| Abstract: |
How far has the credibility revolution spread beyond applied microeconomics? I
update Currie, Kleven, and Zwiers (2020b) using approximately 44, 000
papers—31, 500 NBER working papers (1982–2025) and 12, 300 articles from
eleven top economics and finance journals (2011–2024)—measuring mentions of
empirical methods through keyword matching. Three findings emerge. First,
finance and macro/other fields differ substantially from applied micro in
their mention of credibility revolution methods: as of 2024, 63 percent of
applied micro papers mention experimental or quasi-experimental methods,
compared to 47 percent in finance and 39 percent in macro/other. The current
levels in finance and macro/other are comparable to where applied micro was in
2008–2010, though the long-run trajectories may differ. Second, growth outside
applied micro is driven overwhelmingly by difference-in-differences; including
DiD raises the share of finance papers mentioning any experimental or
quasi-experimental method by roughly 55 percent versus 30 percent for applied
micro. Other quasi-experimental methods—instrumental variables, regression
discontinuity, experiments—have seen far less growth. Third, I document a
striking gap between the methods studied in the Journal of Econometrics—where
nonparametric estimation and asymptotic theory dominate—and those used by
applied researchers, where DiD and identification strategies dominate.
Published journal articles confirm these patterns are not artifacts of the
NBER sample. |