[Baran, Lakonishok, and Ofer [1980] present some empirical evidence on the information content of general price-level adjusted (GPL) earnings relative to historical cost (HC) earnings. Their statistical tests are shown to be invalid, thus leaving unsupported their main conclusion-GPL earnings are superior. We apply an appropriate statistical test to their data and find contrasting results: GPL betas exhibit significantly better correlation with market betas than do HC betas for market value-denominated betas, but not for a non-market-denominated beta. No conclusion can be rendered on the information content of GPL earnings, and the controversy over accounting beta construction is rejoined.]

ABSTRACT: Baran, Lakonishok, and Ofer [1980] present some empirical evidence on the information content of general price-level adjusted (GPL) earnings relative to historical cost (HC) earnings. Their statistical tests are shown to be invalid, thus leaving unsupported their main conclusion-GPL earnings are superior. We apply an appropriate statistical test to their data and find contrasting results: GPL betas exhibit significantly better correlation with market betas than do HC betas for market value-denominated betas, but not for a non-market-denominated β. No conclusion can be rendered on the information content of GPL earnings, and the controversy over accounting β construction is rejoined.