Composition bias in derivatives-market regulation: evidence from India, Korea, and the United States

When market participation changes rapidly, aggregate derivatives volume confounds who trades with how much each participant trades. This composition bias can distort measured regulatory impact. I develop a diagnostic framework and test it in three settings using interrupted time-series and difference-in-differences designs: India (129 monthly observations, April 2015–December 2025), Korea (96 monthly observations), and the United States (132 monthly observations), supplemented by boundary-condition markets. In India, composition-adjusted specifications eliminate the apparent disclosure non-response, showing that aggregate inference is not robust to participation growth. In Korea, stable participation during the 2012 contract-multiplier increase allows aggregate data to identify a genuine 76.8% differential decline. In the United States, surging post-2019 retail participation produces a 54.1% differential equity-options surge, consistent with inflation bias, subject to control-group contamination. Counterfactual-regression analysis confirms that disclosure produced only a marginal departure from the covariate-predicted path, while regulation generated a − 67% divergence ( t = −25.77); placebo-date tests (Bertrand, Duflo, & Mullainathan, 2004; MacKinnon & Webb, 2020) reject the null in all three settings. I provide a four-step diagnostic recipe that any securities regulator can implement with publicly available data.