Do Analysts Correct the Market’s Mispricing of Accruals? via CSInvesting
Yong Keun Yoo
Abstract
Elgers, Lo and Pfeiffer (2003) argue that the bias of analysts’ earnings forecasts is significantly less than the bias of market’s earnings expectations in interpreting accruals. Their argument implies that analysts’ earnings forecasts could potentially mitigate the market’s mispricing of accruals by guiding investors to reduce their earnings prediction errors arising from the misinterpretation of accruals. However, their results call for further investigation owing to the following two questionable research design choices: 1) estimating the magnitude of the market’s bias by using the framework of the Mishkin test based on the earnings response coefficient model, which is vulnerable to the well-known omitted-variable problem; 2) comparing only the bias of one-year-ahead earnings expectations, while ignoring the bias of earnings expectations for longer periods. By taking an alternative approach to address these issues, I find that analysts’ earnings forecasts are more biased than stock prices in interpreting accruals. Thus, contrary to Elgers et al. (2003), I conclude that analysts’ earnings forecasts do not mitigate the market’s mispricing of accruals.
Do Analysts Correct The Market’s Mispricing Of Accruals? – Introduction
This study examines whether analysts’ earnings forecasts mitigate the market’s mispricing of accruals. Since Sloan (1996), accounting research has concluded that investors fail to fully consider the lower persistence of accruals when they are predicting future earnings. The resulting earnings prediction errors lead to predictable stock returns.
This empirical regularity is named the “accrual anomaly”. The accrual anomaly is considered one of the most prominent pieces of evidence of market inefficiency documented in accounting literature. Given the market’s mispricing of accruals, it is of fundamental importance to evaluate which factors mitigate or exacerbate the market’s mispricing. This is the case because the stock price’s deviation from the intrinsic value is a direct challenge to the economically efficient allocation of funds among competing stocks.
This study’s main motivation for evaluating sell-side analysts is that they are the most prominent information intermediaries in the capital market. They receive and process information from diverse sources, communicating it to investors in such concise forms as earnings forecasts and stock recommendations. In particular, analysts’ earnings forecasts can have a large influence upon the market’s earnings expectations by serving as a publicly available and easily accessible benchmark. Thus, it is worthwhile to examine whether analysts’ earnings forecasts either facilitate or impede more accurate pricing of accruals, doing so by influencing the formation of the market’s earnings expectations.’
Elgers, Lo and Pfeiffer (2003) conclude that the bias of market’s earnings expectations in interpreting accruals significantly exceeds the bias of analysts’ earnings forecasts. They suggest that investors forego the opportunity to mitigate their mispricing of accruals by fixating on analysts’ earnings forecasts, which are relatively less biased in interpreting accruals. Such an argument implies that analysts’ earnings forecasts have a potential to mitigate the market’s mispricing of accruals by guiding investors to reduce their earnings prediction errors arising out of the misinterpretation of accruals.
However, the study of Elgers et al. (2003) suffers from some misspecification problems. First, Elgers et al. (2003) compare the proxy of the unobservable market’s earnings expectations with analysts’ earnings forecasts. They estimate the market’s weightings of current accruals for the prediction of one-year-ahead earnings by regressing one-year-ahead stock returns on one-year-ahead unexpected earnings within the framework of the Mishkin test (Mishkin, 1983), which is based on the earnings response coefficient model (hereafter, the ERC model). They, then, show that the market’s over-weightings of accruals significantly exceed the analysts’ over-weightings.
However, the well-known omitted-variable problem of the ERC model causes them to overestimate the market’s weightings of accruals. When the accrual anomaly holds, accruals are negatively correlated with the market’s revisions of earnings expectations beyond one year ahead. Since the ERC model excludes the market’s revisions of earnings expectations beyond one year ahead, the estimated weightings of accruals are inevitably inflated.
Second, Elgers et al. (2003) show only that the market’s one-year-ahead earnings expectations are more biased than analysts’ one-year-ahead earnings forecasts. However, such a result does not necessarily mean that the market’s earnings expectations beyond one year ahead are more biased than analysts’ corresponding earnings forecasts. Since firm valuations are significantly affected by both one-year-ahead and longer-term earnings expectations, the relative bias of the market’s earnings expectations beyond one year ahead to the analysts’ corresponding earnings forecasts should have been examined.
See full PDF below.
