Zum Hauptinhalt springen

I4R Discussion Paper Series #28


Thomas Brailey (University of Oxford), Matthew Hepplewhite (University of Oxford), Scott Moser (University of Nottingham)

Direct Replication and Additional Sensitivity and Robustness Analyses for Frederiksen (2022): A Replication Report from the Nottingham Replication Games

The long-recognized spurious regressions problem can lead to mistaken inference in panel instrumental variables (IV) estimation. Spurious correlations arising from correlated cycles in finite time horizons can make irrelevant instruments appear strong with signable consequences for estimated IV coefficients, or interfere with valid of inference of causal effects from IV coefficients estimated using relevant instruments. The inclusion of time fixed effects in interacted specifications does not always resolve these problems. We demonstrate these concerns by revisiting recent studies of the causal origins of conflict. We offer diagnostic and corrective recommendations for avoiding the pitfalls arising from time series exhibiting persistence.