Midday Meals

It's definitely great to be back in Chennai, India after three years. A lot has changed here and I'll certainly keep you posted on that in the coming weeks.

I'm here for the purposes of my dissertation. I'm writing three papers on "The Economics of Child Health" and one of these will be on studying the health and education effects of large scale nutritional programs in developing countries. One of the most famous programs is administered right here in the state of Tamil Nadu. In 1982, motivated by a mixture of compassion (hopefully) and electoral populism (more likely), then Chief Minister MG Ramachandran put in place a program designed to offer a free lunch to all children aged 2-9, 365 days of the year. A year later, this program was extended to children 10-15 and elderly pensioners.

The anecdotal success of the program has spurred the Government of India at large to mandate all states to implement similar programs. (As yet, not all states have instituted such programs - the political economy of this policy lag is another paper waiting to be written perhaps). In addition, the Tamil Nadu Midday Meals program has motivated other countries to think about developing similar schemes, as well. This is a big deal both politically and fiscally: in Tamil Nadu, at least, this program takes up a sizeable chunk of the social welfare budget. However, there is little empirical evidence documenting the causal effects of the program on nutrition and schooling. Whatever few studies exist have trouble distinguishing between program effects and secular trends in economic development in the state (increase per capita income, better infrastructure, etc).

The aim of my project is to investigate this question more thoroughly. For a variety of reasons, this is going to be very challenging:

-The program is not a randomized experiment, and was implemented all at once across the state. Hence, there is little cross-sectional (i.e., spatial) variation that can be used to identify treatment effects.

-Data on program availability by village is almost necessary in order to develop convincing evidence. Unfortunately, either this evidence does not exist, or it is impossible to get thanks to politics (see below).

My main approach is to milk discontinuities in age eligibility and the timing of the program in order to estimate the causal effects. For example, individuals between the age of 1.5-2 are not eligible for the program, while those aged 2-2.5 are. One can use this information to compare kids in each age group to see if there is a nutritional effect. Of course, if families redistribute resources amongst kids, this comparison becomes much more complicated. Another example: I can look at long term effects, using the fact that individuals of different birth years were exposed to varying lengths of the treatment. This is true because the program came into existence in 1982, and eligibility requirements were changed in different years.

Because its so hard to get data on program center locations, I also have some backup questions, which involve studying the effects of meal programs at an all-India level (much easier from the econometrics point of you), and studying a different health and nutrition program aimed at children aged 0-6 for which there is ample cross-sectional and time series variation both within Tamil Nadu and at the all India level.

About getting data: I went over to the Tamil Nadu Planning Commission and the Department of Social Welfare offices to make my data request. Through the 2005 Right to Information Act, any information request must be addressed within 30 days. Of course, when I went over, I was told that the task was impossible and that the data I need doesn't exist, even at a district (equiv to counties in the U.S) level. I know that this is not true because other researchers (who don't like to share) have used the district level information. I was told later by an academic contact that because the program is heavily politicized, government officials are loathe to give out detailed information. I am making another attempt today, so we'll see.