Do Cryptocurrency Markets Differentiate Infrastructure from Regulatory Shocks? A Multi-Moment Event Study with Dependence-Robust Inference

Abstract

Do cryptocurrency markets process infrastructure failures (exchange outages, protocol exploits) differently from regulatory shocks (enforcement actions, policy changes)? We study the question at both moments of the return distribution over one shared sample (50 events, six assets, January 2019–August 2025), fitting a GJR-GARCH-X variance model under a matched dependence-robust inference design. We lead with a methodological move: treating event inclusion as an explicit, measured design parameter. Rather than asserting the standard selection-on-the-dependent-variable objection away, we trace the infrastructure-regulatory variance differential across the event-inclusion screen and report the selection bias as a measured object. The result is a scope condition. Under curated, high-salience event identification the differential is sizeable—a 4.88× point-estimate multiplier, directionally stable across specifications—but it is selection-conditional: applying the impact filter mechanically to a broad reconstructed candidate pool collapses it to 1.3–1.6×. Identification is only half the story; inference is the other half. The curated multiplier is not statistically distinguishable from zero once cross-asset dependence and heavy tails are respected—a Student-t-copula CCC-GARCH-X bootstrap, our inference of record, returns p≈0.32. Because the six per-asset coefficients are strongly cross-correlated, the effective sample size for the infrastructure-regulatory contrast is closer to three than to six, and even the more favourable design-effect correction reaches only p≈0.07–0.15. A naive i.i.d. test on the same six per-asset coefficients had reported an apparently decisive fivefold effect, but that significance was an inference artefact: pseudoreplication across correlated assets, compounded by a Gaussian bootstrap draw where the fitted innovations are heavy-tailed Student-t at ν≈3. The first moment tells the same story—a +7.19-percentage-point cumulative-abnormal-return difference that an event-level block bootstrap cannot distinguish from zero (p=0.283). On this sample, under correct inference, the infrastructure-regulatory asymmetry is directional, selection-conditional, and unresolved; the contribution is a portable inference toolkit—an inference ladder and a Monte-Carlo size study—for diagnosing how cross-asset event studies in heavy-tailed markets manufacture significance, demonstrated on a worked example where it dissolves an apparently decisive fivefold result the author had himself published; the same discipline measures the selection bias that drives results of this kind rather than assuming it away.