motorised vehicle (three- holds was fixed at seven kg per person per month.
do not meet any of or four-wheeler); house with three or more pacca rooms;
the exclusion criteria. 2.5 acres of irrigated land; any household member earning Apart from Chhattisgarh, Bihar was a pioneer
more than `10,000 per month. in the sense that it was among the earliest to un-
Chhattisgarh All households that Exclusions: Similar to Bihar and Jharkhand. dertake the identification exercise (Drèze et al
meet at least some of Inclusions:* Beneficiaries of the CM Food Assistance Scheme 2015). Bihar applied simple exclusion criteria to
the inclusion criteria, prior to CFSA 2012; landless agricultural labourers; marginal
and none of the or small farmers (up to five acres of land); registered SECC data to prepare the initial list of priority house-
exclusion criteria. casual workers. holds. The village-wise draft lists were displayed in
Jharkhand All households that Exclusions: Permanent government job; motorised four- public, and individual households were given copies
do not meet any of wheeler; more than five acres of irrigated land, or 10 acres
the exclusion criteria. of any land; house with three or more pacca rooms.
of their SECC data. People who had been missed by
Madhya Pradesh All households that Inclusions: Many categories including former BPL the SECC, or who were dissatisfied with their status,
meet at least some of households; all SC/ST households; registered landless could apply for corrections. The reliability of this
the inclusion criteria. agricultural labourers; registered construction workers; correction process, however, is hard to determine.
disabled persons; and pensioners.
Thus, the main problem with this approach is that it
Odisha All households that Exclusions: Permanent government job; motorised vehicle
do not meet any of (three- or four-wheeler); any household member earning relies heavily on the integrity of SECC data.
the exclusion criteria.# a salary or pension of more than `10,000 per month; Jharkhand and West Bengal use models simi-
household paying income tax. lar to that of Bihar. In each case, the difficulty
West Bengal All households that Exclusion: Similar to Bihar and Jharkhand.
meet at least some of Inclusion: Many categories including all SC/ST households;
seems to be in dealing with gaps and inaccuracies
the inclusion criteria, single-room kaccha house; no adult member aged below in the SECC data. For example, privileged house-
and none of the 59; no literate adult above 25 years; landless households holds often try to sneak onto the NFSA list at the
exclusion criteria. living mainly off casual labour.
* These are inclusion criteria for priority households; there are separate criteria for Antyodaya households.
correction stage. Meanwhile, eligible households
“Registered casual workers” refer to those registered under the Unorganised Workers Social Security Act, 2008 often make unsuccessful applications to add
or Buildings and Other Construction Workers Act, 1996.
# In Odisha, exclusion criteria are waived for “automatically included” groups (mainly homeless households,
missing names to their ration cards (for example,
destitute households living on alms, PVTG [Particularly Vulnerable Tribal Groups] households, and former AAY those of children born after 2011). The sheer
households).
These criteria were applied at the time of the survey (June 2016). Some of them have been simplified for
volume of correction work tends to be difficult
presentation purposes. to manage.
38 FEBRUARY 9, 2019 vol lIV no 6 EPW Economic & Political Weekly
 SPECIAL ARTICLE

MP, also among the early movers, relied on a pre-existing Here, “after NFSA” refers to the period after June 2016 (the
database (Samagra Samajik Suraksha Mission, or Samagra for date of the survey), and “before NFSA” refers to two years
short) to identify priority households. Samagra had been built before, based on the recollection of respondents.11 As a first
in the preceding years as a general-purpose database for wel- approximation, we can assume that households with signifi-
fare schemes. Eligible households were identified using this cant PDS entitlements (for example, not restricted to kerosene)
database, based on an extensive list of inclusion criteria (Table 1). had an Antyodaya or BPL card in the pre-NFSA period and
In the initial months, eligible households were given “NFSA Antyodaya or priority cards in the post-NFSA period. Based on
slips” which laid out their new entitlements, along with the this convention, we can conclude that PDS coverage in the
basic details of the household (ration cards were issued later). sample villages rose by 25 percentage points or so—from a
In principle, the Samagra database can be continuously updated base of 58%—between mid-2014 and mid-2016.
at the gram panchayat level, though it is not clear to what Strictly speaking, some households outside these three cate-
extent that is possible at the moment. The importance of gories (AAY, BPL, and priority) also receive significant PDS enti-
decentralised management is explained below. tlements.12 We shall use the term “entitled households” to refer to
Like Chhattisgarh, Odisha adopted a self-declaration app- all those in actual possession of a ration card (or an equivalent,
roach. After eligibility criteria were notified, people were asked such as NFSA slips in MP) that gives them foodgrain entitlements
to self-declare (by filling out a form) whether they were eligi- under the PDS. The proportion of entitled households in the sam-
ble. There were two concerns to this approach—whether mar- ple villages increased from 59% to 85% between the two periods.
ginalised people (including those living in remote areas) The weighted average of the state-specific proportions of
would receive the self-declaration forms and be able to fill entitled households among the sample households (using the
them out, and whether those who were ineligible could be 2011 Census rural population as weight) is 84%. This is quite
trusted to declare their status. The active involvement of gram close to the corresponding weighted average of state-specific
panchayats helped in both respects. To screen for false decla- mandatory rural coverage under the NFSA (82%). Note, how-
rations, the government also launched several verification ever, that the latter applies to population coverage, not house-
drives through which ineligible households that had claimed hold coverage. As discussed below, the actual coverage in
priority cards were deleted and even fined (Mohanty 2017). population terms is lower than the household coverage—by 10
Exemplary action was taken against some of those who had percentage points or so in our sample.
tried to cheat, deterring others to some extent.10 An option The expansion of PDS coverage post-NFSA, as made visible in
was also provided for households to voluntarily surrender the survey, is consistent with secondary data. The National
illegitimate cards. This process, however, took much longer Sample Survey data suggest that the proportion of rural house-
than the SECC-based approach—Odisha adopted it mainly holds using the PDS for foodgrain purchases in the reference
because of repeated delays in releasing the SECC data. states was a little below 50% in 2011–12.13 Today, according to
official data, it is very close to the minimum rural coverage
Public Distribution System Coverage (82%) prescribed by the NFSA; this also applies statewise.14
The biggest gain from the NFSA for the poorest states was the
expanded coverage of the PDS. In fact, the PDS can be descri- Inclusion and Exclusion Errors
bed as “near-universal” in the six reference states, at least in In spite of the difficulties mentioned earlier, the post-NFSA
rural areas where the mandatory coverage varies from 75% in ration card lists in the reference states seem far more reliable
West Bengal to 86% in Jharkhand. and credible than the earlier BPL lists.15 The latter suffered from
This expansion is illustrated in Table 2, where we compare massive inclusion and exclusion errors. To illustrate, three nat-
the PDS status of sample households before and after the NFSA. ional household surveys show that, around 2005, about half of
all poor households in rural India did not have a BPL card.16 The
Table 2: Coverage of the PDS before and after NFSA (Joint Percentage
Distribution of Sample Households by Type of Ration Card) reduction in exclusion errors, of course, was facilitated by the
Before NFSA After NFSA expansion of PDS coverage under the NFSA. The formulation of
AAY Priority Other* None All#
simpler and clearer methods to identify eligible households,
AAY 7 6 0 1 14 (15)
BPL 2 33 2 3 41 (43)
however, also helped to avoid exclusion as well as inclusion errors.
APL 1 13 4 2 20 (21) The rates of inclusion and exclusion errors can be estimated
Other 0 1 0 0 1 (1) for the sample villages since the survey covered (almost) all the
None 1 14 0 3 19 (20) households in each village. The survey teams were trained to
Missing 1 3 0 1 5 (-) assess, based on direct observation and brief enquiries, whether
Total 13 71 6 10 100 (100) a sample household was eligible in terms of the official criteria
* Mainly special categories (RKSY-1, RKSY-2) and “composite” cards in West Bengal (where
different household members occasionally have cards of different types), and lingering BPL
applicable in the concerned state. Exclusion errors refer to cases
and APL cards. where an eligible household is not “entitled” (that is, in possession
# In parentheses, the corresponding figures when missing observations (for pre-NFSA
ration cards) are omitted.
of a valid ration card), and inclusion errors refer to cases where
(1) “After NFSA” refers to June 2016 (when the survey took place). “Before NFSA” means two an entitled household is not eligible. As Table 3 (p 40) shows,
years before that (see note 11).
(2) In Chhattisgarh, Ekal (AAY) and special (AAY) cards were counted as AAY; in West
both exclusion and inclusion errors were below 10% in most
Bengal, “special priority” cards were counted as priority. cases, suggesting the relatively reliable identification of eligible
Economic & Political Weekly EPW FEBRUARY 9, 2019 vol lIV no 6 39
SPECIAL ARTICLE

households. The fact that the exclusion errors are just around SECC list because of migration Table 4: Missing Names
5%, compared to 50% or so in the BPL approach, is particularly or other reasons; occasionally, State Proportion of Missing
Names in Priority
encouraging even after noting that some of this improvement is the SECC would skip an entire Ration Cards (%)
due to expanded coverage. hamlet. As Table 4 indicates, the Bihar 18

Table 3: Inclusion and Exclusion Errors

proportion of missing names in Chhattisgarh 18
State Mandatory Proportion Proportion Inclusion- Exclusion- priority ration cards was a little Jharkhand 15
Coverage of Eligible of Entitled Error Rate: Error Rate: over 10%, rising to 18% in Bihar Madhya Pradesh 7
under NFSA Households Households Proportion Proportion
(% of Rural among among of Entitled of Eligible and Chhattisgarh. On this count, Odisha 10
Population) Sample Sample Households Households West Bengal 12
again, MP is doing better than the
Households Households That Are Not That Are Not Six states* 13 (13)
(%) (%) Eligible# (%) Entitled (%) other sample states, reinforcing * In parentheses, weighted average
Bihar 85 80 81 13 13 the impression that the Samagra of state-specific figures (using rural
population as weight).
Chhattisgarh 84 87 95 8 4 database has some credibility.
Jharkhand 86 98 76 1 24 Because of the missing names, the actual coverage of the
Madhya Pradesh 80 89 87 2 4 PDS in population terms—the proportion of entitled persons in
Odisha 82 95 89 1 8 the sample population—is significantly lower than the cover-
West Bengal 75 83 83 9 9
age in household terms—74% and 85%, respectively, in the
Six states* n/a (82) 89 (86) 85 (84) 5 (7) 11 (10)
* In parentheses, population-weighted average of state-specific figures (using rural population,
sample villages. Thus, at the time of the survey, there was still
based on 2011 Census, as weight). much scope for adding missing names or excluded households
# The “inclusion-error rate” could be also be defined as the proportion of ineligible households
that are entitled. In that case, inclusion-error rates would be high. In Bihar, for instance, about
to the ration card list, within the mandatory coverage provided
half of all ineligible households are entitled. under the NFSA.
“Eligible households” are those that meet official eligibility criteria (Table 1). “Entitled
households” are those actually in possession of a ration card (or equivalent) that gives them
Beyond the issue of missing names, the transition to per
foodgrain entitlements under the PDS (see main text). The same calculations can be done in capita entitlements enhanced the general need for regular and
terms of individuals rather than households, but the results are similar.
reliable updating of ration card lists (in response to changes in
In Bihar and West Bengal, the percentage of inclusion and household composition). So far, this remains an elusive goal in
exclusion errors match, at close to 10% in both cases. This means all the reference states. The transition also means that many
that, in principle, both could be corrected without modifying the small households lost, rather than gained, from the NFSA.
coverage. In Jharkhand, inclusion errors are virtually nil, but
there are several exclusion errors. The reason seems to be that, Purchase–Entitlement Ratios
on the one hand, most rural households in Jharkhand are eligible, A helpful indicator of the integrity of the PDS is what we have
and on the other, the actual coverage of the PDS in Jharkhand was called the “purchase–entitlement ratio” (PER) in earlier writings
still relatively low in June 2016 (it increased later). In Odisha, as (for example, Khera 2011b; Drèze et al 2017). This refers to what
in Jharkhand, most of the sample households were eligible, and entitled households are able to purchase from the PDS as a pro-
targeting errors were heavily tilted towards exclusion errors. portion of their entitlements.17 For instance, if a household that is
It is possibly interesting that MP is doing really well as far as entitled to 30 kg of foodgrain per month actually gets, say, 27 kg
identification is concerned, with inclusion and exclusion errors per month, then the PER would be 90%. Since PDS foodgrain all-
at only 2% and 4%, respectively. This suggests that the Sama- ocations at different levels (such as district, block, and village) are
gra database is reasonably reliable. If so, this is an important typically based on aggregate entitlements at the relevant level,
pointer to the possibility of developing decentralised house- the PER can be read as an indicator of the cumulative leakages
hold lists that are regularly updated at the gram panchayat in the system. If there are any leakages down the line, they
level, instead of relying on “static” and centralised databases would be reflected in a lower average PER in the concerned area.18
such as the SECC. It is also useful to distinguish between the PER so defined
(let us call it the “unconditional PER,” or UPER)—from the
Switch to per Capita System “conditional PER” (CPER), where the latter is calculated in the
Before the NFSA came into force, PDS entitlements in the refer- same way, but after excluding entitled households that did not
ence states, and in most other states, were defined in terms of buy any rice or wheat from the PDS in the relevant month (let
household rather than per capita (typically, for foodgrain, each us call them “zero purchase households”). The idea is that
household was entitled to 25 kg or 35 kg). Under the NFSA, there are many possible reasons for nil PDS purchases (includ-
priority households received entitlements on a per capita basis. ing temporary migration and temporary disruptions in the
The per capita system is certainly more logical and equitable supply chain) that are not necessarily related to leakages.
than the earlier system of allotting entitlements to households Thus, both CPER and UPER are of interest. When zero purchase
as a whole. This transition, however, created a new type of households and other entitled households have the same aver-
exclusion error: “missing names,” that is, the omission of some age entitlements, which is approximately true in our sample, a
household members from the ration card. Children born after simple relationship holds between UPER and CPER:
2011, for instance, were often excluded from ration cards in UPER = (1 – p).CPER
states where the SECC data were used to draft the official list of where p is the proportion of zero purchase households in the
ration cards. Sometimes, adults were also overlooked in the relevant month.
40 FEBRUARY 9, 2019 vol lIV no 6 EPW Economic & Political Weekly
 SPECIAL ARTICLE

We calculated PER s using two different indicators of PDS blocks, most were compensated with double rations in May,
purchases. First, we asked respondents how much they which is why the CPER is above 100% in MP.
“normally” receive from the PDS in a particular month. Sec- Turning to Bihar and Jharkhand, some caution is required
ond, we asked them to tell us specifically how much they had while interpreting Table 5. The figures must be read in light of
bought in the two calendar months that preceded the survey the fact that in both states, PDS leakages were above 80%
(April and May 2016). The results are presented in Table 5, throughout the 2000s (Khera 2011a: Table 2). Against this
where “April–May” refers to the unweighted averages of the background, the fact that the sample households reported get-
month-specific figures.19 ting more than 80% of their entitlements in a normal month
Table 5: Purchase–Entitlement Ratios
seems encouraging. However, the figures in the first column
State “Normal Month” April–May 2016* are probably more like CPERs than UPERs. If a household gets,
(%) Conditional Proportion of Entitled Unconditional say, 80% of its entitlements in nine months out of 12, and noth-
PER Households with PER
(%) “Zero Purchase” (%) ing in the remaining three months, it is likely to report getting
(April/May) (%) 80% in a normal month (the CPER), rather than 60% (the UPER).
Bihar 84 90 20/85 44 Indeed, the figures in the first two columns of Table 5 are quite
Chhattisgarh 98 99 -/4 96
close to each other. The CPER for April–May 2016 for Bihar and
Jharkhand 85 89 26/35 62
Jharkhand is around 90%, which is still encouraging, but both
Madhya Pradesh 98 105 32/12 83
Odisha 99 100 14/5 91
states clearly experience a high incidence of zero purchases.
West Bengal 95 95 1/1 94 The bulk of this problem, especially in Bihar, is due to what
Six states# 93 (92) 97 (96) 19/22 (18/34) 76 (73) might be called “gap months,” when no foodgrain distribution
* The month of May only for Chhattisgarh, where April rations were distributed in March takes place in an entire village or hamlet in a particular month.
(see note 19). March distribution was not recorded, but the survey responses and team’s
observations indicate that most sample households in Chhattisgarh did get their April
In the case of Jharkhand, earlier surveys show that though
rations in advance. Advance distribution also shows in official data. much progress has been made in successfully avoiding gap
# In parentheses, weighted average of state-specific figures (using rural population as
weight). Figures for Chhattisgarh in April are treated as missing observations.
months, the problem has not disappeared. In both sample
blocks, we were told, temporary disruptions in the supply chain
Three of the six sample states (Chhattisgarh, Odisha, and had affected PDS distribution during the reference period. In a
West Bengal) seem to have a relatively well-functioning PDS, follow-up survey, conducted in June 2017 in 32 randomly sele-
with UPERs above 90% in April–May 2016 and almost all cted villages (spread over eight districts) in Jharkhand, we
households reporting that they normally receive their full PDS found that regular distribution every month had become the
rations. For Chhattisgarh and Odisha, this finding is in line norm. Except for a significant minority of households that
with a string of earlier studies.20 West Bengal looks like the lat- experienced transaction problems due to compulsory biometric
est entrant to the league of effective PDS reformers, but this is authentication, most households had received six months’ rations
best considered a tentative finding until it is corroborated (or out of six in the preceding six months.22 Thus, the general situ-
qualified) by other studies. It is worth noting, however, that ation in Jharkhand is likely better than what Table 5 suggests.
our findings for West Bengal are consistent with those of a In Bihar, on the other hand, there is considerable evidence
recent survey on the PDS in the Bankura district.21 that gap months continue to be a problem. Aside from
High UPERs in these states reflect a combination of high frequent disruptions in the supply chain, many PDS dealers in
CPERs and a low incidence of “zero purchase”—most people Bihar regularly divert an entire month’s foodgrain quota
were able to buy their rations in April–May, and got the full without their customers being able to protest. Some dealers in
amount. Shortfalls in the CPER from 100% reflect the fact that Bihar were quite brazen about this practice. One of them
many PDS dealers try to extract “cuts” from their customers, candidly said “Mein do mahine ka kha leta hoon” (I “eat” two
known as katauti in some Hindi-speaking areas. Successive months’ rations every year). Perhaps, he was trying to boast—
surveys over the years show that the cuts tend to decline over or pretend—that he did not go further and, in that sense, was
time, as PDS users become more assertive and PDS dealers relatively “honest.”
more accountable. In Chhattisgarh and Odisha, where private Even by Bihar standards, however, the high proportion of
PDS dealers have been replaced by community institutions zero purchase households in May 2016 (85%), which pulls
(gram panchayats in Odisha, multiple institutions in Chhattis- down Bihar’s UPER, is likely an anomaly. One recent survey of
garh), the cuts are virtually nil now, judging from Table 5. 5,000 randomly selected households in rural Bihar found that
In MP, as in Chhattisgarh and Odisha, almost all the res- priority households miss, on average, four months’ food rations
pondents said that they normally receive their full rations. over the year.23 This seems broadly consistent with other recent
This is consistent with an earlier survey conducted in 2015, surveys of the PDS in Bihar, reviewed in Drèze et al (2015).24
where we found evidence of a dramatic improvement in the The general situation in Bihar is not entirely clear; however, as
PDS in MP post-NFSA (Drèze and Khera 2015b). In April–May with Jharkhand, it is likely better than what we found in the
2016, however, the UPER was a moderate 83%. The shortfall is two sample blocks in April–May 2016.
mainly because a substantial proportion (about one-third) of Much of the lingering ambiguity in these findings relates to
sample households did not receive their food rations in April, the extent and causes of gap months. Siphoning off a whole
for reasons that are not entirely clear. In one of the two sample month’s food rations is a fairly extreme act—not the sort of thing
Economic & Political Weekly EPW FEBRUARY 9, 2019 vol lIV no 6 41
SPECIAL ARTICLE

that people take lying down, or that a vigilant administration large households tend to gain from the per capita system. For the
would fail to prevent. But, it does happen, and even if zero six states together, the proportion of losing individuals is around
purchases are due to supply bottlenecks rather than embezzle- 22%, and would be a little lower still if we take into account the
ment in most cases, a serious delivery failure would remain. reduction in issue prices under the NFSA. But, even if it is only
There is still some homework to do on this in the reference 20% or so, this reduction in entitlements is a matter of concern.
states, especially in Bihar and Jharkhand. The concern only grows when we look at the pre-NFSA status
of the gainers and losers (Table 7). Ignoring price changes, those
Gainers and Losers belonging to households formerly classified (rightly or wrongly)
Many rural households in the reference states (and elsewhere) as BPL had a similar chance—about 40%—of gaining or losing
gained from the NFSA, but some did not. The main gainers from the NFSA. Some of the BPL Table 7: Gainers and Losers by Pre-
NFSA Status (Individuals)
were households that previously had no significant foodgrain losers are relatively well-off Pre-NFSA Percentage Distribution of
entitlements and became priority or Antyodaya households households that should not have Status* Sample Population by Status#
Gainer Loser No
under the act: both poor households that had been left out of been on the BPL list in the first Change
the earlier BPL lists, and less-poor or non-poor households that place, but others are poor BPL 41 38 22
benefited from the enhanced coverage of the PDS. The main households that have been ad- APL 80 5 15
losers were former BPL households with only a few members, versely affected by the per capi- AAY 8 32 60
who used to receive more before the NFSA (25 kg to 35 kg per ta system. APL (above poverty None 100 0 0
month, depending on the state) than they do today under the line) household members, on * For the share of each category in the
sample, see Table 2.
per capita system with a priority card. Some Antyodaya the other hand, were virtually # Row entries add to 100, except for
households also lost, as they were transferred to the priority assured of not losing, and had a rounding.
category, or were even left out altogether. high chance (80%) of gaining. The reason is that most of them
Table 6a: Gainers and Losers We asked sample households did not receive significant foodgrain entitlements before the
(Households) about their foodgrain entitle- NFSA, but they turned into priority or even Antyodaya households
State Percentage Distribution of
Sample Households by Status* ments two years prior to the sur- as PDS coverage expanded under the act.
Gainer Loser No vey, that is, before the imp- The existence of a substantial minority of losers is, to some
Change
lementation of the NFSA (except extent, an inevitable consequence of transitioning to per capita
Bihar 52 21 27
Chhattisgarh 24 33 44 in Chhattisgarh, where the CFSA entitlements as opposed to household entitlements. One important
Jharkhand 53 33 15 was operational by mid-2013). category of victims is widows living alone and elderly couples.
MP 62 18 20 This enabled us to identify the In principle, the Antyodaya programme should have protected
Odisha 45 41 14 gainers and losers by comparing them. As discussed below, however, this safeguard fell short.
West Bengal 54 30 15 food entitlements before and
Six states# 48 29 22 after the NFSA (Tables 6a and 6b). The Antyodaya Programme
(52) (27) (22)
Note that, as before, we are The NFSA prescribes the continuation of the Antyodaya progra-
* Row entries add to 100, except for
rounding. looking here at effective entitle- mme (intended for the poorest of the poor), but for some reason,
# In parentheses, weighted average of
ments (based on ration cards) central and state governments have paid little attention to it.
state-specific figures (using rural
population as weight). rather than legal entitlements. The central government even made a brief attempt—aborted
Figures are based on comparing effective
About half of the sample under public pressure—to phase it out (Drèze 2015).
entitlements in June 2016 (after NFSA)
and two years earlier (before NFSA). households gained from the Perhaps because of this declining policy interest, the num-
In Chhattisgarh, the CFSA applied in
NFSA, and close to 30% lost ber of Antyodaya cards fell sharply in many states immediately
both periods but the transition to per
capita entitlements occurred in between (Table 6a). The rest stayed in the after the act came into force. In the sample villages, the num-
(see main text).
same place in terms of quantity ber of Antyodaya cards in June 2016 was about 20% lower
Table 6b: Gainers and Losers entitlements, but in three refer- compared to two years earlier; only Bihar and Chhattisgarh
(Individuals)
ence states, they benefited to more or less preserved their pre-NFSA coverage. More
State Percentage Distribution of
Sample Population by Status* some extent from the reduction importantly, almost 50% of households that used to have an
Gainer Loser No in issue prices under the NFSA.25 Antyodaya card had lost it by June 2016 (Table 8).
Change
Bihar 58 15 27
If we also count them as gainers, Table 8: Antyodaya
State Proportion of Absolute Number of Antyodaya Proportion of
Chhattisgarh 25 30 45 most households benefited from
Entitled Households Cards among Sample Households Pre-NFS Antyodaya
Jharkhand 55 30 15 the act. Even after taking into in Antyodaya Households That
Category Before NFSA After NFSA Had No Antyodaya
MP 72 10 18 account the benefits of lower
(%) (Mid-2014) (Mid-2016) Card Post-NFSA (%)
Odisha 48 36 16 issue prices, however, the pro- Bihar 8 51 47 16
West Bengal 55 28 16 portion of losers would not be Chhattisgarh 28 141 147 40
Six states# 54 23 23 much below 30%. When calcu- Jharkhand 16 108 74 73
(56) (22) (22)
* Row entries add to 100, except for
lations are done in terms of indi- MP 11 102 65 39
rounding. viduals rather than households Odisha 7 63 46 40
# In parentheses, weighted average
of state-specific figures (using rural
(Table 6b), the proportion of West Bengal 15 76 56 74
population as weight). gainers is a little higher, because Six states 14 541 435 49

42 FEBRUARY 9, 2019 vol lIV no 6 EPW Economic & Political Weekly

 SPECIAL ARTICLE

It is possible that some of those who lost their Antyodaya distribution schedules, electronic weighing, computerisation,
cards had not been eligible in the first place.26 This is difficult to and a range of transparency and grievance redressal measures.
verify as the eligibility criteria are not clear.27 The observations However, arguably, the most critical reform in Chhattisgarh was
of the survey teams suggest that the post-NFSA distribution of the deprivatisation of ration shop management; private dealers
Antyodaya cards was somewhat haphazard, possibly due to this were removed in favour of collective institutions such as gram
lack of clear eligibility criteria. Quite likely, many households panchayats, cooperatives, and women’s self-help groups (SHGs).
lost their Antyodaya cards for no valid reason after the act came Deprivatisation has been less widely emulated, perhaps be-
into force (we met some of them during the survey). cause it tends to be fiercely resisted by private dealers and
The Antyodaya programme is an important means of provi- requires firm political will. Among the reference states, Odisha
ding enhanced support to the poorest households, but it requires is the only one (other than Chhattisgarh) that went the whole
urgent attention. The eligibility criteria lack transparency. hog; most ration shops there are now managed by gram pan-
Supreme Court orders are not being followed. Some households chayats. In MP, many ration shops are managed by coopera-
were arbitrarily deprived of their Antyodaya card post-NFSA. tives.33 In some states, new ration shop licences are often
The entire programme seems to have fallen off the radar.28 reserved for women’s SHGs or cooperatives, but existing private
dealers are rarely removed. In Jharkhand and (especially)
Quality of PDS Foodgrain Bihar, the overwhelming power of private dealers over their
In line with our previous PDS surveys, we found that a large customers remains a major problem. Even in West Bengal,
majority of cardholders were reasonably satisfied with the quality some of the private dealers we met were far from inspiring.
of PDS foodgrain—79% of respondents felt that it was “good” or One of them casually told the survey team, “Of course, I cheat—
“fair” (Table 9).29 Two states, however, had serious problems with so did my father, and hopefully my son will do the same.”
regard to low-quality PDS foodgrain—Bihar and West Bengal. Another area for improvement in PDS reform is effective ration
Table 9: Quality of PDS Foodgrain card management. Computerised ration card management,
State Percentage Distribution of Entitled Households by Perceived Quality of which began in a handful of states in the early 2000s, was well
PDS Foodgrain*
Good Fair Poor Rice Was OK Wheat Was OK
underway in many others by the time the NFSA was implement-
but Not Wheat but Not Rice ed. The rollout of the act forced laggard states to catch up, with
Bihar 20 37 23 6 13 impressive results in some cases (for example, in Jharkhand and
Chhattisgarh 63 34 2 0 0 West Bengal). Computerisation enables transparency and is an
Jharkhand 40 52 5 2 0 important step in keeping corruption at bay. None of the refer-
MP 20 53 16 2 9
ences states, however, have been able to establish a system that
Odisha 50 36 12 0 1
continuously and reliably updates the list of ration cards—add-
West Bengal 27 31 10 32 0
Six states# 38 (30) 41 (40) 11 (15) 6 (9) 4 (6)
ing the names of newborns, updating the list when people
* Based on the most recent quota. Row entries add to 100, except for rounding. move (for example, after marriage), deleting the names of peo-
# In parentheses, weighted average of state-specific figures (using rural population as weight). ple who die, and so on. As noted earlier, an attempt was made
In West Bengal, the main problem was the poor quality of in MP to use the Samagra system for this purpose, but not every-
flour packets (on this, see Drèze and Maji 2016; Haldar and one was aware of this facility; those in charge of updating
Basu 2018). Some respondents said that they fed the flour to records did not always do their job (and some, we were told, were
cattle, or that they had to sieve the flour twice for it to be edi- charged for it); and there were some complaints about Sama-
ble. Private contractors, it seemed, were making money by gra data not being well-integrated with food department data,
supplying low-quality flour packets for public distribution.30 so that revisions in the former were not always reflected in the
Perhaps the flour packets should be replaced with wheat or latter. Aside from effective ration card management, other areas of
rice. If flour distribution continues in West Bengal, the quality ongoing reform efforts include reliable recording of last-mile
of the flour packets should be strictly monitored. transactions, portability of benefits, and grievance redressal.34
Bihar quite likely has an adulteration problem; downgrading The active involvement of gram panchayats in the rollout of
PDS foodgrain—for example, by mixing rice with straw, or sub- the NFSA was evident in at least three states (Chhattisgarh, MP,
stituting PDS rice with inferior rice—enables corrupt dealers to and Odisha), and those states certainly seemed to benefit from it.
make some extra money. A full 42% of respondents in Bihar Given certain constraints, they were able to decentralise impor-
complained about the low quality of PDS wheat, or rice, or tant aspects of implementing the act, such as the identification
both.31 Here, again, Bihar seems to lag far behind the other ref- of eligible households and the management of ration shops. As
erence states, despite significant improvements in recent years. discussed below, however, the NFSA has also created some
centralisation tendencies.
PDS Reforms
All the reference states have emulated, to a varying extent, the State-specific Issues
PDS reforms initiated in Chhattisgarh about 12 years ago.32 We proceed with a brief account of some state-specific issues.
These include broader coverage, clearer entitlements, separa-
tion of transport agencies from distribution agencies (known as Odisha: The survey suggests that the PDS in rural Odisha
“doorstep delivery”), viable distribution commissions, fixed works reasonably well. This adds to a series of recent studies
Economic & Political Weekly EPW FEBRUARY 9, 2019 vol lIV no 6 43
SPECIAL ARTICLE

(cited earlier) that reached a similar conclusion.35 Odisha, it with ABBA, this system has the major advantage of avoiding a
seems, was the first state to emulate Chhattisgarh’s PDS dependence on connectivity and biometrics at the time of
reforms, with similar results. transaction. So far as we can tell from the limited observations
The main flaw we observed in Odisha in June 2016 was that generated by the survey, it was working reasonably well in MP
PDS distribution was lagged by a month. The monthly distribu- at that time, but this requires further corroboration.
tion schedule (which used to be diligently observed across the
state) was disrupted, temporarily at least, during the transi- Jharkhand: In Jharkhand, the rollout of the NFSA was still in
tion to the NFSA system. progress at the time of the survey, and a number of transition
problems were yet to be resolved. For instance, the online
West Bengal: At the time of the survey (and even more so ration card lists did not always match what we found on the
before the NFSA came into force), West Bengal had a unique ground—some households we encountered had new ration
problem—unnecessary complexity in the PDS. Prior to the cards but were yet to be added to the online list, while others
NFSA, household entitlements were virtually impossible to un- were still waiting for their new ration cards despite being on
derstand even for the people concerned—there were all sorts of the list and receiving food rations. When a follow-up survey
bewildering complications, variations, and exemptions. After was conducted one year later (Drèze et al 2017), it was found
the NFSA, the system was simplified to some extent, but it is still that there was much greater consistency between the virtual
far too complex, with, for instance, individual instead of house- and actual situations, though some households were still wait-
hold ration cards (in some households, different members even ing for their printed rations cards.
have different types of cards), multiple ration card categories, Some of the transition problems we encountered in Jhar-
double-decimal issue prices, weekly rather than monthly distri- khand in June 2016 were exclusion errors, missing names in
bution, and special packages for particular areas or groups.36 ration cards, and incomplete PDS reforms (for example, door-
All this begs further simplification. Indeed, the first step in step delivery was still lacking in some areas). The battle against
eradicating corruption is to ensure that people clearly under- corruption, of course, was also far from over despite the state
stand their entitlements and other basic features of the PDS. having made considerable progress in recent years. Similarly,
As mentioned earlier, except for Haldar and Basu (2018), we much progress has been made in ensuring regular monthly dis-
are not aware of other studies on the PDS in West Bengal in the tribution; however, supply disruptions still occur. With accelerated
post-NFSA period. Further research is needed to corroborate PDS reforms, Jharkhand may well be able to achieve the same
the findings reported in this paper and take them forward. standards of PDS management as, say, Chhattisgarh and Odisha.

Madhya Pradesh: In MP, the survey revealed two innovations Bihar: Bihar is a case of the proverbial glass that is half empty
of interest—the use of the Samagra database for ration card man- or half full. On the one hand, Bihar is the clear laggard among
agement, and the use of point of sale (PoS) machines in offline the six reference states, especially when it comes to removing
mode for last-mile authentication and to record PDS transactions. PDS corruption. On the other hand, Bihar has improved a
The Samagra database, as discussed earlier, is used to update great deal since the 2000s, when leakages were as high as
the PDS database from time to time. This is an important 80%–90% year after year (Khera 2011a; Drèze et al 2015).38 It
innovation, compared with the use of a static ration card data- is also worth noting that Bihar is one of the states that took the
base (derived, say, from a baseline survey such as the SECC). lead in implementing the NFSA. And indeed, it set a useful prec-
However, information from the Samagra database does not edent in using SECC data for the selection of eligible households.
automatically reflect in local PoS machines. Sometimes, spell- Aside from persistent corruption, Bihar appears to have a
ing mismatches between the two databases lead to the denial specific problem with regards to limited logistic capability. In
of PDS rations. Further, some respondents said that the gram most of the reference states, foodgrain distribution witnessed
panchayat secretary demanded money (up to `100 in one a quantum jump with the implementation of the NFSA. Bihar, it
village) to correct or update Samagra entries. Samagra is best seems, had great difficulty coping with the logistics. As men-
seen as an ongoing experiment, awaiting closer scrutiny. tioned earlier, the problem of “gap months” arises both from
The system of using offline PoS machines is also of interest. disruptions in the supply chain and from the tendency of PDS
Unlike the PoS machines used in some states today for Aad- dealers to use that opening to divert an entire month’s
haar-based Biometric Authentication (ABBA)—for example, in foodgrain quota from time to time.
Jhark hand and Rajasthan—the PoS machines used in MP at The fact that Bihar, and (to a lesser extent) Jharkhand, emer-
the time of the survey did not require internet connectivity or ged as the two laggard states in this survey will come as no sur-
biometrics.37 Upon entering a household’s Samagra number prise to anyone familiar with the feudal social structures and
(printed on the provisional NFSA slips that acted as temporary low standards of governance in those states. One aspect of the
ration cards in MP in June 2016), the PoS machine confirms the governance problem is that Bihar and Jharkhand’s food depart-
household’s entitlements, generates a printed receipt, and ments are simply lacking; there are staff shortages and vacancies
records the transaction. Uploading transaction records takes at all levels. In this situation, they tend to rely on PDS dealers to
place separately (along with downloading updated household carry out tasks that government functionaries in other states are
information), as and when connectivity is available. Compared responsible for, such as collecting ration card applications or
44 FEBRUARY 9, 2019 vol lIV no 6 EPW Economic & Political Weekly
 SPECIAL ARTICLE

seeding Aadhaar numbers. Given the corrupt tendencies of PDS 2016 survey and instructed them to take a photograph of each
dealers (who are often upper-caste exploiters with political PDS transaction, to be uploaded on the net in due course. Two
connections), this approach is bound to be problematic. Some other options are offline PoS machines (as in MP at the time of
of this, of course, is a reflection of the state government’s low the survey) and smart cards. The smart card option, it seems,
political commitment to food security issues. There have been is already being used with good effect in Tamil Nadu (Khera
episodes of political interest in these issues in Bihar and 2018).40 The central government, however, is relentlessly
Jharkhand—for example in the run-up to the asembly elec- promoting Aadhaar-based technology.
tions—but they have tended to be short-lived. In states like
Chhattisgarh and Odisha, by contrast, there is more sustained Concluding Remarks
interest in food security issues. Correspondingly, the food dep- India’s PDS has come a long way in the last 10 years or so. It has
artments have strong teams with reasonable tenure, and there turned from a morass of leaky and poorly targeted transfers to a
are better administrative capabilities at lower levels too. critical form of social support for a vast majority of poor house-
None of this, of course, is immutable. Time will tell whether holds. In some states, the market value of the monthly PDS trans-
Bihar and Jharkhand are able to achieve the same level of PDS fers for eligible households is now roughly equivalent to one week’s
functionality and integrity as the leading states. All said and earnings under the Mahatma Gandhi National Rural Employment
done, they are certainly moving in that direction. Guarantee Act, without the beneficiaries having to work.41
This transformation did not begin with the NFSA, but the act
Chhattisgarh: We end this section on a happy note. In June contributed to it. In the six reference states, it certainly helped
2016, as in earlier surveys, we were impressed with the PDS in to put the PDS on a more solid footing, with lower exclusion
Chhattisgarh. Regular distribution every month, without cuts, errors, enhanced transparency, greater political visibility, and
was the norm. Ration card entries were complete and accurate higher operational standards. Bold PDS reforms are in progress
in most cases. Further, the respondents were generally happy in all the reference states. Chhattisgarh, the pioneer in that
with the local PDS manager. We did, of course, encounter various field (at least in the eastern region), has set standards that now
issues such as missing names in ration cards, some puzzling com- seem achievable in the other states as well.
plaints of cards being cancelled, and the first signs of Aadhaar- Having said this, the transition is far from complete. Bihar
related woes. The mandatory taking of photographs at the time and Jharkhand are yet to achieve the standards set by other
of distribution was also causing problems for the elderly, who states, and in Bihar especially, the hold of corrupt PDS dealers
were not always able to go to ration shops in person. Overall, how- is proving hard to break. Further, much remains to be done
ever, the system’s effectiveness was remarkable. everywhere to ensure effective ration card management, reli-
able recording of last-mile transactions, and prompt grievance
Recent Developments redressal, among other remaining challenges.
As mentioned earlier, some states have been able to decentralise We end by sharing a few thoughts on possible courses of
important aspects of the implementation of the NFSA. Having action, not only for the government, but also for researchers
said this, the NFSA has also created some centralisation tendencies. and concerned citizens.
Over time, the central government has increasingly used its powers First, further evidence on the status of the PDS and NFSA, not
to remotely control the PDS across the country. One striking only in the reference states but across the country, is urgently
example concerns the recent imposition of Aadhaar integration required. It is hard to think of an issue of equal import that has
on the PDS. This requires all members of entitled households to received such little attention so far from the research commu-
submit their Aadhaar numbers so that they can be linked with the nity. Our own study has its limitations, given the relatively
PDS database. Aadhaar integration had already started causing small sample size. For many other states (including some like
difficulties in some of the reference states in mid-2016.39 Uttar Pradesh, where the NFSA could make a big difference),
Going one step further, the central government recently even limited evidence of this sort is not available.
started pressurising state governments to introduce ABBA veri- Second, the success of the PDS depends on the provision of
fication to collect PDS entitlements. This requires fingerprint simple, transparent, and durable entitlements. Some states are
authentication of at least one household member each month, already introducing complications, such as multiple categories
at the time of buying PDS rations. In Jharkhand, compulsory of ration cards and variable issue prices. Clarifying entitle-
ABBA was introduced in the Ranchi district in August 2016 and ments is the first line of defence against corruption.
was gradually extended to the whole state by May 2017. The Third, private dealers are begging to be removed, in favour
results were sobering: major exclusion problems, enhanced of public institutions.42 States like Chhattisgarh and Odisha
transaction costs, and—if anything—a revival of corruption (and others before them, notably Tamil Nadu) have shown the
(Drèze et al 2017). In short, this move turned out to be a set- way—it is largely a matter of political will.
back rather than a step forward for PDS reforms in Jharkhand. Fourth, the Antyodaya programme needs a new lease on life.
Jharkhand’s experience shows the dangers of a dependence on With a coverage of about 20 million households in the whole
internet connectivity and biometrics for PDS authentication, country, it could go a long way towards eliminating destitution.
especially in rural areas. Chhattisgarh, hoping to avoid this However, this requires clear eligibility criteria and a reliable
pitfall, supplied ration shops with tablets shortly before the selection process, consistent with Supreme Court orders.
Economic & Political Weekly EPW FEBRUARY 9, 2019 vol lIV no 6 45
SPECIAL ARTICLE

Fifth, the selection of eligible households requires more Finally, despite recent improvements, the PDS still has a long
thought. Very soon, the current lists of ration cards will be out way to go in terms of standards of transparency and accounta-
of date in many states. Is it time to plan another SECC-type bility. Strict action against corrupt dealers, for instance, would
national survey? Or is there a better way, based for instance on help—they still tend to go scot-free even after being caught.
self-declaration, or the agency of gram panchayats, or even— Similarly, most of the reference states are yet to enforce basic
in some areas at least—on the universalisation of the PDS? The transparency norms at the ration shop, such as the mainte-
answer is far from obvious. nance of clear and legible information boards. Effective griev-
Sixth, technology can help reform the PDS and reduce corrup- ance redressal facilities are also lacking at all levels.
tion, but only if it is appropriate, reliable, and people-friendly. Ex- These are just a sample of the issues that emerge from the
ploring or evaluating alternatives to ABBA, based, for instance on findings of the survey. They hardly span the NFSA’s wide can-
smart cards or offline PoS machines, would be particularly useful. vas, but we hope that some of the issues are clearer.

Notes with the selection of eligible households). In 23 See Dar et al (forthcoming), an insightful study
1 Maternity entitlements were yet to be operation- the reference states, other than Chhattisgarh, of the PDS in Bihar. In their sample, as in ours,
alised at the time of writing. Child nutrition pro- this happened sometime within the two years the CPER was close to 90%. Incidentally, their
grammes (mainly school meals and the Integrated preceding the survey. study also sheds light on the continued inci-
Child Development Services [ICDS] programme) 12 To illustrate, West Bengal has an additional dence of hunger in Bihar—in 36% of the sam-
were largely in place before the NFSA. category called RKSY–I (Rajya Khadya Suraksha ple households, at least one member had gone
2 Earlier studies on the PDS in the post-NFSA Yojana) with the same entitlements as priority hungry (for lack of food in the house) in the
period include National Council of Applied households, but which are paid for by the state preceding 30 days.
Economic Research (2015) and Puri (2017). For government. 24 An even more recent study of the PDS in Bihar,
more on the PDS and PDS reforms before the 13 Calculated from unit-record NSS (68th Round) covering about 1,200 households in 95 villages,
NFSA, see Khera (2011a, 2011b), Drèze and data; similarly, the proportion of households arrived at a surprisingly high estimation of
Khera (2013, 2015a, 2017), Drèze (2017a), and with a BPL or AAY card was around 45% in the UPERs—about 90% on average (National Coun-
the literature cited therein. For an introduction reference states at that time. For related statis- cil of Applied Economic Research 2015).
to the wider literature on food subsidies and tics, see Himanshu and Sen (2013), Puri (2017), 25 In the other three states (Chhattisgarh, Jharkhand,
related issues, see Bhattacharya et al (2018) and Bhattacharya et al (2018). and Odisha), issue prices had already been re-
and Alderman et al (2018). 14 Detailed information on statewise PDS coverage duced before the NFSA.
3 State governments, of course, are free to expand in November 2016 is available from the central 26 Some Antyodaya households probably surren-
the coverage or entitlements beyond NFSA norms government’s Foodgrains Bulletin (2017: 47). dered their cards voluntarily, in exchange for a
at their own expense. Among the six states dis- Note, however, that the coverage figures assume priority card. Households with more than sev-
cussed in this paper, however, only Chhattisgarh the 2011 population as the denominator. en members stand to gain from doing this, and
(and to a lesser extent, West Bengal) had done 15 For an enlightening case study on BPL lists, see some states (for example, Madhya Pradesh)
so at the time of the survey. Agrawal and Agrawal (2014). That study per- allowed it. Only a small proportion of Antyodaya
4 Most rural households in Chhattisgarh fall under tains to Uttar Pradesh, but similar problems households, however, are likely to have used
one of the two standard categories (priority and used to apply in the reference states, where the this provision.
Antyodaya), though the CFSA also provides for BPL methodology was much the same. 27 In the initial scheme, Antyodaya households
excluding well-off households using simple ex- 16 These are the National Sample Survey, the were supposed to be identified by the local
clusion criteria, and for a residual category of third National Family Health Survey, and the community. Later on, a Supreme Court order
“general households” with reduced entitlements. first India Human Development Survey. See (dated 2 May 2003) made Antyodaya cards an
5 In Bihar, since the first district (Gaya) was ac- Drèze and Khera (2010a: Table 1) and Desai entitlement for six specific groups of vulnera-
tually quite close to the state capital, we chose et al (2010: Table A.13.1a). ble households, such as Particularly Vulnerable
one away from the state capital (Jamui) as the 17 For the purpose of PER computations, household Tribal Groups (PVTG) and single women with
second district. entitlements are calculated using the number no regular support. But this order was largely
6 They also looked for duplicate ration cards but of members on the ration card (this is the basis ignored in most states, except for the universal
found very few. of foodgrain allocations to ration shops). These coverage of PVTG households under Antyodaya.
7 It is also worth noting that the NFSA requires “effective entitlements” are not necessarily the 28 The recent wave of hunger-related deaths in
ration cards to be issued in the name of the eldest same as their legal entitlements, because of the Jharkhand (at least 12 have been reported in
woman in the household. This is a progressive possibility of missing names on ration cards. the last few months) is another sign that the
step which made it much harder to match the 18 Strictly speaking, this statement may not apply Antyodaya programme is not working the way
NFSA list with the door-to-door survey because if a significant proportion of ration cards are it should. Most victims belonged to families
women’s names are not always well known in fake. As mentioned earlier, however, we found that should have been (but were not) receiving
Indian villages. That was a problem in all the little evidence of this. food rations under the Antyodaya scheme.
reference states, but perhaps especially in Bi- 19 In Chhattisgarh, most sample households did 29 For similar findings from two earlier surveys
har and Jharkhand where a woman’s name not purchase PDS foodgrain in April, but this (in 9 and 10 states, respectively), see Khera
tends to be considered very private. is because they received their April rations in (2011b) and Drèze and Khera with the PEEP
8 The contrast between the two approaches is dis- advance (in March). According to food depart- Team (2014). Muralidharan et al (2018b) also
cussed in Drèze and Khera (2010a). Inclusion ment officials in Chhattisgarh, this is a common find relatively high levels of satisfaction with
and exclusion criteria can be used and combined practice—April’s rations are often distributed the quality of PDS foodgrain in Jharkhand.
in diverse ways. As discussed in that paper, how- in March to ease the storage constraints that 30 The flour is supposed to be fortified (with iron
ever, simplicity and transparency are paramount. crop up in April. and vitamin A). Ironically, the alleged need for
9 There are a few other exclusion criteria, such 20 See Drèze and Khera (2010b), Aggarwal (2011), fortification is often used as an entry point by
as payment of income tax, but these are largely Puri (2012), Garg (2013), Chatterjee (2014), private contractors, who actually supply low-
redundant. For example, any household paying Drèze and Khera with the PEEP Team (2014), quality food packets. There have been many
income tax in rural Jharkhand is likely to also among others. examples, in recent years, of contractors mak-
meet some of the other exclusion criteria. 21 See Haldar and Basu (2018), based on a survey ing money by supplying substandard food
10 See, for example, Press Trust of India (2015), of 424 randomly selected households spread items to the PDS, the midday meal scheme, and
among other similar newspaper reports of over four blocks of Bankura. Among other rele- the ICDS. (Among other notorious cases is the
that period. For a detailed account of the se- vant insights, the authors found a UPER of 96% plundering of ICDS funds in Uttar Pradesh by
lection of eligible households in Odisha, see and reasonably effective targeting. This is the Ponty Chadha, a flamboyant gangster-business-
Satpathy (2017). only post-NFSA study of the PDS in West Bengal man close to the then chief minister, Mayawati,
11 The NFSA officially came into force in mid- that we are aware of. For a useful supplement of under the garb of supplying a ready-to-eat food
2013, but PDS-related provisions were imple- personal observations, see Rana (2018). mixture called panjiri.) West Bengal’s experience
mented later, at different dates in different 22 See Drèze et al (2017: Table 4). In their sample, is another healthy warning about this problematic
states, whenever they were ready (for example, the CPER was 93%. aspect of fortification schemes.

46 FEBRUARY 9, 2019 vol lIV no 6 EPW Economic & Political Weekly

 SPECIAL ARTICLE
31 In some areas, including in large parts of Bihar Agrawal, Aman and Ashutosh Agrawal (2014): Chhattisgarh, India,” Poverty in Focus, No 25,
and Jharkhand, people strongly prefer usna “Manufacturing Confusion: BPL Lists in Uttar 28 August, International Policy Centre for In-
(parboiled) rice to arwa (non-parboiled) rice, and Pradesh,” Economic & Political Weekly, Vol 49, clusive Growth, Brasilia.
complain when they get arwa rice from the PDS. No 14, pp 21–24. Haldar, U and D Basu (2018): “Implementation Status
Although the issue here is not so much “low Bhattacharya, S, V L Falcao and R Puri (2018): “The of the National Food Security Act (NFSA) in
quality” as the provision of an unsuitable or un- Public Distribution System in India: Policy Evo- Drought Prone Bankura District of West Bengal,”
appreciated variety of rice, complaints of this sort lution and Program Delivery Trends,” The 1.5 Bil- International Journal of Management Studies,
would come across in survey responses (Table 9) lion People Question: Food, Vouchers, or Cash Vol 5, Nos 1–3, pp 126–34, http://research-
as instances of low-quality PDS foodgrain. Transfers? H Alderman, U Gentilini and R Yemtsov ersworld.com/ijms/vol5/issue1_3/Paper_15.pdf.
32 For further discussion of PDS reforms in Chhat- (eds), World Bank, Washington, DC, pp 43–105. Himanshu and A Sen (2013): “In-Kind Food Trans-
tisgarh see, for example, Drèze and Khera Chatterjee, Mihika (2014): “An Improved PDS in a fers,” Economic & Political Weekly, Vol 48,
(2010b), Puri (2012), Garg (2013), and Bhat- ‘Reviving’ State: Food Security in Koraput, Nos 45–46, pp 46–54.
tacharya et al (2018). There have been further Jain, P, R Gupta and Y Haider (2014): “Chhattisgarh’s
Odisha,” Economic & Political Weekly, Vol 49,
developments in the last few years, such as the Food ATMs: Portable Benefits Minus Biometrics,”
No 45, pp 49–59.
deployment of the “CORE PDS” in urban areas India Together, 15 September, http://indiatogeth-
(Joshi et al 2016). Choithani, C and B Pritchard (2015): “Assessing
Bihar’s Coupon-Based PDS,” Economic & Politi- er.org/core-pds-smart-system-in-raipur-chhat-
33 One of the ration shops in our West Bengal tisgarh-food-security-portability-government.
sample (Kasol village, Bankura district) was cal Weekly, Vol 50, No 3, pp 61–68.
Dar, A, C Kumar and P Verma (forthcoming): “Is Joshi, A, D Sinha and B Patnaik (2016): “Credibility
managed by a cooperative. Interestingly, it per- and Portability? Lessons from CORE PDS Re-
formed better than other ration shops in West Bihar the New Poster Boy for Public Distribu-
forms in Chhattisgarh,” Economic & Political
Bengal (Drèze and Maji 2016). tion System Reforms?” Working Paper, Inter-
Weekly, Vol 51, No 37, pp 51–59.
34 Across reference states, whenever an attempt national Growth Centre, New Delhi.
Khera, Reetika (2011a): “Trends in Diversion of Grain
was made to update, deduplicate, or verify ra- Desai, S B, A Dubey, B L Joshi, M Sen, A Shariff,
from the Public Distribution System,” Economic
tion cards, it led to the unfair exclusion of some and R Vanneman (2010): Human Development
& Political Weekly, Vol 46, No 21, pp 106 –14.
eligible households. None of the states appear in India: Challenges for a Society in Transition,
— (2011b): “Revival of the Public Distribution Sys-
to have a process of serving show-cause notices New Delhi: Oxford University Press.
tem: Evidence and Explanations,” Economic &
before cancellation, to give the concerned Drèze, Jean (2015): “No Mercy for the Poor,” Wire, Political Weekly, Vol 46, Nos 44–45, pp 36–50.
households a chance to object. 17 June, https://thewire.in/economy/no-mercy-
— (2017): “Impact of Aadhaar on Welfare Pro-
35 One qualification is due—much of the recent evi- for-the-poor.
grammes,” Economic & Political Weekly, Vol 52,
dence pertains to eastern and southern Odisha, — (2017a): Sense and Solidarity: Jholawala Econom- No 50, pp 61–70.
especially the former “KBK” (Kalahandi–Bol- ics for Everyone, Ranikhet: Permanent Black.
angir–Koraput) region. The KBK region used to — (2018): “Smarter Than Aadhaar: Government’s
— (2017b): “Chronicle of a Starvation Death Fore- Insistence on Disruptive Option Is Bewilder-
benefit from special PDS provisions, and it is told: Why It Is Time to Abandon Aadhaar in the
possible that the PDS functions better there ing,” Business Standard, 13 March, https://
Ration Shop,” Scroll.in, 21 October, https:// www.business-standard.com/article/opinion/
than in western Odisha. scroll.in/article/854847/chronicle-of-a-starva- how-successfully-last-mile-authentication-has-
36 During the survey, individual ration card details tion-death-foretold-why-it-is-time-to-abandon- recorded-pds-118031301260_1.html.
(for example, entitlements) were aggregated aadhaar-in-the-ration-shop.
by the household to calculate the correspond- Mohanty, Meera (2017): “Aadhaar Seeding Respon-
Drèze, Jean and R Khera (2010a): “The BPL Census sible for Only 4% of the Cancelled Ration Cards
ing details at the household level.
and a Possible Alternative,” Economic & Politi- in Odisha,” Economic Times, 14 November, htt-
37 On ABBA in Jharkhand, see Drèze et al (2017) cal Weekly, Vol 45, No 9, pp 54–63. ps://economictimes.indiatimes.com/news/
and Muralidharan et al (2018b). One year after
— (2010b): “Chhattisgarh Shows the Way,” Hindu, politics-and-nation/aadhaar-seeding-responsi-
being introduced, this experiment looked like
13 November, https://www.thehindu.com/fea- ble-for-only-4-of-the-cancelled-ration-cards-
a case of “pain without gain.”
tures/magazine/Chhattisgarh-shows-the-way/ in-odisha/articleshow/61645590.cms.
38 For more on the PDS in Bihar, see also Mooij
article15685530.ece. Mooij, Jos (2001): “Food and Power in Bihar and
(2001), Rahman (2014), Choithani and Pritchard
— (2013): “Rural Poverty and the Public Distribu- Jharkhand: PDS and Its Functioning,” Economic
(2015), and Dar et al (forthcoming). PDS re-
tion System,” Economic & Political Weekly, & Political Weekly, Vol 36, No 34, pp 3289–99.
forms, it seems, were particularly intensive be-
fore the assembly elections in 2015, but it is not Vol 48, Nos 45–46, pp 55–60. Muralidharan, K, P Niehaus and S Sukhtankar (2018a):
clear to what extent the momentum was sus- — (2015a): “Understanding Leakages in the Public “Balancing the Costs and Benefits of Aadhaar,”
tained after that. Distribution System,” Economic & Political Livemint, 6 March, https://www.livemint.com/
Weekly, Vol 50, No 7, pp 39–42. Opinion/qFeWbHCXPq1wNZ43wfk3RP/Bal-
39 Among these difficulties is the possible need to
ancing-the-costs-and-benefits-of-Aadhaar.html.
remove inconsistencies between the Aadhaar — (2015b): “Thought for Food,” Outlook, 31 August,
and PDS databases. On related issues, see https://www.outlookindia.com/magazine/ — (2018b): “Authentication and Targeted Transfers,”
Khera (2017). story/thought-for-food/295144. draft paper, University of California San Diego.
40 See also Jain et al (2014) on the use of smart — (2017): “Recent Social Security Initiatives in NCAER (2015): Evaluation Study of Targeted Public
cards in Chhattisgarh’s “CORE PDS.” India,” World Development, Vol 98, pp 555–72. Distribution System in Selected States, National
Council of Applied Economic Research, New Delhi.
41 This applies in Chhattisgarh (Drèze and Khera Drèze, Jean, J Pudussery and R Khera (2015):
2013). Muralidharan et al (2018b) find that PDS Press Trust of India (2015): “Nine Govt Employees
“Food Security: Bihar on the Move,” Economic
cardholders in Jharkhand would be willing to Suspended for Applying NFSA Ration Cards,”
& Political Weekly, Vol 50, No 34, pp 44–52.
exchange their PDS rations for cash transfers of 21 August, http://zeenews.india.com/news/odis-
Drèze, Jean and R Khera, with the PEEP Team ha/nine-govt-employees-suspended-for-apply-
`1,028 per household per month on average, (2014): “A PEEP at Another India,” survey re-
which is also similar to a week’s earnings un- ing-nfsa-ration-cards_1650598.html.
port, Indian Institute of Technology Delhi; Puri, Raghav (2012): “Reforming the Public Distribu-
der NREGA. Note, however, that this is 36% partly published in Outlook, 24 March, https://
higher than the market value of PDS entitle- tion System: Lessons from Chhattisgarh,” Eco-
www.outlookindia.com/magazine/story/wa- nomic & Political Weekly, Vol 47, No 5, pp 21–23.
ments in Jharkhand, which is `754 per card-
ter-for-the-leeward-india/289801.
holding household per month on average, ac- — (2017): “India’s National Food Security Act
cording to the same study. Drèze, Jean and S Maji (2016): “PDS Has Improved (NFSA): Early Experiences,” LANSA Working
in West Bengal, but It’s Still Not Up to the Paper Series, Vol 2017, No 14, MS Swaminathan
42 Fostering competition between dealers (as
Mark,” Indian Express, 16 September, https:// Research Foundation, Chennai.
Chhattisgarh did under the CORE PDS) is an-
other possibility. However, this is not easy to do indianexpress.com/article/opinion/columns/ Rahman, Andaleeb (2014): “Revival of the Public
in rural areas with small and distant settlements. mamata-banerjee-public-distribution-system- Distribution System: Expansion and Outreach,”
west-bengal-food-security-act-3033234/. Economic & Political Weekly, Vol 49, No 20,
Drèze, Jean, N Khalid, R Khera and A Somanchi pp 62–68.
(2017): “Aadhaar and Food Security in Rana, Kumar (2018): “Why the Lies and Violence in
References Jharkhand: Pain without Gain?” Economic & Mamata’s Bengal?” National Herald, 28 April,
Aggarwal, Ankita (2011): “The PDS in Rural Orissa: Political Weekly, Vol 52, No 50, pp 50–60. https://www.nationalheraldindia.com/india/a-
Against the Grain?” Economic & Political Weekly, Dutta, Siraj (2018): “Aadhaar-Enabled Starvation in hunt-lies-and-violence-in-mamatas-bengal.
Vol 46, No 36, pp 21–23. Narendra Modi’s ‘New India,’” Wire, 18 January, Satpathy, S K (2017): “Implementation of National
Alderman, H, U Gentilini and R Yemtsov (eds) https://thewire.in/government/aadhaar-ena- Food Security Act (NFSA) in Odisha,” Odisha
(2018): The 1.5 Billion People Question: Food, bled-starvation-narendra-modis-new-india. Review, February–March, http://magazines.
Vouchers, or Cash Transfers?, World Bank, Garg, Samir (2013): “Twin Strategies for Food Secu- odisha.gov.in/Orissareview/2017/Feb-March/
Washington, DC. rity and Productive Inclusion: PDS Reforms in engpdf/43-48.pdf.

Economic & Political Weekly EPW FEBRUARY 9, 2019 vol lIV no 6 47