Robust Inference in Single Firm / Single Event Analyses

Abstract

Single firm / single event (SFSE) studies are relevant in corporate finance. Since inference on
abnormal returns in this context necessarily relies on the time series variance of these abnormal
returns, the implied problem of heteroscedasticity is obvious, although hard to solve. We analyze
robust inference in an SFSE setting using Monte Carlo and resampling experiments. Estimation is
biased when the calibration and event period occur in different volatility regimes. We develop a
unique specification test for these structural breaks. The most robust inference is obtained by using
intraday data and a multiplicative component GARCH estimator.

Keywords: Event studies, Inference, Monte Carlo simulation, Volatility, Structural breaks

JEL Classification: C12, C15, G14, K41