Declining PDS leakages: A look at the numbers

Estimates have shown that leakages of foodgrain from the Public Distribution System declined from 42% in 2011-12 to 22-28% in 2022-23. Analysing new data from the Household Consumption and Expenditure Survey, Puri and Pingali show that leakages have further reduced to 8.8% in 2023-24. Raising concern regarding the accuracy of the latest estimates, they conclude that the correct estimate for 2023-24 is likely to be in the range of 10-20%.

With the much-anticipated release of data from the Household Consumption and Expenditure Survey (HCES), we have seen two estimates of foodgrain leakages from the Public Distribution System (PDS). Using the first round (2022-23) of the HCES, Khera (2024b) estimated PDS leakages at 22.1%. A second estimate of 28.1% was reported by Das et al. (2024). In this post, we present estimates of PDS leakages from the first and second (2023-24) rounds of the HCES. We find that PDS leakages declined from 24.1% in 2022-23 to 8.8% in 2023-24.

A large part of this decline in PDS leakages between 2022-23 and 2023-24 was due to the end of the distribution of free foodgrain, that had been instituted during the Covid-19 pandemic, at the end of 2022. Although we observe a decline in PDS leakages, our findings raise concerns regarding the reliability of recent estimates since five states report negative PDS leakages. This underestimation of PDS leakages is likely due to an overestimation of PDS consumption resulting from incorrect population projections or the overestimation of per capita PDS foodgrain consumption by the HCES.

Estimating PDS leakages

PDS leakages are measured as the difference between the foodgrain provided for distribution through the PDS (henceforth referred to as 'PDS offtake') and the total PDS foodgrain consumed by households (henceforth referred to as 'PDS consumption'). While PDS offtake data are publicly available, PDS consumption is estimated using nationally representative consumption data, such as the HCES.¹

Table 1 summarises estimates of PDS leakages over the past two decades. These estimates are based on two rounds of the Consumption and Expenditure Survey (CES) and two rounds of the HCES conducted by the National Sample Survey Office (NSSO). The reduction in PDS leakages between 2004-05 and 2011-12 has been credited to state-level reforms that expanded PDS coverage, lowered prices, and improved foodgrain delivery (Bhattacharya et al. 2017, Drèze and Khera 2015, Puri 2022). The decrease in PDS leakages after 2011-12 has been attributed to the implementation of the National Food Security Act (NFSA), which scaled up these state-level reforms across the country (Khera 2024a).

Table 1. Estimates of PDS leakages, 2004-2024

Source	2004-05	2011-12	2022-23	2023-24
Drèze and Khera (2015)	54	41.7
Bhattacharya et al. (2017)	58.6	43.1
Gulati and Saini (2015)		46.7
Khera (2024b)			22.1
Das et al. (2024)			28
Puri and Pingali (2025)			24.1	8.8

New estimates of PDS leakages for 2023-24

We use data from the first (2022-23) and second (2023-24) rounds of the HCES to estimate PDS leakages. PDS consumption is estimated by multiplying state-wise estimates of annual per capita consumption of PDS food grain (rice, wheat, and coarse grains) from the HCES with projected population numbers for each state in 2022 and 2023. We use population projections from a report by the National Commission on Population (Registrar General of India, 2020).

PDS offtake is calculated by adding state-wise NFSA PDS rice and wheat offtake numbers from the Department of Food and Public Distribution's Monthly Food Bulletins (MFBs) and additional allocations of foodgrain by states for non-NFSA PDS foodgrain distribution (or ‘state allocations’).² As the MFBs do not provide information on coarse grain offtake for 2022-23, we use coarse grain distribution numbers from the IMPDS website.³ In addition to the NFSA offtake and state allocations, we also include the offtake of free foodgrain distributed in response to the Covid-19 pandemic from the MFBs.

We find that PDS leakages declined from 41.7% in 2011-12 to 24.1% in 2022-23 and 8.8% in 2023-24. Table 2 presents our estimates of PDS leakages at the national and state levels for 2022-23 and 2023-24, along with the 2011-12 estimates from Drèze and Khera (2015).

Table 2. Estimates of PDS leakages

State	NSS 2011-12*	HCES 2022-23^	HCES 2023-24^
Andhra Pradesh	22	21.3	15.6
Assam	50.7	25.9	-5.2
Bihar	24.4	25.4	-0.4
Chhattisgarh	9.3	36.2	8.2
Goa	-	21.9	8.2
Gujarat	67.6	36.9	29.7
Haryana	49	20.2	3.8
Himachal Pradesh	27.1	34.8	20.9
Jharkhand	44.4	22.1	-7.4
Karnataka	34.7	17	1.9
Kerala	37.1	35.3	23
Madhya Pradesh	51.5	25.5	11.1
Maharashtra	48.2	33.8	19.7
Odisha	25	8.8	13.1
Punjab	58.8	-4.9	8.1
Rajasthan	60.9	9.1	-6.3
Tamil Nadu	11.9	27.4	18.9
Telangana	-	26.8	6.7
Uttar Pradesh	57.6	23.6	-2.4
Uttarakhand	34.9	35.3	22.5
West Bengal	65.3	15	14.6
Union Territories	-	35.1	25.5
Northeastern States (excluding Assam)	-	27.2	19.5
India	41.7	24.1	8.8

Source: Authors' calculations (see data file for details).

Notes: (i) * NSS 2011-12 estimates are from Drèze and Khera (2015). (ii) ^ The reference period for PDS offtake and consumption is August-July.

Between 2011-12 and 2022-23, most states experienced a significant decline in PDS leakages. Some states, such as Rajasthan and West Bengal, saw PDS leakages decline by more than 50 percentage points (pp). Others, such as Uttar Pradesh, Gujarat, Haryana, Madhya Pradesh, and Assam, saw declines ranging from 35 to 25 pp. These declines have been attributed to the implementation of NFSA, which increased coverage of the PDS, reduced prices, simplified identification criteria, and improved the delivery of foodgrains through reforms such as the digitisation of ration cards and the end-to-end computerisation of the PDS supply chain (Khera 2024a, Puri 2022). Our estimates for 2022-23 are largely similar to those estimated by Khera (2024b).

We find an additional decline in PDS leakages between 2022-23 and 2023-24 across most states, ranging from 10 to 25 pp. We attribute this to the ending of the distribution of free foodgrain as part of the government's Covid relief programme. It is concerning, however, that five low-income states have negative estimates for PDS leakages. We discuss three possible explanations for these peculiar results and their implications for recent estimates of PDS leakages.

Impact of Covid relief foodgrain distribution on PDS leakages

In response to the Covid-19 pandemic, the government provided 5 kg of free foodgrain per person under the Pradhan Mantri Garib Kalyan Anna Yojana (PMGKAY), in addition to regular NFSA PDS foodgrain, between April 2020 and December 2022. Unlike NFSA PDS foodgrain, which is distributed every month, the PMGKAY food grain was distributed as and when the programme was extended. The uncertainty and irregular nature of the distribution of Covid relief foodgrain could have resulted in higher leakages, similar to the high leakages in PDS foodgrain in the Above Poverty Line (APL) category during the pre-NFSA period (Drèze and Khera 2015, Khera 2024b).

Figure 1. PDS foodgrain consumption and offtake (August-July)

Figure 1 shows the total consumption and offtake of PDS foodgrain from August to July (the reference period of the HCES) for 2022-23 and 2023-24. While there is a large decline in the total offtake of PDS foodgrain (due to the ending of Covid relief under PMGKAY) between the two HCES rounds, the decline in total consumption of PDS foodgrain is relatively small.⁴ The large declines in PDS leakages between 2022-23 and 2023-24 are likely due to the ending of the distribution of food grain for Covid relief under PMGKAY. This is a notable finding, as data from the second round of the HCES shows that PDS leakages are lower than those estimated using the first round of the HCES.

Negative leakage numbers for five states

Although the estimates for PDS leakages are lower in the second round of the HCES, they are negative for five states: Assam, Bihar, Jharkhand, Rajasthan, and Uttar Pradesh. These negative numbers raise concerns as they bring into question the reliability of the data used to estimate PDS leakages in 2022-23 and 2023-24. For estimates of PDS leakages to be negative, we are either overestimating PDS consumption (we use two numbers to estimate PDS consumption: per capita PDS consumption estimates from the HCES and the population projections from the Registrar General of India (RGI)) or underestimating PDS offtake. We discuss three possible explanations for these negative figures. These include incorrect population projections (due to a delay in the 2021 Census), an overestimation of per capita PDS consumption by the HCES, and an underestimation of PDS offtake data.

1. Reliability of population projection: A major limitation of any estimates of PDS leakages is the unavailability of updated state-wise population numbers. Due to the delay in the 2021 Census, we had to rely on population projections by the RGI, released in 2020, to estimate PDS consumption. Although these projections use the latest data on fertility and mortality from the Sample Registration System (SRS), they rely on migration data from the 2011 Census. If these population projections overestimate the population of certain states, it could result in the overestimation of PDS consumption and an underestimation of PDS leakages.

In the past, RGI's population projections have underestimated India's population (Bhattacharya and Mishra 2024). However, there was considerable variation at the state level. Table 3 presents the projected and actual population numbers for the five states with negative PDS leakages in 2011. These projections were made in 2006 using the 2001 Census data. Except for Uttar Pradesh, the populations for all other states were underestimated.

Table 3. Comparison of RGI population projections and Census 2011

State	RGI Projections for 2011	Census 2011
Assam	30,695,000	31,205,576
Bihar	97,720,000	104,099,452
Jharkhand	30,568,000	32,966,238
Rajasthan	67,830,000	68,548,437
Uttar Pradesh	200,764,000	199,812,341
India	1,192,506,000	1,210,854,977

Source: RGI projections are from Table 8 of RGI (2006).

It is unclear whether the projections made in 2020, using 2011 Census data, overestimate or underestimate state-wise populations for 2022 and 2023. The increase in migration from low-income states to high-income states, combined with the impact of the pandemic, could have led to a decline in actual population numbers in some states relative to the projections made in 2020. If the 2020 projections overestimate the population of these states in 2022 and 2023, we would be overestimating PDS consumption and underestimating PDS leakages for 2022-23 and 2023-24.

2. Overestimation of PDS consumption in the HCES: Another explanation for the negative estimates of PDS leakages could be an overestimation of the per capita consumption of PDS food grain in the HCES. While the HCES has improved the quality of consumption data by introducing multiple visits to sample households (compared to the CES), concerns have been raised about the potential overestimation of consumption due to changes in the HCES sampling approach.

For instance, Anand (2024) highlights how the HCES' emphasis on selecting more villages closer to district headquarters, along with no specific efforts to ensure that the poorest are part of the sample, might lead to higher consumption estimates. Similarly, Mehrotra and Kumar (2024) suggest that there appears to be a higher representation of well-off groups, which could result in higher consumption estimates. Although this would lead to an overestimation of consumption in general, it is unclear how this would impact the consumption of PDS foodgrain. While a village closer to the district headquarters is more likely to have a better-functioning PDS, leading to higher PDS consumption, well-off groups are less likely to consume PDS foodgrain.

3. Underestimation of PDS offtake data: Finally, an underestimation of PDS offtake could result in negative leakage numbers. There are two ways this could happen: first, an underestimation of state allocations, and second, a mismatch in the PDS offtake and consumption numbers due to lags in the distribution of foodgrain.

We use state-level economic survey reports and state PDS websites to determine the total offtake of PDS foodgrain for states with state allocations.⁵ While we have made every effort to ensure all state allocations are accounted for, we could be missing numbers for some states, which could result in an underestimation of PDS leakages. This is unlikely to explain the negative leakage numbers for the five states, as only Jharkhand has a state-level allocation for its PDS (which is included in our calculations).

Our PDS leakage estimates in Table 2 use PDS offtake numbers from August to July, consistent with the reference period for the HCES. However, it is likely that if a state receives food grain in July, it distributes that food grain in August. To account for this difference, we estimate PDS leakages using a one-month lag for PDS offtake and report similar results in Table 4. This suggests that lags in PDS distribution do not meaningfully explain our finding of negative leakages for the five states. We also estimate PDS leakages using monthly offtake data from the Food Corporation of India (FCI) and find similar results.

Table 4. Alternative estimates of PDS leakages

Estimates	HCES 2022-23	HCES 2023-24
Offtake from MFBs (August-July)	24.1	8.8
Lagged Offtake from MFBs (July-June)	23.6	7.8
Offtake from FCI Reports (August-July)	23.5	8.0

Source: Authors' calculations (see data file for details).

Conclusion

Data from the first and second rounds of the HCES show noteworthy improvements in the functioning of the PDS since the implementation of the NFSA. Whether it is the significant increase in households reporting consumption of PDS foodgrain or the substantial decline in PDS leakages over the past decade, the PDS has improved considerably. However, it is essential to ensure that we have accurate estimates of PDS leakages to quantify the performance of the PDS and further enhance the functioning of this pivotal food-based safety net.

Our new estimates of PDS leakages from the second (2023-24) round of the HCES highlight two key points. First, the estimates from the first round (2022-23) of the HCES do not reflect the regular or ‘business as usual’ PDS, as part of the leakages are due to the distribution of PMGKAY foodgrain. The differences in the distribution of PMGKAY foodgrain versus regular PDS foodgrain could be due to the former's ad-hoc nature as a temporary relief measure.

Second, our findings raise concerns about the data used to estimate PDS leakages in 2022-23 and 2023-24, as five states have negative estimates in 2023-24. This is likely due to an overestimation of PDS consumption. It is unclear if this overestimation is due to the HCES consumption data or the projected population numbers (or both). In either case, PDS leakages are underestimated for 2022-23 and 2023-24. A combination of these factors suggests that PDS leakages are likely to be in the range of 10-20%.⁶ This is a substantial decline from the estimated PDS leakages of 42.1% in 2011-12.

The authors would like to thank Naveen Kottayil for providing research support for this study. The views expressed in this post are solely those of the author and do not necessarily reflect those of the I4I Editorial Board.

Notes:

PDS consumption is estimated by multiplying the annual per capita consumption of PDS foodgrain (rice, wheat, and coarse grains) by the total population.
There are three types of allocations for the PDS. First, all states receive foodgrain for NFSA PDS beneficiaries from the Food Corporation of India (FCI). Second, some states (such as Himachal Pradesh) receive a ‘tide-over’ allocation of foodgrain in case their NFSA allocation of PDS foodgrain was lower than their pre-NFSA allocation of PDS foodgrain. Third, some states provide PDS foodgrain to non-NFSA beneficiaries (such as Andhra Pradesh) or additional foodgrain to NFSA beneficiaries (such as Chhattisgarh and Mizoram). See Khera and Somanchi (2020) for more details.
The IMPDS distribution data underestimates coarse grain offtake. However, this is unlikely to significantly impact our results as coarse grains account for less than 2% of total PDS food grain distribution.
Ideally, we could have estimated PDS consumption separately for regular PDS food grain and PMGKAY/Covid foodgrain for 2022-23 and used that information to estimate PDS leakages separately for the two types of foodgrain. However, the HCES questionnaire does not make a clear distinction between these two types of foodgrain (rice, wheat, and coarse grains). For example, HCES respondents are asked about three types of rice consumption: PDS, free, and other. ‘Rice-Free’ is defined as rice ‘received free of cost from PDS shop/FPS or Point of Sale (POS) under PMGKAY or from any other state-specific schemes.’ Due to the definition used for free rice, if a state was already providing PDS rice at no cost (rather than Rs. 3 per kg – the central issue price), this rice is listed as ‘Rice-Free’ rather than ‘Rice-PDS.’
Some states report total allocation or distribution of food grain (rather than offtake). For those states, we use the allocation or distribution numbers as proxies for offtake.
A ten (five) percent reduction in the PDS consumption estimates results in the following estimates of PDS leakages: 31.5 (27.7) percent in 2022-23 and 17.9 (13.4) percent in 2023-24.