While I have tried to provide a conceptual overview in this essay, recent research in development economics has been almost entirely empirical. A veritable explosion in computing power, the expansion of institutional datasets and their increased availability in electronic form, and the growing ease of collecting one’s own data has bred a new generation of development economists. Their empirical sensibilities are of a high order; they are extremely sensitive to issues of endogeneity, omitted variables, measurement error and biases induced by selection. They are constantly on the search for good in- struments or natural experiments, and when these are hard to find, they are adept at creating experiments of their own.
There is little doubt that we know little enough about the world we live in that it is often worth finding out the simple things, rather than continue to engage in what some would term flights of theoretical fantasy. Are people really credit-rationed? Does rising income automatically make for better nutrition and health? If we had the option to throw in more textbooks, or reduce class size, or add more teachers, or install monitoring devices to track teacher attendance, which one should we do? Do women leaders behave differently from men in the policies that they adopt? Do households behave as one frictionless unit? Or, if one is the big-picture sort, have countries indeed converged over the last 200, 500 years? are richer countries more democratic? How many excess female deaths have occurred in China or India because of gender bias? Are poorer countries more “corrupt”?, and so on. The list is practically endless.
Why can’t well-trained statisticians answer these questions?, the somewhat churlish theoretically-minded economist might ask. Why do we need economists, who are sup- posed, at the very least, to combine two observations to form a deduction? The answer, at one level, is very simple and not overly supportive of the churlish theorist’s complaint. While the questions are straightfoward, the answers are often extremely difficult to tease out from the data, and you need a well-trained economist, not a statistician, to under- stand the difficulty and eliminate it. Because of the aforementioned econometric issues, not a single one of the questions asked above admit a straightforward answer. Devel- opment economists spend a lot of time thinking of inventive ways to get around these problems, and it is no small feat of creativity, dedication and extremely hard work to pull off a convincing solution.
It is true that the very desire to obtain a clean, unarguable answer — with its attendant desire to have control over the empirical environment — sometimes narrows the scope of the inquiry. There is often great reluctance to rely on theoretical structure (for such reliance would contaminate the near-lexicographic desire for an unambiguous result). This means that the question to be asked is often akin to that for a simple production function (e.g., “do students do better in exams if they are given more textbooks?”) or is focussed on the direct effect of some policy intervention (“does the provision of health checkups improve health outcomes?”) So it is that a boring but well-identified empirical question will often be treated with a great deal more veneration (especially if a clever instrument or randomization device is involved) than a model which relies on intuitive but undocumented assumptions.
That said, it is also a fact that we know very little about the answers to some of the most basic questions, such as the ones we’ve listed above. The great contribution of empirical development microeconomics is that we are building up this knowledge, piece by piece. Whether the search for that knowledge is informed by theory or not, there will be enough theorists to attempt to put these observations together. There will be enough empirical researchers to keep generating the hard knowledge. Development economics is alive and well.