I am teaching an undergraduate class on development microeconomics starting in a few weeks, so I have been reading and thinking about the field quite a bit in preparation for the semester.
One of the things I find striking is how economic theory has regretfully been given short shrift over the past few decades in development economics. This is likely almost entirely due to increases in computing power in the early 1990s,* which led to a much greater demand for data, which in turn fostered more systematic efforts at collecting household survey data in developing countries.**
This made the field of development economics much more empirical than it used to be, and the Credibility Revolution, combined with the rediscovery of randomized controlled trials, came as a one-two punch that pushed the field into almost exclusively empirical territory. And obviously, it didn’t help that the returns to empirical research were much larger than the returns to theoretical research, which had already declined considerably by then.
(To convince yourself of the foregoing, pick up an issue of the Journal of Development Economics from the early 1990s and compare it to one of the most recent ones; the theory-to-empirics ratio has virtually been inverted. Or just think about how the standard text for graduate development micro remains Bardhan and Udry–a book that was published in 1999!)