Replicating History? The Uniqueness of Historical Context and Data Scarcity in Economic History

Gay Victor, Toulouse School of Economics
Mohamed Saleh, Toulouse School of Economics

Economics is in a state of credibility crisis, as both publication bias and the large number of scholars examining the same empirical relationship tend to increase the number of publications containing ``false positives.’’ Replication offers a sound solution to this crisis; a task that several fields in economics, such as experimental and labor economics, have recently embarked on. In this paper, we first explore theoretically the credibility crisis from the perspective of the field of economic history. We build upon a Bayesian framework to derive the likelihood that an empirical finding is true (Maniadis et al., 2014), taking into account two main specificities that characterize economic history. First, what constitutes a replication in economic history, given the uniqueness of the historical context and the consequent challenge of defining external validity? Second, how can we replicate economic history studies given the scarcity of historical data and where novel data discoveries can have a large impact on results? We then explore these two issues empirically. We first replicate Boehnke and Gay (2019)’s study on the impact of World War I military fatalities on female labor force participation during the interwar period. We conduct this replication exercise across various contexts. Second, we replicate the papyri-based empirical evidence in Saleh and Tirole (2019)’s study on the impact of the religious fervor of tax authorities on taxation and conversions under the early Arab Caliphate. We finally evaluate whether and how replications in economic history can inform the credibility crisis, not only in economic history, but in economics more generally

See extended abstract

 Presented in Session 39. Problems with Data and Measurement