抄録
In the broad field of social sciences, including economics and social epidemiology, empirical micro-analyses, especially social experiments, have been making rapid progress. This paper summarizes the evolution of suchempirical research and discusses recent trends over the three generations.
In the first generation, which developed from 1970 to the 1990s, the field of micro-econometrics was developed to solve problems specific to existing observational data. In the second generation after 1990, the “empiricalization“ of economics accelerated further, and research trends evolved toward the introduction of randomized controlled trials (RCTs) to promote evidence-based policy making (EBPM) and more general causal inferences. This is called the Credibility Revolution. In addition to RCTs, the second generation features quasi-experimental methods such as Regression Discontinuity Design (RDD), Difference in Differences (DID), Propensity Score Matching (PSM), Instrumental Variable Approach (IV), and Synthetic Control Method (SCM). In general, “quasi-experimental methods“ are studies based on observational data that estimate causal effects with an accuracy comparable to that of RCTs, based on carefully “designed“ statistical models.
The second generation, or the credibility revolution, contributed to making empirical economics as a science, at the same time, but it faced the problems of reproducibility and replicability. Typical problems include p-hacking, HARKing (Hypothesizing after the Results are Known), and other Questionable Research Practices (QRPs). In addition, the “impracticability “ of policies and the “exclusivity“ of stakeholders have been exacerbated.
While the issues of “reproducibility and replicability“, “impracticability, “ and “exclusivity“ are still being addressed, RCTs and quasi-experimental studies, which were central to the credibility revolution in the second generation, have evolved further by setting pre-analysis plans and pre-results reviews. Regarding the “impracticality“ and the “exclusivity, “ one of the keys seems to be the augmented use of big data (or alternative data) in the third generation. The trend of third-generation empirical research, which seeks to apply new data such as big data to existing fields and new issues, has been further accelerated following the outbreak of COVID-19. This major trend of using big data, a.k.a., Computational Social Science, has rapidly expanded the frontiers of empirical research in the social sciences.
In sum, through these three generations of empirical research in economics, it has been clear that three elements of “the strengthened credibility of analysis “, “the arrival of the era of big data,“ and “the changing role of economics in society“ are key foundations for the next generation empirical research. Particularly, it will be imperative to strengthen the productive collaboration among industries, governments, academia, and private stakeholders. By doing so, academic frontiers will continue to expand.
