Triple Differencing Bikes in India

In 2007 in an effort to increase the number of girls enrolled in school the government of Bihar in India gave each schoolgirl of age 14 a bicycle. The excellent Karthik Muralidharan and co-author Nishith Prakash set out to discover whether the program was effective. To jump to the conclusion they found that the program increased the enrollment of girls by 41% reducing the gender gap by almost half.

The reason for this post, however, is not the result–important as it is–but the two videos the International Growth Center made to explain Muralidharan and Prakash’s research methods. The first video explains the background of the research and then gives a very elegant explanation of triple-differences as an estimation strategy.

The second video explains that the researchers still weren’t completely happy that they had truly identified a causal effect (or perhaps the referees were not completely happy) so they hit on a complementary approach, looking for a dose-response relationship. With the collection of more data Muralidharan and Prakash were able to ask whether the program was more effective for the students who were neither so close nor so far from the school that a bicycle wouldn’t make a difference. Indeed, the program was most effective for students who lived at bicycle-relevant distances.

These videos are an interesting peek at some of the questions economists ask and the methods they use to answer those questions. The videos would be excellent for classroom use–challenge your students after the first video to come up with potential problems with the triple difference method and see if they can identify another research design that would address these problems!

Addendum: Here are previous MR posts on Karthik Muralidharan’s important research program.


That's awesome. Imagine if we did that in the U.S.? India per capita GDP is much lower than the U.S. so a bike is a lot more, relatively speaking, than it would be in the U.S. Imagine if a University gave a low enrollment demographic group, a free car to everyone within 15 miles of the University campus as part of a giving back to the community experiment. Would the enrollment of that demographic group go up at that University?

A quick smell test is in order for this idea. The Indian program budgeted $45 (and later, less than $60) for each bike. The per capita GDP of India is $1500.

The cheapest new car sold in the US is about $10,000; the US PC GDP is about $53,000. So a cheap new car is about 20% of PC GDP in the US. In India, a new bike is about 3% of PC GDP.

It's probably both easier and cheaper to figure out cheap accommodations on or near campus, and indeed that is what many universities offer.

They do? In my experience universities charge at least the market rate for housing


Excellent. I don't know how generalizable the results will be (it seems doubtful that simply handing out bicycles can overcome other obstacles besides distance, e.g. sexism -- boys don't seem to have needed the bicycles, quite possibly because their families regard their education as more important than the girls' education). But good for them, and as Alex says, big props to the IGC for these videos. They're the best video presentations of econometrics that I've seen (granted I've seen very very few econometrics videos, but excellent job regardless.)

Hmmm - not sure about "owning a Bicycle" and being a "Rockefeller" - seems to me that bicycles have been, since the advent of the "Safety Bicycle" (the diamond frame bicycle) cycling has been more closely associated with very egalitarian trends. I.e. affordability for the masses.
Interesting discussion of history in this matter @

The historian on that Bike Show episode also notes the impact of cycling on women and the economy (check out starting minute 10:30). There are economic bits throughout, tho.

Sounds like overall good news, something we can all be optimistic about. But since it has nothing to do with race, politics or Paul Krugman, this post only gets 8 replies.

Allow me put in my two cents worth of twenty-twenty hindsight. Does anybody agree with me that these two gentlemen should have started the second research in the first place (measuring the commuting distance) than taking route to via DID or DIDID method?

Difference in Difference should be used when you are sure that covariances are there but not observable (if they can be observed, you can simply stick them in to a model). DID will cancel out all the effects from these unobservable covariances, without revealing what they are or how they work. Distance between a school and a student's residence *is* a measurable (observable) covariance that governs the dropout ratio. If the two researchers knew it, they wouldn't take trouble of testing DiDiD method or creating animation for this method (though I admit that the clip is excellent).

Or, maybe I'm wrong. DiD (DiDiD) might have provided some heuristic value for the researchers and they wouldn't come up with the idea of the second research if it were not for the first. What do you think?

Comments for this post are closed