credibility revolution
A movement in microeconometrics emphasizing research designs (natural experiments, instrumental variables) that credibly identify causal effects with minimal assumptions, which Heckman criticizes for sacrificing economic interpretability.
As defined in James Heckman on Facts, Evidence, and the State of Econometrics
Where it comes up · 5
- Raj Chetty on Economic Mobility3 claims
- James Heckman on Facts, Evidence, and the State of Econometrics2 claims
- Hal Varian1 claim
- Noah Smith on Whether Economics is a Science1 claim
- 232 | Amy Finkelstein on Adverse Selection and Hidden Information1 claim
Most important claims · 8 of 8Search all claims for “credibility revolution” →
- Empirical economics has undergone a 'credibility revolution' (Angrist's term) emphasizing isolating causal effects from spurious correlation, with the randomized experiment as one of the most valuable tools — applying medicine's randomized-trial logic to policy, as in the 2008 Oregon Medicaid lottery, where the state randomly allocated Medicaid to some low-income uninsured adults and not others to measure true causal effects.232 | Amy Finkelstein on Adverse Selection and Hidden Information
- Contrary to the interpretation that his work means 'just look at the data,' theory remains crucial in three domains: hypothesis formulation (you cannot organize data without a theory—e.g. testing Loury's cross-class interaction theory versus tight-knit-community theories), extrapolation to new policies/environments (which requires a theory to translate past findings), and general-equilibrium analysis (since the partial-equilibrium treatment-control paradigm cannot handle price changes and re-sorting because the control group gets contaminated).Raj Chetty on Economic Mobility
- Natural experiments (quasi-experiments) exploit the real-world policy-making process to achieve clean causal identification; they cannot be replicated like lab experiments since each happens only once, making them inferior to lab science, but they are substantially better than merely looking at correlations, and this 'credibility revolution' has greatly improved economists' ability to understand policy effects.Noah Smith on Whether Economics is a Science
- Since the mid-1990s, economics has seen a great flourishing of empirical work driven by the credibility revolution and better econometric models, rather than a dearth of theory being a problem—and econometrics work has been central to building coherent statistical reporting within an organization like Google.Hal Varian
- Progress in empirical economics is verifiable via replication and experiments: overlaying the randomized Moving to Opportunity experimental results on the predictions from the observational Opportunity Atlas work produces a scatter plot showing the two line up well, meaning the team could forecast how randomly-assigned movers would do on average, giving confidence the estimates are stable.Raj Chetty on Economic Mobility
- A core problem with the empirical 'credibility revolution' is that econometric results are hard to verify because they rely on many complex assumptions and cannot be checked against ground truth the way physics can (e.g. predicting exactly where a spacecraft lands); psychology has seen much of its empirical work fail to replicate, and economics may face a similar problem.Raj Chetty on Economic Mobility
- The credibility revolution reduced bias but became obsessed with controlling for unobservables without asking what is actually being estimated, trading economic interpretability for statistical credibility, so its results identify hard-to-interpret statistical artifacts that cannot answer the structural economic questions needed for policy.James Heckman on Facts, Evidence, and the State of Econometrics
- Instrumental variables, central to the so-called credibility revolution, are not new techniques—they trace back to Philip Wright or Sewall Wright around 1928 and have been a central part of econometrics for 70-80 years.James Heckman on Facts, Evidence, and the State of Econometrics