実証開発経済学の分析手法の最近の動向について

計量経済分析における「内生性」問題を中心に

抄録

In recent years, there has been a rise in the scientific threshold for establishing causality in the empirical development economics literature. In light of this development, this article discusses some specific contexts where 'endogeneity problems' arise (where systematic correlation between a right-hand-side variable and the error term in a regression equation is suspected) in microeconometric analysis using household-level data in developing countries. The article also reviews a few of the recent trends in the methodological advances in the empirical development literature that are likely to be behind the rising scientific standards in our profession. These are: the explosion in the availability of micro (household)-level data in developing countries; innovations in econometric methods (especially the search for ever 'cleverer' and 'more credible' instrumental variables, including 'natural experiments'); and innovations in data collection designs and set-ups (including randomized policy evaluations and field experiments by experimental economists). The article further argues that such recent developments in the literature could potentially provide great opportunities for agricultural economists in Japan, who have long traditions of conducting fieldwork and of being actively involved in primary data collection at the micro level. At the same time, however, we will need to pay a much closer attention to the distinction between the observation of simple statistical correlation, on the one hand, and the establishment of causality, on the other, in our empirical work.