Reducing the true cost of food-based safety nets: evidence from India’s subsidized food program

Public procurement of food plays a pivotal role in determining the production and consumption of various food items. This is particularly true for staple grains in countries such as India, where the government procures over 40% of rice and wheat. This grain is redistributed to approximately 60% of India’s population through the public distribution system (PDS), the national food-based safety net. Although the PDS plays a critical role in ensuring food security for millions of beneficiaries, there is scope for reducing the hidden costs associated with the production of PDS food grain. As costs such as unsustainable use of groundwater and subsidies for power and fertilizers are not accounted for in the ‘financial’ cost of the program, we use true cost accounting to identify, quantify, and monetize these hidden costs associated with the PDS. In 2021–22, the Indian government and PDS beneficiaries spent US$16.5 billion and US$0.9 billion, respectively, on the PDS. However, accounting for hidden costs associated with the production of PDS food grain reveals that the true cost of the PDS is more than twice its financial cost. Using true cost estimates of distributing rice, wheat, and millets through the PDS, we find that replacing one kg of rice with one kg of millets for a quarter of PDS beneficiaries every month has the potential to reduce the true cost of the PDS by US$1.37 billion annually.


Introduction
Public policies play an important role in determining what is produced and consumed across countries.
Two important examples of such policies are public procurement and public distribution of food items by governments through food-based safety nets.As governments have a direct role in purchasing and distributing food items through food-based safety nets, the design of these programs can play a pivotal role in making food systems more sustainable.
Though a large number of studies have focused on the successes and failures of food-based safety net programs (see, for example, Alderman et al 2018), few have examined the negative impacts associated with the public procurement and distribution of food.We aim to address this gap in the literature by examining the hidden costs associated with food-based safety nets and propose ways to make these programs more sustainable.Hidden costs are costs associated with the production and consumption of food that are not accounted for in its price.Our study focuses on the public distribution system (PDS), India's national food-based safety net, which provides highly subsidized food grain (a combination of rice and wheat) to over 800 million individuals every month.
Previous studies have called attention to the unintended negative impacts of the production of food grains on India's food system.They have highlighted the high dependency of cereal consumption on production in states with stressed groundwater reserves (Kayatz et al 2019, Harris et al 2020, Rajan et al 2020) and rising greenhouse gas (GHG) emissions from agriculture (Rao et al 2019, Nayak et al 2023).Some studies have proposed solutions such as reducing groundwater use by removing power subsidies for agriculture (Bhattarai et al 2021); providing incentives for water conservation (Shah 2009, Fishman et al 2015); replacing rice production with alternate grains (Davis et al 2018(Davis et al , 2019)); and adopting sustainable agricultural practices to mitigate GHG emissions (Sapkota et al 2019).
With growing concerns about the long-term sustainability of the food system, there is an urgent need to address the negative impacts resulting from the production and consumption of food.One way to address these impacts is by estimating the true cost of food by identifying, quantifying, and monetizing the hidden costs associated with the food system.True cost accounting (TCA) is emerging as a potential methodology to fill this gap (de Adelhart Toorop et al 2021, FAO 2023, Hendriks et al 2023).
TCA identifies various costs associated with the production and consumption of food for four focus areas (examples in parentheses): economy (subsidies to the agricultural sector), environment (GHG emissions from rice production contributing to global warming), health (respiratory illnesses due to air pollution from crop residue burning), and social (child labor).These costs, which are not accounted for in the price of food items, are quantified and monetized using available data.For costs that cannot be quantified, such as the suffering of animals, the true cost can be assessed qualitatively.
Recent TCA studies have estimated the true cost of food worldwide (FOLU 2019, FAO 2023), in the United States (Rockefeller Foundation 2021), and the United Kingdom (Fitzpatrick et al 2019).Some TCA studies have focused on specific crops, such as sustainable rice production in Thailand (Khon Kaen University 2022), while others have looked at foodbased safety nets, such as school meals in the United States (Rockefeller Foundation 2021a).
Our study contributes to two streams of literature (negative impacts of food grain production and the use of TCA to improve the sustainability of food systems) by using TCA to estimate the true cost of producing and distributing rice, wheat, and millets through the PDS in India and proposing solutions to making this food-based safety net more sustainable.Our study builds on previous work that recommended gradually replacing rice and wheat production in India with more climate-resilient crops and reducing unsustainable use of groundwater (Fishman et al 2015, Davis et al 2018, Ashok et al 2021, Chakraborti et al 2023).We do this by proposing a solution that accounts for the political economy associated with price guarantees for farmers in India, which limits the government's ability to undertake major reforms to reduce staple grain production (Pinstrup-Andersen 2014, Pingali 2015).
The rest of the paper is divided into four sections.The following section provides the context for our study by discussing the unintended impacts of the PDS.The third section describes the methods and data.Results are presented in the fourth section, and we conclude in the fifth section.

Unintended impacts of the PDS
The PDS provides five kg of food grain per person to most PDS households every month.Each PDS beneficiary receives a combination of rice and wheat, depending on their state of residence.Rice and wheat are sold at Rs. 3 and Rs. 2, respectively.The corresponding retail prices for these commodities were approximately Rs. 35.4 andRs. 28.1, respectively, in 2021-22 (Government of India 2022a).To learn more about the PDS, see Bhattacharya et al (2017), Puri (2022) or Rahman and Pingali (2024).
Food grain distributed through the PDS is procured from farmers at guaranteed prices, also known as minimum support prices (MSPs).Figure 1 shows the production and procurement of PDS food grain between 2012 and 2022.In 2021-22, the government procured 44% and 41% of total rice and wheat production, respectively.Although the PDS plays a pivotal role in ensuring food security for India's growing population, it has resulted in certain unintended negative impacts (Pingali 2012, Pingali et al 2019, Davis et al 2022).
Production of rice and wheat in India is heavily concentrated in certain states (see figures S1.1-S1.3 for state shares of the production, procurement, and consumption of PDS food grain in .Relative to their size (area and population), two states in the northwest (Punjab and Haryana) account for a disproportionately higher share of production and procurement of rice and wheat for the PDS.In contrast, Uttar Pradesh, the state with the highest production and consumption of PDS food grain, is responsible for a relatively smaller share of procurement.
The imbalance in the production and procurement of food grain for the PDS has its roots in the green revolution in the early 1960s.States with better agro-climatic endowments, irrigation facilities, and road networks (such as Punjab and Haryana) benefited immensely from the Green Revolution.In contrast, states such as Bihar and Odisha were largely left out (Pingali 2012, Pingali et al 2019).Although the increased production of rice and wheat benefited farmers in Punjab and Haryana, it also resulted in negative environmental impacts, such as a large decline in groundwater levels and soil degradation from the overuse of fertilizers and pesticides (Bhattarai et al 2021).
To address this imbalance, the government introduced the decentralized procurement (DCP) program in 1997-98 to promote local (within state) Figure 1.Production and procurement of PDS food grain .This data was obtained from the Unified Portal for Agriculture Statistics (UPAg), Government of India.
procurement of food grain.While the program has increased procurement of PDS food grain from states such as Chhattisgarh and Madhya Pradesh, five states continue to account for 71% of total PDS procurement.
In addition to an imbalance in the procurement and consumption of PDS food grain, the PDS has also been criticized for promoting the production and consumption of rice and wheat over more nutritious alternatives such as millets (Eliazer Nelson et al 2019).Over the past decade, the production and procurement of rice and wheat have increased significantly, while that of the three major millets (Bajra or pearl millet, Ragi or finger millet, and Jowar or sorghum) have remained stagnant (see figure 1).
Adding millets to the PDS food basket has many potential health and environmental benefits, providing better nutritional value than rice and wheat.Pearl millet, finger millet, and sorghum have approximately 6, 5 and 4 times, respectively, more iron content than rice (Longvah et al 2017).Millets have also been described as 'climate-smart' crops as they consume less inputs such as water (Davis et al 2019) and can help reduce the true cost of the PDS by reducing its water and carbon footprints.Given that millets primarily grow in areas with limited water availability, such as in western India (see figures S2.1-2.3),including them in the PDS has the potential to increase the local procurement of food grain and improve livelihoods of farmers in areas that do not benefit from price guarantees provided through public procurement.

Methods and data
3.1.TCA TCA has emerged as a holistic tool to estimate hidden costs associated with food systems.TCA quantifies costs from farm to plate that are not reflected in prices paid upfront by consumers.Multiple frameworks exist that estimate the true cost of food, such as the TEEBAgri framework (Obst and Sharma 2018), the Rockefeller foundation framework (Rockefeller Foundation 2021), and the FAO two-phase agrifood systems assessment process (FAO 2023).Broadly, they focus on identifying, quantifying, and monetizing the economic, environmental, health, and social costs of the food system.
In this paper, we use the framework developed by the Rockefeller Foundation (hereafter referred to as the 'RF Framework') to estimate the true cost of India's PDS.We use this framework because of its straightforward approach to estimating the true cost of food.The RF framework is also easy to adapt to different contexts (such as food-based safety nets in India) because of its publicly available model and technical appendix (Rockefeller Foundation 2021b).
Applying the RF framework involves three steps.First, we select the main focus areas.Second, we identify the important impacts or 'metrics' for each focus area.Third, we quantify and monetize each of the metrics to estimate the true cost.Metrics are selected based on the ease of quantification, size of primary impact, and feasibility of intervention to modify impact in the future (Rockefeller Foundation 2021b).

Quantifying and monetizing metrics
In this study, we examine two focus areas: the economy and the environment.We begin by estimating the 'financial cost' of the PDS.This includes the cost borne by the government to procure and distribute PDS food grain and the price paid by PDS beneficiaries to purchase their PDS entitlements.For the economy-related impacts of the PDS, we estimate the cost of fertilizer and power subsidies.Although these costs are included in the government's budget, they are not attributed to the PDS.For environmentrelated impacts of the PDS, we estimate the cost of scarce water use and GHG emissions.Table 1 lists the metrics and monetization factors used in this study.One of the major challenges of applying TCA is monetizing metrics.We use actual costs, where possible, to monetize the hidden costs associated with the PDS.These numbers are available in government documents such as the Ministry of Finance's annual budget and the Ministry of Food and Civil Supplies' food bulletins.For environment-related metrics, we use India-specific monetization factors that were estimated by the Impact Institute for the year 2023 (True Price 2021).See S7 and S8 for more details on the monetization factors used for these metrics.

Financial cost of the PDS
The financial cost of the PDS includes the cost borne by PDS consumers and the government.PDS consumers pay a highly subsidized price for food grain purchased from the PDS.In some states, state governments further subsidize this cost by providing a state-level subsidy.In 2023, the Government of India announced that all PDS beneficiaries would receive free PDS food grain until 2028, essentially converting the cost borne by consumers to a cost for the government (PIB 2023).The cost incurred by the central government includes costs associated with procurement and distribution of PDS food grain (table S4.1).

Economy-related metrics
In addition to the direct subsidies provided by the government through the PDS, it also provides input subsidies that further reduce the price of growing rice and wheat.These include subsidies for inputs such as fertilizers (table S5.1) and power (table S6.1).
One of the major challenges of estimating costs associated with the production of food grain is attributing agriculture-related costs to specific crops.The methodology and assumptions used to estimate cropwise subsidies for fertilizer and power are described in sections S5 and S6 of the supplementary materials, respectively.

Environment-related metrics
We use blue (i.e.irrigation) water footprint estimates from a study by Kayatz et al (2019) to calculate the scarce water use for our study.Kayatz et al (2019) used the cool farm tool water (CFTW) to model state-wise crop-specific blue water footprints for rice, wheat, and millets.In order to estimate the share of blue water use that is 'unsustainable' or 'scarce' , we multiply the blue water footprint with a 'scarcity factor' that ranges from 0 to 1 (Galgani et al 2021).We use water depletion categories from the World Resources Institute's Aqueduct 4.0 database (Kuzma et al 2023) as the scarcity factor.
The Aqueduct 4.0 database estimates water depletion as the ratio of total net water consumption to available blue water supplies.Water depletion risk is categorized into five categories: low (<5%), lowmedium (5%-25%), medium-high (25%-50%), high (50%-75%), and extremely high (>75%).We use the lower end of this range as the scarcity factor for each state.For a more detailed description of our estimation of this metric, see section S7 of the supplementary materials.
We use state-wise estimates of GHG emissions from a study by Nayak et al (2023), which uses the partial life cycle analysis method to estimate GHG (CH 4 , CO 2 , and N 2 O) emissions associated with different agricultural inputs for rice and wheat.As this study does not estimate GHG emissions for millets, we use estimates from Rao et al (2019).Section S8 of the supplementary materials provides a detailed description of the methodology used to estimate this metric.While we do include GHG emissions from the production of these food grains, our estimates do not account for the emissions resulting from the procurement, transportation, and distribution of PDS food grains.
Our estimates only account for the current costs associated with scarce water use and GHG emissions related to PDS food grain.We do not account for past inaction or future deficits resulting from these metrics.

True cost of the current PDS food basket
We estimate that in 2021-22, the total financial cost of the PDS was US$17.4 billion, and the true cost was US$45.3 billion.As shown in figure 2, the largest hidden costs are associated with scarce water use, followed by GHG emissions and fertilizer subsidies.
Although there are no previous studies that provide estimates of the hidden costs of the PDS or similar programs in other countries, estimates for the hidden costs of India's agrifood system as a whole show that health, environment, and social costs are responsible for 62.5, 24.5 and 12.9%, respectively (FAO 2023, Lord 2023).The high share of health costs is primarily due to obesity and noncommunicable diseases resulting from food consumption (Lord 2023).We do not expect similar health-related impacts in the PDS as it provides 5 kg of food grain per person per month, which is 2 kg below the monthly recommended calorie intake from cereals in the EAT-Lancet diet (Gupta et al 2021).
A large part of our estimate of the hidden costs of the PDS is attributable to environment-related costs.The cost of scarce water use is primarily due to the large blue water footprint of wheat production as it is cultivated in the dry season (winter) and the common practice of flooding paddy fields to avoid the growth of weeds during rice cultivation.We validate these findings, which are based on blue water use estimates from Kayatz et al (2019), with state-wise blue water use estimates from Nayak et al (2023), and find the results to be similar.
Costs associated with GHG emissions are attributable to high methane (CH 4 ) emissions from the flooding of paddy fields during rice cultivation and the resulting anaerobic decomposition of organic material (Rajendran et al 2023).Using estimates from Nayak et al (2023), we find that GHG emissions from rice and wheat are 0.87 and 0.36 kg CO 2 eq.per kg, respectively.Our estimate for GHG emissions from wheat is similar to the results from Vetter et al (2017), but the estimate for rice is five times lower.
It is important to note that there is uncertainty associated with these cost estimates.Table S7.2 in the supplementary materials presents state-wise confidence intervals of our scarce water use cost estimates.The 95% confidence interval for the national scarce water use cost estimate ranges from US$14.9 billion to US$22 billion.We are unable to present confidence intervals for the cost estimates for GHG emissions as Nayak et al (2023) do not present standard errors.

True cost of the PDS by state
Figure 3 highlights the differences in per capita true costs of the current PDS food basket for each state based on the origin of the cost.Though the costs of state-level GHG emissions are felt beyond their borders, the costs associated with scarce water use and power subsidies are borne by each state's residents.
As expected, states with the highest share of procurement have the highest per capita true costs of the PDS.However, it is important to note the disproportionately large share of per capita hidden costs associated with the production of PDS food grain in Punjab and Haryana.Studies have documented a large decline in groundwater levels in Punjab over the past decade (Harris et al 2020) and the resulting decline in cropped area (Bhattarai et al 2021), attributing the decline to the intensive cultivation of rice and wheat in these states.
Although procurement from Punjab and Haryana benefits farmers in the states, the high per capita true costs associated with the environment and economy adversely impact all residents of the states.As state governments fund power subsidies, the large subsidies for farmers in Punjab and Haryana are borne by the state exchequer.Similarly, overuse of groundwater leads to 'virtual water flows' from Punjab and Haryana to other states (Harris et al 2020).

Reducing the true cost of the PDS
There are several options available to reduce the true cost of the PDS: (1) adopting practices that reduce production-related costs, such as alternate wetting and drying, system of rice intensification, incentives for using better irrigation technology (Fishman et al 2015), and taxes or quotas to discourage the overuse of scarce water (Singh et al 2023); (2) promoting DCP by developing infrastructure (such as procurement centers) to purchase rice and wheat from farmers in states with high production but low procurement of food grains; and (3) replacing part of the current allocation of PDS rice and wheat with alternative grains, such as millets.
We focus on the third option as it is the most politically and administratively feasible option due to existing legal provisions in the National Food Security  .Table 1 provides sources of impact quantities and monetization factors for each metric.See section S3 of the supplementary materials for assumptions used to estimate the true costs.See sections S4-S8 of the supplementary materials for the methodology to estimate costs associated with government, fertilizer subsidy, power subsidy, scarce water use, and GHG emissions, respectively.Act (NFSA) of 2013 for the inclusion of millets in the PDS.This option also fits well with the government's push to promote the production and consumption of millets in India.We estimate the true cost of distributing a kilogram of rice, wheat, and millets through the PDS (see figure 4).In addition to the metrics and monetization factors used in table 1, we indicate the MSP for each food grain.This distinction helps highlight the large differences in the procurement and distribution costs among the three food grains.Rice has the highest procurement and distribution costs, as it loses 30% of its quantity during the milling process.
Despite having lower MSPs than millets, the true cost of rice and wheat is 2 and 1.8 times, respectively, that of millets.The MSP of millets is a weighted average of the MSP of the three major millets, which range from Rs. 22.5 for pearl millet to Rs. 33.8 for finger millet.The main difference in true cost arises from the large differences in scarce water use.Though millets have significantly lower scarce water use, GHG emissions, and subsidies per hectare, we use per kg estimates to understand better the potential for replacing one with the other.The high costs of fertilizer subsidies per kg of millets are due to the lower yield of millets relative to rice and wheat.

Adding millets to the PDS food basket
Although the NFSA has had a provision for the distribution of millets through the PDS since 2013, few states have included millets in their PDS (see figure S11.1 for millet distribution by state for 2021-22).Only Karnataka provided millets at scale in the PDS in 2021-22.Karnataka was responsible for procuring and distributing more than 60% of millets in the PDS.Other states that provided millets include Madhya Pradesh, Maharashtra, and Haryana.
We use the true cost estimates of providing rice, wheat, and millets in the PDS to calculate the savings from including millets in the PDS.We perform calculations based on a scenario in which 1 kg of rice is replaced by 1 kg of millets for a quarter of PDS beneficiaries (∼200 million beneficiaries) every month.We limit the scenario to a quarter of the population to account for the availability of millets for procurement and a strong taste preference for rice and wheat in many states (Makkar et al 2019, Nuthalapati et al 2023).We assume that millet-producing states are more likely to introduce them into their PDS.This scenario would require an additional procurement of 2.6 million tons of millets (the official target for millet procurement in 2023-24 was 2 million tons).
We find that replacing 1 kg of rice with 1 kg of millets can potentially save US$1.37 billion (approximately 3% of the true cost of the PDS) yearly when we take true costs into account.Most of these savings result from reductions in scarce water use and GHG emissions.If the analysis is limited to the 'financial cost,' the savings are US$231.6 million (approximately 1.3% of the financial cost of the PDS).A major assumption in these estimations is that beneficiaries will replace 1 kg of rice with 1 kg of millets in their diet.
As the production of millets is concentrated in states that do not have a large share of rice and wheat  S9.1 shows the data used in this figure.See S4-S8 for the methodology to estimate costs associated with government, fertilizer subsidy, power subsidy, scarce water use, and GHG emissions.
production (see figures S1.1 and S2.1), the inclusion of millets in the PDS can promote local procurement and increase the incomes of farmers in states that have not benefitted from the public procurement system.
The major challenge to increasing the procurement of major millets is related to their low level of production.The main reason for this stagnancy in the production of major millets is the declining area under cultivation and the slow increase in yields over the past few decades.Figure S12.1 shows the changes in the area under cultivation and yields for rice, wheat, and millets over the past five decades.While yields have increased for all three grains, the area under cultivation has declined significantly for millets.Introducing millets in the PDS can help stop this decline.However, significant investments would be required to increase yields, develop infrastructure (procurement, storage, and processing), and promote millets to a population with a strong taste preference for rice and wheat.

Limitations and way forward
Although we have made every effort to ensure our findings are rigorous, they are subject to three important limitations.First, our estimate of the true cost of the PDS is conservative as our analysis is limited to two focus areas and four metrics.This is primarily due to the limited availability of data.As more data becomes available, future TCA studies of  S10.1 shows the data used in this figure .the PDS should include health-related costs, such as impacts from air and water pollution.For example, air pollution from burning crop residue has been linked to acute respiratory illnesses (Chakrabarti et al 2019), and water pollution from farm run-off has been associated with contamination drinking water (Sarkar et al 2021).Similarly, various other metrics, such as soil erosion due to crop cultivation and emissions from the PDS infrastructure (transportation and processing), can be included in environmentcosts.Second, our true cost estimates are sensitive to the choice of monetization factor.There is no consensus on monetizing metrics such as scarce water use and GHG emissions.We use monetization factors from the Impact Institute, which provides countryspecific costs for restoring scarce water and abating GHG emissions.As more monetization factors become available, these estimates can be updated.
Third, we are unable to assess the uncertainty associated with our cost estimates of GHG emissions.Our estimates for crop-wise and state-wise GHG emissions are from a secondary source that does not provide standard deviations.This limits our ability to calculate confidence intervals for our cost estimates of GHG emissions.We do provide confidence intervals for our cost estimates of scarce water use in table S7.2 in the supplementary materials.

Conclusion
Public procurement and distribution of food through food-based safety nets play an important role in determining the production and consumption of food, and the design of these programs can have a significant impact on the sustainability of national food systems.In this study, we highlighted the hidden costs associated with food-based safety nets by estimating the 'true' cost of the PDS, India's national food-based safety net program.We found that the program's true cost was more than twice its financial cost when hidden costs associated with economic subsidies (fertilizer and power) and environmental impacts (scarce water use and GHG emissions) were included.
We further examined the possibility of reducing the true cost of the PDS by replacing part of the current allocation of food grain with millets.Our estimates show that replacing kg of millet with 1 kg of rice for a quarter of PDS beneficiaries every month can reduce the true cost of the PDS by US$1.37 billion (or 3% of its true cost).
While previous studies recommended the promotion of millets as an alternative crop to rice and wheat in India (Davis et al 2018, Ashok et al 2021), our study lays out a clear pathway to promote the production and consumption of millets.Given the political economy of input subsidies that incentivize rice and wheat cultivation and the high costs of production of millets (Nuthalapati et al 2023), the likelihood of rice and wheat farmers shifting to other crops is limited.Thus, a gradual inclusion of millets in the PDS in western and central India would be more likely to reduce the emphasis on rice and wheat production in northwest India (Makkar et al 2019).
Our study highlights the urgent need for policymakers to consider costs that are not included in traditional accounting.While millets may seem expensive as a replacement for rice and wheat in the government's budget documents, the additional benefits from lowering scarce water use, GHG emissions, power subsidies, and fertilizer subsidies are immense.

Figure 2 .
Figure 2. True cost of the PDS.Table1provides sources of impact quantities and monetization factors for each metric.See section S3 of the supplementary materials for assumptions used to estimate the true costs.See sections S4-S8 of the supplementary materials for the methodology to estimate costs associated with government, fertilizer subsidy, power subsidy, scarce water use, and GHG emissions, respectively.

Figure 3 .
Figure 3. State-wise capita true costs of the PDS (2021-22).TableS9.1 shows the data used in this figure.See S4-S8 for the methodology to estimate costs associated with government, fertilizer subsidy, power subsidy, scarce water use, and GHG emissions.

Figure 4 .
Figure 4. True cost of procuring and distributing PDS food grains (US$/kg).TableS10.1 shows the data used in this figure.