Moretti replication published in AER
a day ago
- #Economics Methodology
- #Academic Integrity
- #Research Replication
- Replication of Moretti (2021) published in AER reveals ten issues, focusing on two major problems in the event study and IV strategy.
- Event study for selection bias uses an incorrect model, yielding null results when corrected; IV for omitted variable bias fails due to a coding error, also resulting in null findings.
- Additional minor problems include unreproducible results from many-to-many merges, reversed patent quality effects due to incorrect log transformations, and coding errors in heterogeneity analyses.
- Distributed lag model errors and misinterpretations of time unit variations further undermine the paper's robustness and causal claims.
- Author highlights challenges in getting responses from original researchers and stresses the importance of replication for policy-relevant studies.