This time next week, Diane and I will be back in India. For now, I’m blogging from a hotel room near the American Economic Association conference in Chicago. Avinash is napping – recovering from a long day of interviews – and I’m still mentally at the first panel I attended today: four similar presentations of the effect of a disaster on children’s health.
I should preface all of these comments with the warnings that I haven’t read any of the papers, and that I might not be qualified to evaluate most of them even if I had. Based on the presentations, they all seemed nicely and carefully done, and surely – in focusing on young children’s health – are speaking to what we at the RICE blog think has got to be one of the most important topics out there. Lest there be any ambiguity, based on what I know, I am happy to call all of these good papers.
For the uninitiated, all four papers followed an increasingly popular pattern (at least I seem to remember so). First, they noticeed some terrible event in a poor country that could have impacted children’s health, but only while the event lasted: one financial collapse in Argentina; two armed conflicts, one each in the West Bank and Cote d’Ivoire; and one seasonal over-application of fertilizer in India. Then, they found data on anthropometric health indicators for many children: height-for-age, generally, but also birth weight.
Next they figured out which children were in utero in which trimesters during the disastrous events. Finally, they computed how average health outcomes for children in the womb during the event differed from average outcomes among other similar children who were gestated in similar places, perhaps in similar times of other years.
The difference is interpreted as the effect of the event. The big threat to all of this making sense is that some other, correlated change was going on at the same time. And the big difficulty for audience members like me is that – not being experts on the countries and events involved – it is hard to judge whether that was likely to have been the case. That, I suppose, is why we trust peer reviewers.
It is no criticism that the papers were methodologically similar. Indeed, that is a big part of what makes them science: we would not be impressed by a biologist with a too creative method of sterilizing petri dishes. They join a quickly growing set of other, similar papers that are collectively (and I believe rightly) persuading economists of important mechanisms of early life health.
So it is no surprise that, this evening, Avinash and I were trying to think of our own special event: “exactly how much do people clean their houses for Diwali,” I asked, “and how much does the holiday move around in the year?” Not enough on either count.
But Diane asked me a question that I didn’t immediately know how to answer: what did I learn about children’s stunting that I didn’t already know? Of course we learned about these events, but the Country X crisis of Year Y will never happen again. Knowing about this extra bad consequence is unlikely to increase a government’s motivation to avoid a financial meltdown, although somebody observed that it might increase a mother’s motivation to instead get pregnant some other time. Moreover, as is often true in economics papers – including mine – there was much more attention paid to the direction of effects than to the size.
A good answer to Diane’s question might have to do with learning about the effects of a disaster in, say, the first trimester relative to the third trimester, although there seems to be evidence there already, too. Of course, more evidence for things we already know something about is still very good – and not to be dismissed lightly in a world where data that let us pin down cause and effect are hard to come by. These papers have further advanced the “theory” (in quotes for economists because it does not involve math) that what happens in utero has lasting effects on health. They add a little more weight to my slightly updated beliefs, and that is valuable.
But Diane’s question stays with me. What exactly did I learn? Should I wake up Avinash, to keep searching for our event? If we had found one, I could dispense with this hand-wringing and blog about that instead.
To be sure, learning is not so binary, is not like flipping a switch. But in the absence of a study with the precision to be able to advance our understanding of the details about which Diane was understandably curious – for example, about critical periods in development, or the possibility for catch-up growth – how should we think about the next paper that makes use of the next special event? Is it right to think about it as a “replication” – a perhaps unfortunately nasty term in academics – and should that matter either way?
I bet Diane’s suggestion would be to aim for those details. Nice work if you can get it. Another possibility – the agenda of people seeking evidence-based development policy – is to choose to study events that governments could and plausibly would elect to implement or prevent. Along these lines, I admit my own biases: I might have been even more excited if one of the events was a disruption in sewer service, or an airlift of hand-washing soap. Of course, good disasters are hard to find.