The Credibility Revolution in Empirical Economics
Causal Question / Estimand
A methodological manifesto rather than a single estimand. The question is whether empirical microeconomics has become more credible since Leamer’s (1983) “take the con out of econometrics” critique, and what made that happen. The implicit estimand throughout is a well-defined treatment effect (Causal-Estimand) recovered by a transparent research design.
Identification Strategy
Argues that the engine of improved credibility is research design — leaning on sources of variation that approximate Randomization — not Leamer’s proposed remedy of sensitivity analysis. Surveys the design-based toolkit: randomized field experiments, instrumental variables / natural experiments, regression discontinuity, and difference-in-differences. Contrasts this “experimentalist” paradigm with structural/CGE modeling, arguing design-based work makes its identifying assumptions explicit and defensible.
Key Assumptions
The paper is about assumptions in general: Potential-Outcomes, Randomization (and as-good-as-random natural experiments), Parallel-Trends (for DiD), Continuity-at-Cutoff (for RDD), and Exclusion-Restriction (for IV). Its thesis is that credible work surfaces and scrutinizes these, rather than hiding them in functional form.
Threats to Validity
n/a — methodological essay. Acknowledges the standard critique that design-based estimates buy internal validity at the cost of external validity / generality, and defends the trade-off as worthwhile.
Setting / Data
n/a — survey of applied microeconomics (labor, education, development), citing exemplary studies rather than running its own analysis.
Key Claims
- The “credibility revolution” is real and driven by research design quality, in the spirit of Leamer’s critique but not his remedy.
- Randomized and quasi-experimental designs make identifying assumptions transparent; this is the path “out of econometrics” of the con.
- Structural macro/CGE modeling is criticized for opaque, hard-to-falsify assumptions.
- Internal validity (a credible causal claim in some population) is a prerequisite for useful external extrapolation.
Connections
- See also: Angrist2022-EmpiricalStrategiesIlluminatingPath, AngristKrueger1999-EmpiricalStrategiesLaborEconomics, Meyer1995-NaturalAndQuasiExperiments, Foundations
- Builds on: the Potential-Outcomes tradition of Rubin1974-EstimatingCausalEffects and Holland1986-StatisticsAndCausalInference
Citation
Angrist, J. D., & Pischke, J.-S. (2010). The Credibility Revolution in Empirical Economics: How Better Research Design Is Taking the Con out of Econometrics. Journal of Economic Perspectives, 24(2), 3–30. https://doi.org/10.1257/jep.24.2.3