The Robustness of Conditional Logit for Binary Response Panel Data Models with Serial Correlation

Abstract

We examine the conditional logit estimator for binary panel data models with unobserved heterogeneity. A key assumption used to derive the conditional logit estimator is conditional serial independence (CI), which is problematic when the underlying innovations are serially correlated. A Monte Carlo experiment suggests that the conditional logit estimator is not robust to violation of the CI assumption. We find that higher persistence and smaller time dimension both increase the magnitude of the bias in slope parameter estimates. We also compare conditional logit to unconditional logit, bias corrected unconditional logit, and pooled correlated random effects logit.

Accounting Quality and Household Stock Market Participation

“…This is because, under conditional or fixed effect logit and probit models, any families that have no variation in outcomes will be dropped from the estimation which induces a selection bias towards larger families(Miller, Shenhav, and Grosz, 2018). Moreover, according toKwak, Martin, and Wooldridge (2018), when dependent variables are serially correlated, it violates a key assumption of logit estimation, Fixed-Effect OLS is preferred to either conditional or unconditional logit.16 In an untabulated analysis, I scale the sum of audit fees by state-level market capitalization of public companies, which does not change my inferences.…”