GAAP Versus Street Earnings:The Role of Accounting Conservatism

JIN, Qinglu | HSU, Charles | HEFLIN, Frank

There has been a lively discussion recently within the financial world on ways in which company earnings are reported; in other words, what are the merits and otherwise of GAAP versus “Street” earnings?  GAAP provides uniformity through the application of standardized “generally accepted accounting principles” developed by the Financial Accounting and Standards Board. However, analysts, along with investors, managers and financial media, consider that GAAP earnings do not necessarily reflect a company’s core performance. Therefore, the numbers are adjusted to exclude transitory income-reducing items; the resulting Street earnings are increasingly being used in companies’ press releases and in analysts’ reports.

Researchers Frank Heflin, Charles Hsu and Qinglu Jin examined the effect of accounting conservatism on GAAP earnings and analysts’ forecasts, arguing that these effects help explain differences between GAAP and Street earnings, and create predictable differences between GAAP and Street earnings’ properties. They define accounting conservatism as book values written down under sufficiently adverse circumstances but not written up under favorable circumstances, i.e., conditional conservatism. Their basic tenet is that while both equity holders and debt holders demand conditional conservatism in GAAP earnings for various reasons – mostly to do with contracting – conditionally conservative GAAP earnings likely do not satisfy all the informational needs of investors; Street earnings help fulfil this requirement.

They analyzed earnings data within the period 1995–2009; to examine the relation between accounting conservatism and earnings persistence, earnings smoothing, and earnings informativeness, they used at least eight quarters of earnings data to estimate earnings properties and conservatism measures at the firm level.

In the study, the authors conjecture that Street earnings are, in part, motivated by the desire of investors for earnings numbers that exclude some of the consequences of conditional conservatism on GAAP earnings and analysts’ accommodation of that demand. They found that conditional conservatism was associated with larger errors in analysts’ forecasts of GAAP earnings and with more disagreement among analysts. These properties give analysts incentives to forecast less conditionally conservative earnings.

Their evidence showed that Street earnings were less conservative than GAAP earnings, and that Street earnings were more likely non-GAAP and that the magnitude of the difference between the two increases as conservatism increases. Finally, the study found that the quality of the exclusions used to calculate Street earnings increases as conservatism increases, consistent with the notion that analysts cater to investors’ needs by removing transitory items from GAAP earnings to arrive at less conservative Street earnings. The results are robust to controlling for special items but are stronger without special items, consistent with special items capturing some, but not all, of the effect of conditional conservatism. Overall, their results suggest that Street earnings are, in part, a response by analysts to investor demand for more valuation-useful information, and that analysts remove some of the consequences of GAAP conditional conservatism when forecasting earnings.

HSU, Charles