The Relation between Aggregate Earnings and Security Returns over Long Intervals

Research output: Contribution to journalArticlepeer-review

2 Scopus citations

Abstract

This paper provides a theoretical explanation and consistent empirical evidence for the increase in the contemporaneous correlation between returns and aggregate earnings as the return interval is lengthened. Consistent with intuition and with Easton, Harris, and Ohlson 1992, the analysis shows that aggregation over time renders the lag in accounting recognition relatively less important and thus improves the returns-earnings R2. Interestingly, the analysis also reveals that aggregating earnings over longer periods increases the positive covariance between aggregate earnings and the accounting lag, which may further increase the R2. This positive covariance can lead to an earnings coefficient greater than one over some range of aggregation, which is consistent with the findings of Easton et al. that over the 10-year interval the returns-earnings regression slope coefficient is greater than one (1.7). The empirical results highlight the fact that the slope coefficient, which is greater than one and increasing with the interval, accounts for much of the increment to the returns-earnings R2. In fact, constraining the slope coefficient to be one results in an R2 of 11 percent for the 10-year interval, which is considerably lower than the R2 of 47 percent when the regression is unconstrained. Hence, the positive covariance between current earnings and the accounting lag, rather than the diminishing effect of the accounting lag, appears to be the dominant explanation for the observed high R2 over long intervals.

Original languageEnglish (US)
Pages (from-to)147-164
Number of pages18
JournalContemporary Accounting Research
Volume19
Issue number1
DOIs
StatePublished - 2002

Keywords

  • Accounting lag
  • Aggregation
  • Long intervals
  • Returns-earnings regression

Fingerprint Dive into the research topics of 'The Relation between Aggregate Earnings and Security Returns over Long Intervals'. Together they form a unique fingerprint.

Cite this