The heterogeneity of Okun's law: A metaregression analysis

Abstract

Okun's law, a significant parameter in empirical research and policy analysis, faces considerable heterogeneity. This stems from its dual interpretation in the literature, with one implying unemployment's effects on output and the other suggesting output's effects on unemployment. Consequently, comparing results from these approaches is not straightforward. Even within each approach, variability persists. Through meta-analysis and correction for publication bias, we identified the primary factor contributing to heterogeneity in both approaches: labor market characteristics (e.g., self-employment, labor laws, productive structure), leading to varying reactions of unemployment to cyclical output changes across different labor markets. The second most influential factor was methodological issues (data type, frequency, spatial coverage, sample period, etc.), highlighting how researchers' decisions impact results. Lastly, the underlying theoretical model also accounted for some variability. Okun proposed three models to estimate the relationship, which yielded comparable results for the US economy, but for other economies this was less evident.