This paper presents a statistical procedure to identify effects of three potential structural changes on accounting earnings – temporary, short-run, and long-run. The procedure is applied to quarterly accounting earnings of 39 utility companies. Structural changes are found to be commonplace. Statistical forecasting models that explicitly incorporate structural change effects are found to generate more accurate forecasts than other statistical models in the literature. Although no statistical model significantly dominates Value Line, a firm-specific model with structural change adjustment forecasts as well as Value Line. Moreover, all statistical models examined have significant marginal forecasting power to complement Value Line forecasts.

We applied the VARMA test to examine the dynamic relation between prices and interest rates. The dynamic relation, which is important to characterize the nature of the Gibson paradox, provides economists new insight in discriminating against competing theories. In light of our empirical findings, all theories in the literature lose their persuasiveness. We found some evidence of unidirectional relation from prices to interest rates, but we found no evidence of unidirectional relation from interest rates to prices. Hence, the business cycle explanations advanced by Wicksell (1907), Keynes (1930), Lee and Petruzzi (1986), and Barsky and Summers (1988) are especially in jeopardy. A century and a half after its birth, this paradox is more puzzling than ever. Copyright 1990 by MIT Press.

We examine how the qualitative characteristics of a non-GAAP earnings disclosure reveal the quality of the non-GAAP performance metric itself. We measure the qualitative aspects of the non-GAAP disclosure using an index of 12 hand-coded characteristics and presentation choices that provide context to non-GAAP disclosures. We find that more transparent qualitative information in non-GAAP disclosures is associated with more transitory non-GAAP exclusions and a lower likelihood that managers will aggressively exclude expenses in calculating non-GAAP earnings to meet or beat analysts’ forecasts. We also find that the association between non-GAAP exclusions and earnings announcement period returns is less negative for firms with more transparent qualitative information accompanying their non-GAAP disclosures. Overall, these results suggest that qualitative characteristics of non-GAAP disclosures convey useful information about the quality of non-GAAP earnings metrics.