THE ROLE OF TIME SERIES ANALYSIS IN STUDIES OF ACCOUNTING POLICY CHOICE: A COMMENT

The study by Anderson and Zimmer [1992] of goodwill accounting policies uses a pooled time series experimental design. This approach can add substantially to our understanding of accounting policy choices, but not in the manner used by Anderson and Zimmer. Where accounting policy choices are believed to be independent from one period to the next, then a time series approach can greatly enhance our ability to capture the influence on such policy choices of changing circumstances, more so than a simple cross-sectional test. Conversely, if accounting policy choices are not independent between periods, pooling over time can overstate significance levels of statistical tests. The nature of Anderson and Zimmer's data makes the impact indeterminate. However, even under an extreme assumption, pre-regulation evidence remains significant at conventional levels.