Bengaluru: In 2021, three in four people in India could not afford a healthy diet. In South Asia, the pandemic disrupted work and disproportionately impacted poorer families.

Following the pandemic in 2020, the Indira Rasoi Yojana [now called Shree Annapurna Rasoi Yojana] was started by the Congress-led government in Rajasthan to provide subsidised nutritious meals for the urban poor. This was similar to earlier initiatives in Tamil Nadu, Odisha and Karnataka.

Reetika Khera, economist and faculty at the Indian Institute of Technology Delhi, said that “canteens have fostered the creation of democratic spaces” in societies divided along caste and class lines. One of the biggest shortcomings of the National Food Security Act (NFSA), 2013 is that it did not include a provision for community kitchens or canteens which provide heavily subsidised hot, cooked food, especially in urban areas.

Immediately after the pandemic-induced lockdown, Khera (along with Jean Drèze and Meghana Mungikar) had highlighted the exclusion of 100 million people from the government’s subsidised food grain programme which has more than 800 million beneficiaries. It is exacerbated by the indefinite delay in the population Census 2021, she said.

As India votes for the 18th Lok Sabha, we speak with Khera on various social security schemes including maternity entitlements, healthcare, food-related welfare like community kitchens, which are important in the context of the decline in real wages.

Edited excerpts:

The government’s One Nation One Ration Card (ONORC) scheme, which was expected to improve foodgrain access to migrants, is mostly limited to Delhi NCR. You have talked about community kitchens, their potential in providing nutritious food, and how they alleviate women’s workload in households. How can it support migrant households at a time when there is concern about real wages? Could you share more details based on your research, and the impact it can have in urban areas?

One of the biggest shortcomings of the NFSA as passed in 2013 is that it did not include a provision for ‘community kitchens’ or canteens. Many states have been experimenting with these over time. Canteens provide heavily subsidised hot cooked food especially in urban areas.

Amma’s canteens in Tamil Nadu were the first “on scale” initiative and were set up in 2013 (there have been such initiatives even before that). After this, Indira canteens were set up in Karnataka. Most recently, and perhaps overtaking these two states in terms of scale, was Rajasthan’s Indira Rasoi scheme that began in 2020 in response to the humanitarian crisis created by the Covid-19 lockdown. In November 2023, there were more than 1,100 Indira Rasois in Rajasthan. These canteens provide breakfast/lunch and dinner at Rs 3 to Rs 8 per plate.

In our recent survey in six cities of Rajasthan and three each in Karnataka and Tamil Nadu, we were struck by the wide range of guests at the canteens.

Canteens have fostered the creation of democratic spaces in deeply divided (on caste and class lines) societies. One of the most remarkable was a canteen near a hospital in Jaipur where we saw patients, doctors, lab technicians and cleaning staff among others, eating in the same place. It was quite a moving sight. For some elderly and single persons, canteens have become a welcome daily outing.

On a Sunday in Chennai, I was touched when several men who were having breakfast at the Amma’s unavagams [canteen] said that they were eating there even though they have kitchens at home because they didn’t want to trouble their working wives for breakfast on her weekly holiday. In a few canteens, we also met housewives at lunch time. Canteens are creating opportunities for paid work for women. In Tamil Nadu, the entire team is only women. (Though Odisha’s canteen scheme Aahaar is much smaller in scale, there too, all canteens are run by women’s self-help groups, Mission Shakti.)

For many migrant workers, with or without cooking facilities, there is little ONORC can do that canteens cannot. Many working professionals, mostly men, appreciated the canteens because as migrants living alone in the city, their accommodation did not always have cooking facilities, and even when they do, they often don’t have time.

Across states, many regular guests of canteens across the three states appreciated that the food was like home-cooked food, something they could eat on a daily basis. Monetary savings add up to quite a lot. Most respondents reported incomes between Rs 10,000 and Rs 18,000 per month. A conservative estimate suggests that for those relying on canteens for even one meal per day regularly, monthly savings would be 5% to 10% of their salaries.

Experts have pointed out that the new household consumption survey expenditure is not comparable with the previous ones due to changes in the methodology. But considering the reported reduction in real wages, households would ideally be spending a larger proportion on food. How do you explain the reduction in percentage of expenditure on food to 46.4% per month compared to more than half in rural areas in the previous surveys?

The share of food expenditure in total expenditure is the ratio of absolute expenditure on food (numerator) divided by total expenditure (denominator). For the ratio to decline, either the denominator has to increase or the numerator has to decrease. Or both.

However, absolute food expenditure is the product of food quantity consumed and their prices. A decrease in food quantities or prices can lead to a reduction in absolute food expenditure. (For example, if prices rise and people respond by reducing their intake, absolute food expenditure can decline.) If this is the case, then a decline in the share of food expenditure in total is not necessarily a good thing.

I suspect that when people see that food expenditure share has declined, they are assuming that prices have fallen, but they are not accounting for the possibility that a decline in quantity may be behind the observed decline of food share in total expenditure. It would have been far more informative to have information on quantities and prices.

I am unable to understand why the government released the NSS (National Sample Survey) consumption expenditure data selectively. Why not share the raw/granular data using which these summaries were prepared?

Three in four Indians could not afford a healthy diet in 2021. In March 2019, the Supreme Court said that nearly 80 million unorganised workers and migrants who did not have ration cards must be provided one. You, along with Drèze and Mungikar, had calculated the exclusion of 100 million people due to the outdated Census 2011 on which the government relies for the PDS. The government’s Pradhan Mantri Garib Kalyan Anna Yojana now provides priority households with only 5 kg of foodgrains under the PDS, but at no cost [a change from the Rs 2 or Rs 3 per kg for rice and wheat]. What are the present challenges in terms of food security and allocation for households. Does the PDS need to be revamped to include more nutritious food options?

For years, even during the discussions on the NFSA, we have been raising the issue of lack of diversity of food items provided through the PDS. The inclusion of dals and oil is important, as well as diversification out of wheat and rice into millets such as ragi, bajra, etc. In some states, there has been progress on both fronts; for example, Karnataka and Odisha have been giving some ragi (it is not clear how regularly) through the PDS. In Tamil Nadu and Himachal Pradesh, dals and oil have been provided for years, but we have not seen any move from the Union government to learn from these state-level initiatives. (Interestingly, the Congress party has promised to add dal and oil to the PDS in its 2024 election manifesto.)

Besides these older issues, we now have the problem of the exclusion of over 100 million people from the PDS because a fresh census in 2021 did not happen. Both these aspects--diversity of food basket provided through the PDS and expansion of coverage to account for the population increase since 2011--need to be remedied urgently.

Your survey and research on health centres in Bihar, Chhattisgarh, Himachal Pradesh, Jharkhand and Rajasthan show that the pace of improvement is slow. Your analysis said, “Both NRHM and Ayushman Bharat can be seen as a useful but very limited demonstration of the possibility of improving PHCs.” What are the significant gaps in primary healthcare infrastructure, and what investment must be made to improve primary healthcare access in less developed states?

The usage of public health facilities is much higher than what is popularly believed, and it has been rising in most states. According to NFHS, the proportion of households that report using a public health facility when they are sick has risen from 34% (in 2005-06) to 45% (2015-16) and to 50% in the latest round (2019-21). States like Himachal Pradesh have remained above 80% throughout this period. The most remarkable increase is recorded in Chhattisgarh where it has doubled (from 36% in 2005-06 to 69% in 2019-21).

However, the health services that they get are uncertain (the nurse or doctor, the medicines and tests they need may or may not be available) and remain quite basic when available. Though this did not reflect in the numbers in our survey, my sense was that the appointment of more nurses at the sub-centre has been helpful in bringing very basic health services to people (e.g., for small wounds, cold-cough and fevers, diarrhoea and headaches).

Unfortunately, this [Union] government tends to get too caught up in branding/ rebranding exercises, and that wastes resources everywhere. For instance, sub-centres that were “upgraded” to “Health and Wellness Centres” after 2019 are now to be rebranded as “Ayushman Aarogya Mandirs”. A recent report suggested that funds for Kerala were delayed because the state refused to comply (for linguistic reasons) with rebranding norms issued by the Union government.

You and Drèze analysed the government's maternity entitlements scheme, Pradhan Mantri Matru Vandana Yojana (PMMVY) and found that it has “gone into reverse gear”. You also mention challenges in accessing data for the scheme. Is the PMMVY particularly concerning in some states? How adequately has the government allocated resources for welfare schemes like PMMVY?

An overarching issue with cash-based entitlements is that they lose value over time because governments do not have a mechanism in place for indexation or ensuring an increase in real terms. The central contribution to some social security pensions has not increased since 2006; it is Rs 200 per month even today. (Again, the Congress manifesto promises to increase the Union government’s contribution to Rs 1000; the BJP manifesto is silent on this.)

As per the NFSA, maternity entitlements were supposed to be Rs 6,000 per pregnancy. When the scheme was operationalised in 2017, using warped logic, the government reduced it to Rs 5,000 and limited it to the first child only. (That restriction has been partially relaxed recently.)

Today, more than 10 years later, the amount remains stuck at Rs 5,000 and Rs 6,000. [In] some states with a better track record on social policy issues, things are somewhat better--Tamil Nadu has Dr Muthulakshmi Reddy scheme for maternity entitlements that provides Rs 18,000 per child. Even in such states, the problem of stagnation of cash benefits remains. In 2021, the manifesto of the Dravida Munnetra Kazhagam promised to increase maternity entitlements to Rs 24,000, but as far as I am aware, this has not been done.

Coming back to PMMVY, the central scheme, it has remained underfunded from day 1. The extent of underfunding is depicted in the accompanying chart – the horizontal line on top is an estimate of the budget that would be required for full coverage and the line at the bottom shows you the actual budget.

If we use the ‘wage compensation’ principle to fix maternity entitlements (as is done for women in the formal sector), then at Rs 400 per day for six months, the PMMVY amount would be Rs 72,000 per pregnancy.

When an entitlement is supposed to be universal but adequate funds are not made available, the government uses (hidden) rationing mechanisms to manage the situation: Aadhaar and the Aadhaar-enabled payments system serve this purpose.

Aadhaar is compulsory for PMMVY. On top of that, women are told to correct/update their demographic details on their Aadhaar cards: Their Aadhaar should have their in-laws’ address (not of their natal village); or “w/o” (wife of) rather than “d/o” (daughter of) on their Aadhaar card. Maternity entitlements are supposed to make their lives easier during pregnancy; they end up giving women a runaround for a token amount that is losing value over time.

The combined result of under-funding and this rationing system is that in 2021-22, less than half of all eligible women actually got their maternity benefits.

At 93 million, the total number of people who demanded MGNREGA work in 2023 was similar to 2019, before the pandemic. The household demand was 6% higher in 2023. Further, in 2023-24, at 59%, women persondays generated in MGNREGA was the highest in a decade. What does the reduced demand for MGNREGA, and more women persondays, indicate in terms of employment and the rural economy?

I do not believe that the “employment demanded” figures on the MGNREGA website are reliable. When the Act was passed, awareness among workers about their right to demand work was poor. The norm was for Gram Panchayats and other implementing agencies to open worksites when budgets were available. In small pockets where there were local workers unions, work was being demanded but there were difficulties in getting either a dated receipt or getting work within 15 days, the time-frame stipulated by the Act.

Under pressure from such groups, the employment demanded feature was introduced in NREGASoft, the MIS for MGNREGA. It became compulsory to first enter demand everywhere--even in areas where work was being provided proactively without anyone demanding it. The right to demand work under MGNREGA became a ‘responsibility’.

MGNREGA functionaries reacted to the new ‘work demand’ module in the MIS by entering the work demanded once work had begun in their village. This protected them from having to pay an unemployment allowance (that kicks in if work is not provided within 15 days of demand). Such retrospective data entry practices render this module of the MGNREGA MIS unreliable.

If demand for MGNREGA work is depressed, the most plausible explanation is that MGNREGA wages are not only below market rates but also below wages for agricultural workers (generally the least paid category of workers) in most states. Recently in rural Barmer, I learnt that for harvesting operations on private land, the wage rate for agricultural labour was Rs 500 per day and on MGNREGA, de facto they get only Rs 250 per day.

If you add the hassles and delays MGNREGA workers face to get paid once they have worked, you can see why they may not want to rely on it.

For women, the story is a bit different: MGNREGA has always been an attractive option for them, because it allows them to combine their household responsibilities with some paid work. It is rare for them to find other forms of paid work outside the house that would allow them to do so. In any case, opportunities for paid work outside the house are not easy to come by.

I believe that if MGNREGA is implemented in spirit--that means, if work is provided close to home, timely payments are assured, etc.--the demand for MGNREGA work and budgets would be much higher than we have seen.

We welcome feedback. Please write to We reserve the right to edit responses for language and grammar.