Matching and the Properties of Accounting Earnings Over the Last 40 Years

Abstract

This paper presents a theory and empirical evidence that investigate the effects of poor matching on the properties of accounting earnings. The key intuition of this theory is that poor matching manifests as noise in the economic relation between revenues and expenses. As a result, poor matching decreases the correlation between contemporaneous revenues and expenses and increases the correlation between non-contemporaneous revenues and expenses. With regards to earnings effects, poor matching increases earnings volatility, decreases earnings persistence, and induces a negative autocorrelation in earnings changes. Since poor matching resolves over time, we expect that all of these effects are less pronounced over longer-horizon definitions of earnings. The empirical tests concentrate on documenting the effects of matching and the associated properties of earnings in a sample of the 1,000 largest U.S. firms over the last 40 years. We find a clear and economically substantial trend of declining contemporaneous correlation between revenues and expenses, while the correlation between revenues and non-contemporaneous expenses is increasing. We also find strong evidence of increased volatility of earnings, declining persistence of earnings, and increased negative autocorrelation in earnings changes. As expected, these trends are less pronounced for longer-horizon definitions of earnings. Based on this evidence we conclude that accounting matching has become worse over time and that this trend has produced a pronounced effect on the properties of the resulting earnings. This evidence also suggests that the FASB's stated goal of moving away from matching and towards more fair-value accounting is likely to continue and deepen the identified trends in the properties of earnings.

Keywords: matching, earnings properties

JEL Classification: M41, M44, C21