Overview

The average earnings CAR is only as precise as the dependence the variance formula admits. This topic writes the cross-sectional variance of the CARs, separates 600 announcements from 30 firms, and switches design when the event is a policy with a control group rather than a market-model residual. Covariance estimators that implement those dependence structures sit in the causal-extensions topic.

By the end of this note you should be able to:

• estimate the variance of an average CAR from the cross section, and state that clustering breaks that estimator;
• count firms and announcements as different sample sizes in the Dow file;
• compute a two-group, two-period difference in differences from four cell means.

1. Contents
2. The big picture