How much processed food do Indians eat?

“We discourage children from eating processed foods.  In fact, we don’t allow children to bring any processed foods as snacks.  No bread, no pizza, no birthday cake.  Only home-made Indian foods are allowed, like idlis, roti, etc.”  Announced a primary school principal recently.  This got me thinking: what is considered “processed” food?  If I buy dosa batter instead of making it home, does it make dosa “processed food”? Does making bread at home shift it from processed food category to home food?  Is processing a question of who does it and where (home vs. “production unit” or factory) or what ingredients are added to a food? And is all processing bad?

The question of food and food processing in India is something I have been interested in over the last several years.  One topic which I have been particularly curious about is how much processed or unprocessed food do Indians actually eat?  We like to believe that we eat very little processed food, but is this backed by data?  Have changing food habits shifted the norms significantly?

Last week, I came across academic papers which analysed household food expenditure data to show that only 10% of food consumed in urban India is highly processed (with added ingredients), while the majority 73% is minimally processed, and 17% is unprocessed.  For rural India, the comparable figures are: 6%, 79% and 15% (see chart below).  These figures are in terms of money spent, not quantity.

Food consumption processing 1508

Note one slightly tricky point about fluid milk: Fluid milk is treated as unprocessed in rural areas because it is an unpasteurized raw product, but as food with high-value first processing in urban areas.

The first question that arises from the above chart is why are rural Indians eating slightly less unprocessed food (excluding milk) than urban ones?  This could be because the data is based on expenditure, not quantity of food consumed.  Since prices of unprocessed foods are lower in rural areas where they are grown, it could account for the difference.  Another contributor could be that many high-value produce such as fruits and vegetables and milk are intended for urban markets.  Poor households (which constitute ~40% of the producers of the food we eat) often have dire need for cash, and therefore, sell almost all the produce, keeping very little for own consumption. Or this could simply be the margin of error in the data.  The authors didn’t comment on this point.

The second interesting thing to note is that almost 6% of food expenditure in rural households goes towards highly processed foods.  Though this is a significant amount, it is still quite small.

Trend over time

The authors compare expenditures with similar data from 1999-2000 and report some to-be-expected trends.  Both urban and rural people are spending more on foods with high-value first processing and second processing. The interesting data is again in unprocesssed foods: Why are rural households spending slightly more than they used to on unprocessed foods?  Could it be because rural Indias are increasing intake of higher value items such as vegetables? (see “What’s on the Indian Thali?“)

Food expenditure trend by level of processing

Food consumption processing table 1508

The study goes on to show (not surprisingly) a correlation between income and higher expenditure on processed foods.  It also shows that states with greater urbanization also show higher expenditure on processed foods.

Those interested in digging deeper, can check out the two papers in endnotes.


Morisset, M., and P. Kumar. (2008). “Structure and Performance of the Food Processing Industry in India.” Available at

Reardon, T., & Minten, B. (2011). The quiet revolution in India’s food supply chains (Vol. 1115). IFPRI Discussion Paper. Available at:

How many farmers does India really have? Or, what percent of the population does farming?

Yesterday, Hindustan Times published an article titled “How Many Farmers Does India Really Have?”.  This is a complex question involving definition of a farmer and the workforce of the country.  The article makes a valiant attempt to fairly present data from Census 2011 until the second last para, but ends with a wrong answer by oversimplifying the final calculation.

The article concludes:

So, if we add the number of cultivators and agricultural labourers, it would be around 263 million or 22% of the population (1.2 billion).  Then where does the common perception of 53% of population being involved in agriculture come from? It needs to be remembered that over 600 million Indians dependent on agriculture are not farmers. They are deployed in an array of related activities including fisheries.  And this confusion is widespread and innocent!

I will try to answer this question, since I pulled out half my hair trying to do so last year. But before we start, let me state that adding fisheries and other “allied” activities will not take 263 million to 600 million by any stretch of imagination.

Per occupation data from Census 2011, here are the all-India numbers which add up to 263 million engaged in agriculture either as cultivators or as casual labor.

Cultivators   82,706,724   35,985,916 118,692,640
Agri Labor   82,740,351   61,589,482 144,329,833
Total  165,447,075   97,575,398 263,022,473

But the real question the article asks is “where does the 53%” come from?  And this is where things start getting even more complicated.

What should we divide the 263 million by?  The total population of India (which includes home-makers, old people and young children), as the HT author does?

Or should we divide 263 million by only India’s workforce?  Or perhaps we should divided by all adults in the country (say, 15-59)?  The lower age bound is because Indian law guarantees Right to Education, and prohibits child labor until age 14. The upper bound is determined by the official retirement age in India. The table below summarizes some potential denominators.

Total workforce (all ages, including children)             331,865,930             149,877,381            481,743,311
Children in workforce (14 or less)    5,628,915    4,499,748    10,128,663
Total working age population (15-59, inclusive) 375,474,130 354,597,889    730,072,019
Total population 623,121,843 587,447,730 1,210,569,573

Given the sad state of affairs in our country, roughly 10 million people in our workforce of 482 million are actually children below 14 years of age.  Working children are included both in the agri-workforce number (263 million) and total workforce (482 million).  While Census 2011 data for number of children in total workforce has been released (10 million), the corresponding number for agri-workforce has yet to be released.


I think the real percentage we all want to know is one of the following three:

  1. Of the total workforce (any age), how many are employed in agriculture?
  2. Of the total number of working age population (aged 15-59), how many are employed in agriculture?
  3. or perhaps: Of the total # of households, how many depend on agriculture (let’s define this as, how many get at least half their income from agriculture)?

The first one, we can calculate from the two tables above. We find that 52% of the workforce (any age) is employed in agriculture.

For the second, we know the denominator (working age population) but not the working age population among the agri-workers.  That data is yet to be released.  But we can calculate a range: since we don’t know what portion of the working children are working in agriculture, we can assume that it is 0% or a 100% to get a range.  In fact this correction has to be made for the elderly who are in the workforce (those beyond the retirement age, who add up to 43 million).  So the range turns out to be 27-35%.

And finally, as a percentage of total population, the % employed in agriculture is 21% (in comparison, the percent employed in any occupation is about 40%).

% of people employed in agriculture MALE FEMALE TOTAL
As a % of total workforce 48% 63% 52%
As a % of working age population (aged 15-59, inclusive) 32-42% 22-27% 27-35%
As a % of total population 25% 16% 21%

Regardless of which number we want to look at, it is clear that agriculture is by far the biggest and most important occupation in the country, both for men and women.  It is the sector (much more than any other) that most of the population of the country depends upon for income and livelihood.

And, finally, the question about how many families (or people), “depend on agriculture” is an insanely complex question that I’m hoping some researcher answers once more Census data is released.

[Note: The occupation data is based on preliminary Census tables, which may get revised. And, thanks to my team member Garima for the initial analysis, which I have updated here.]

Why women are withdrawing from agri labour

A recent article from Livemint examines the reduction in availability of female agricultural labor.

India witnessed a massive exit of female agricultural labourers between 2004-05 and 2011-12, according to India’s official employment estimates published by the National Sample Survey Office (NSSO). Thirty three million agricultural workers, mostly women, quit the labour force between 2004-05 and 2011-12.

… Landless labourers, considered to be the bottom of the rural pyramid since they have the lowest incomes and typically belong to historically disadvantaged caste groups such as the Dalits, are at the forefront of the changes sweeping the rural economy today. Even a few years ago, women in Mamta’s circumstances would have joined their parents to work in others’ farms. Not anymore.

Dalits are increasingly turning away from farms because they associate farm jobs with the indignities of the past, said Chandrabhan Prasad, a Delhi-based Dalit intellectual. Dalit women, in particular, have long borne the brunt of caste and sexual oppression in rural India.

Read more: Why women aren’t taking up farm jobs


PDS efficacy

In response to an email exchange regarding my previous post on the impact of the PDS on food security, I looked up some data on the efficacy of the program, as experienced by BPL beneficiaries.



% of BPL households who did not get any foodgrains from the PDS in the previous 3 months



Average foodgrain purchases of BPL households from PDS in previous 3 months
–          Kg/month



–          As % of entitlements



% of BPL respondents who said that they ‘normally’ get their full PDS entitlements



% of BPL respondents who agree with the entries in their ration cards



% of BPL households who skipped meals in the previous 3 months



% of BPL households who would support the PDS being replaced with equivalent cash transfers



Source: PDS Survey cited in “An Uncertain Glory” by Jean Dreze and Amartya Sen, page 208. The survey was conducted by Khera et al in 2011 based on a random sample of 264 households in 24 villages of Bihar and Chhattisgarh (six villages per district in two districts of each state).

What is interesting about this data is the stark contrast in the experience of respondents in the two states.  The Chhattisgarh PDS program underwent a revival starting mid-2000s, based on a firm political decision to make it work, in the belief that restoring the system was critical for winning votes.  I won’t go into the details of the revival (you can read about it in the book).

The point I want to make with the above table is twofold, that a) it is possible to fix the leakages in the PDS system, and b) if the system works well, the beneficiaries actually prefer PDS food in-kind rather than “equivalent” cash.

PDS impact on food security and poverty

In my previous blog about the National Food Security Act, I had discussed whether cash transfers are an effective replacement for providing food through PDS.  Here are a few related facts from recent research which provides fascinating data on the impact of PDS over the last 15 years.

I have culled out the  points which quantify the impact of PDS in an attempt to answer the question: Is it worth keeping (and expanding) the admittedly leaky PDS system we have in place? I have combined ideas from various research, but almost all the data cited is the from the sources listed at the bottom of the post.

PDS impact – on calorie intake

  1. Calorie intake of the country was about 6% higher in 2009-10 than would have been without PDS.  Cereal consumption is about 7% more than without PDS and non-cereal calories are also higher though not significantly.
  2. For the same income (inclusive of value of PDS food), those without any PDS access had about 10% lower calorie intake than those who did have PDS access (2009-10).
  3. The proportion of people NOT getting FAO-determined absolute minimum of 1800 calories was 40% in 2009-10, and would increase to over 50% in absence of PDS. The proportion below the recommended 2200 calories would go up from 70% to over 80%
  4. Just having PDS available locally, even if the household does not actually purchase PDS food, also increases calorie intake, possibly because of the assurance that PDS provides.
  5. Over the last 15 years (1994 to 2010) there has been a puzzling 15% decline in per capita calorie consumption despite economic growth. This reduction was only 8% for those who bought PDS cereals but over 20% for those with no PDS purchase at all. This is another way in which PDS is contributing to food security.

PDS impact – on poverty reduction

  1. 30% of the total reduction in poverty headcount ratio from 2005 to 2010 can be attributed purely to PDS, and the rest to increases in out of pocket MPCE (due to economic growth or other anti-poverty programs).
  2. The study also finds that for the period from 1993 to 2010, the PDS and MDM (Mid Day Meals) lifted more people out of poverty than income growth, particularly so in rural India. This is because population growth continued to erode the poverty reducing impact of GDP growth. (I have to admit that I couldn’t quite figure out the analysis behind this claim, but that’s for further reading…)

PDS versus cash-transfer

  1. The percentage increase in calorie intake due to PDS is in almost all cases is much larger than the cash value of PDS transfers as a percentage of MPCE (Monthly Per Capita Income) So if PDS were to be replaced with cash transfers, it would need to be replaced by 3.5 times the current value of PDS grain.
  2. Another way to think about this is that if you give people cash instead, they may not spend it on food. So we have to be clear whether the purpose of PDS is food/nutrition security or income enhancement. If the former, there is plenty of evidence besides this research (including the ones cited in my previous blog), that cash transfers cannot compensate for supply food items.

The recently passed National Food Security Act slightly expands the current PDS program. Although it is not perfect by any means, and doesn’t fix the already exiting 40% leakage, given the high impact of PDS on national food security, it seems a worthwhile way to spend ~1% of the country’s GDP.



  1. “In-Kind Food Transfers – I: Impact on Poverty”, Himanshu and Abhijit Sen, EPW, Nov 2013
  2. “In-Kind Food Transfers – II: Impact on Nutrition and Implications for Food Security and its Costs”, Himanshu and Abhijit Sen, EPW, Nov 2013

Mobile phones and women farmers

Narbada Bai and Meera BaiThere is a lot of interest in using mobile phones as the “last mile” connectivity for providing various services in rural India:  Services ranging widely from education, awareness building, and a variety of health services.  Many of these services are intended for women.

My team’s study of women farmers’ access to and usage of mobile phones has thrown up some fascinating insights.

For example, while only 10-25% of women farmers own mobile phones, the majority have access to phones at some point in the day.  Our survey shows that women have access to mobile phones typically from 3 – 5 pm and 7 – 11 pm in rural Karnataka and from 7 – 9 am in Odisha.

But access does not mean attention: Competing with TV watching time and saas-bahu dramas can only lead to ineffective outcomes. Therefore, our study concluded that the best time to provide women farmers information through alerts or pre-recorded messages is between 3 – 5 pm and 9 – 11 pm in rural Karnataka and between 7 – 9 am in rural Odisha. 

Intriguing, isn’t it?

Programs targeting women often fail to understand the nuances of how women actually live and work, even though attention to detail such as this is critical for their success.

Read the full report with information on rural women’s ownership, access and usage of mobile phones here.

Caution: Women at work

“No, Mamma has to cook and Papa has to work on the computer,” corrected my daughter’s friend, to bring about the proper stereotype in their Mummy-Papa role play.  So dutifully the father typed away frantically on a pretend laptop while the mother fretted away in the kitchen trying to get dinner ready.

I used to think that times are changing as a greater percent of women are entering the workforce, and by the time today’s kids grow up, the reality of women and work would be different.  However, a friend who researches women and work pointed out that women’s participation in the workforce in India has actually been declining.

At first I assumed that this was a purely rural phenomenon, a result of no employment options for women.  But when I dug up national level data and research, I realized that the trend is much stronger than I originally suspected.

The Facts

The labor force participation rate (LFPR) of rural women has been declining from 31% in 1983 to 24% in 2012, and roughly stable for urban women at ~18%.  These numbers are for “principal status”.  Principal status is defined as the main activity undertaken by an individual for more than six months in the year.  If we take into account subsidiary status (activities taken up for less than six months), the rural LFPR dropped from 37% to 30% and the urban one from 21% to 20%.

In rural areas there is actually a decline in absolute number of women in the workforce over the last decade.  In urban areas, while there has been an increase in absolute numbers, the growth rate has barely kept pace with increase in population of women, leading to a more or less flat LFPR across the years.

Here are the pertinent numbers:

Women work 1

The LFPR% chart below shows a long term trend of withdrawal of women from the workforce, with a small exception of an upward trend in the last two years in urban areas.

Women Work 2

The ‘U’ curve

Economic and sociological theory holds that as the income of households at the bottom of the pyramid increases, women withdraw from the workforce to focus almost entirely on domestic activities.  At even higher income levels, households produce highly educated women who participate in the workforce, competing with men for similar kinds of jobs. This results in a ‘U’ shaped labor force participation curve mapped against income deciles.

The ‘U’ shape more or less holds for urban India, although two noticeable changes have taken place in the last decade and half.

Urban Female LFPR by MPCE
Urban Female LFPR by MPCE

First, the overall U sits lower along the y-axis due to overall reduction in labor participation in the last decade.  Second, the upper income (right) portion of the curve now rises lower as more and more high income women are withdrawing from the workforce. Among urban high-income households, it is unclear to me why there is a discreet jump from ~30% LFPR in the 80s and 90s, to ~15% in 2000s rather than a spread, as in the case of urban poor.

Rural Female LFPR by MPCE percentile
Rural Female LFPR by MPCE percentile

In rural areas, the ‘U’ curve was never in place in India due to no economic opportunities for educated women.  This sad reality hasn’t really changed in recent years.  As in urban areas, the drop in labor participation of urban women is higher in the higher income brackets.

‘Producing’ status

Data shows that this withdrawal from workforce cuts across all age groups, and is not limited to typical child-bearing age in rural or urban India.

And, the reverse also holds true: that is, female participation in domestic activities as their primary occupation has been increasing, after adjusting for increase in enrollment in institutes of higher education.

So what value does a household get out of withdrawal of women from the workforce?

The act of withdrawal of women from wage-earning work signals financial security.  Secondly, the women’s focus on domestic affairs enables greater investment in children’s education and household wellbeing. Thus instead of producing income, women focused on domestic work produce ‘status’.

In 1983, 61% of urban women graduates participated in the workforce, compared to only 26% now.  Education is widely regarded as one of the key tools of empowerment of women to enhance their autonomy and agency.  But if the women are focusing entirely on domestic affairs and the production of household status, how does education matter?

According to research cited in this paper, “Education among women does not necessarily increase their ‘autonomy’ in substantive ways, rather it may only lead to modernization and internalization of patriarchal norms”.  In fact, “ ‘Schooling seems to inculcate discipline, self-restraint, patience, routine and obedience to authority among girls.’”

More marginalized work

In the agriculture sector, which is the main employer of rural women, we find that women who work are withdrawing from agriculture as a primary activity and focusing on home-based activities such as preparation of seeds, cleaning, storing, etc. or on livestock. And those at the bottom of the economic pyramid are engaging in more work as ‘casual labor’.  So not only are women withdrawing from work, the ones who do engage in production activities are engaging in marginal activities rather than core activities.

What does this add up to?

So, we have fewer and fewer % of women working overall.  And in rural areas, those who are working are increasingly working in supplemental activities or as casual labor.  In urban areas, 85% of educated women are choosing to invest their effort exclusively on household activities.

Whether this reflects their actual preference or is a reflection of societal pressure or lack of spousal support in household matters, I can’t say.

But what I can say is that the stereotype scenario played by young girls the world over is more prevalent in India now than it was 20 years ago.

[Note: LFPR data and charts cited in this article are reproduced from “Missing Labour or Consistent ‘De-Feminisation’?” by Vinoj Abraham, Economic & Political Weekly, August 3, 2013. Agriculture related information is from NSSO reports.

A version of this article also appeared at]