We show that the vast majority of investors ignore value-relevant accruals information when it is first released, but that investors who initiate trades of at least 5,000 shares tend to transact in the proper direction. These investors trade on accruals information only when the previously-announced earnings signal is non-negative. Unconditionally, those investors initiating the smallest trades appear to respond to accruals in the wrong direction, but further investigation suggests this behavior is explained by their attraction to attention-grabbing stocks. Finally, we find that those who trade on accruals information have insufficient market power to mitigate the accruals anomaly.
The accrual anomaly is the tendency for stock prices to lag information in firms' accruals levels. Specifically, when the accruals component of a firm's earnings is relatively high (low), future returns tend to be relatively low (high). Sloan (1996) shows that the accruals component of earnings is less persistent than the cash flow component of earnings and that investors apparently do not fully appreciate this difference. Given the importance of accruals levels in predicting future returns, knowledgeable, active investors should transact immediately upon receiving accruals information. The primary research question of this paper is, therefore, who, if anyone, reacts to accrual information when it first becomes available? Specifically, we examine whether any investors respond to accruals information upon the filing of the firm's 10-K/Q. We find circumstances under which investors who initiate very large trades (at least 5,000 shares) act almost immediately to exploit the information in accrual levels, while those who initiate very small trades (fewer than 500 shares) seem to react in the wrong direction. Investors in intermediate trade-size classes appear to completely ignore accruals information. While we document that a statistically significant fraction of one subset of investors acts immediately to take advantage of accruals information, perhaps as interesting is that the vast majority of traders in each size category behave as if they are unaware of the release of accruals information and/or the implications of accruals for future price movements.
We also find that the large-trader response to accruals information is restricted to cases where the previously announced earnings news for the quarter is non-negative. In other words, large traders seem to ignore the information in accruals following negative earnings surprises. Given that abnormal returns following negative earnings surprises are non-positive (Collins and Hribar, 2000), even for low accrual stocks, this behavior is predictable in light of the inability or unwillingness of most mutual funds to go short. Short sales restrictions may take the form of the legal prohibition faced by mutual funds during most of our sample period or contractual restrictions that often appear in institutional charters. For long-only investors, the accrual signal is not important for those firms previously experiencing a negative earnings surprise and it is, therefore, rational for them to ignore it. While there may be other explanations, the data are clear. Accruals-based trading (in the correct direction) is detectable only for those investors initiating large trades and only for firms whose previously announced earnings meet or beat analysts' forecasts.
While investors who initiate trades greater than 5,000 shares exhibit a statistically significant propensity to respond to accruals information, this tendency is not nearly large enough to eliminate or even significantly reduce the magnitude of the anomaly. Specifically, we show that when large traders are correct on average, the immediate stock price response to accruals information is greater and the anomaly is smaller. But, on net, the aggregate actions of large traders do not increase the immediate stock price response to accruals information or reduce the magnitude of the drift. We interpret these results as indicating that, while there exists a statistically significant subset of investors initiating large trades who are cognizant of the release and importance of accruals information, this group is too small to have an economically significant impact on prices. The vast majority of investors (across all trade-size categories), who trade at the time of the 10-K/Q filing, trade for reasons unrelated to accruals levels. In other words, while we document that some investors transact immediately in an attempt to profit from accruals information—and we can identify those investors as among those initiating trades of at least 5,000 shares—their influence is insufficient to significantly affect the magnitude of the accrual anomaly. This observation, along with the finding that the traders in all other size categories exhibit no awareness of the accrual signal sheds light on why the accrual anomaly has proven to be such a persistent phenomenon.
Investors who initiate the smallest trades—fewer than 500 shares—seem to trade in the wrong direction at the time of the accruals signal, but we find this result is weakened substantially and rendered insignificant when controlling for two variables suggested by Barber and Odean (2008) and two additional variables that we propose. Specifically, Barber and Odean hypothesize that individual investors buy stocks that exhibit the attention-grabbing characteristics of large absolute returns and high levels of trading volume. We further hypothesize that by examining large-investor trading behavior around the filing date and small-trader buying behavior around the prior earnings announcement, we can better determine which stocks (or types of stocks) are likely to be purchased by small-traders around the filing date. As predictors of small-trader buying behavior around the 10-K/Q filing date, we find strong support for all four variables. Inclusion of these variables renders the effect of accruals economically small and statistically insignificant. These results suggest that the unconditional positive correlation between small-trader purchases and accruals is probably driven by the attention-grabbing behavior of individual investors as documented by Barber and Odean.
This paper, therefore, contributes to the literature in a number of ways. Because it is the first to examine trading behavior immediately following 10-K/Q filing dates, we believe it represents the most convincing evidence to date that any subset of investors exhibits a significant tendency to trade specifically in response to accrual information. Further, we identify those investors as ones with the resources to trade at least 5,000 shares of a single stock at once—almost assuredly institutions—and show that they trade almost immediately in response to accruals. In contrast, we show that investors in all other trade-size groups behave as if accruals information is not important. We show that the tendency for large traders to react to accruals information is insufficient to eliminate or measurably reduce the accrual anomaly. Applying the logic and results of prior research, we hypothesize that those investors who respond to accrual information may do so only following non-negative earnings surprises. The data support this hypothesis. Finally, our small-trader results confirm Barber and Odean's (2008) attention-grabbing hypothesis in a new context and using different methods.
The rest of the paper is organized as follows. Section 2 reviews the relevant literature and motivates the hypotheses. Section 3 describes the sample and defines the variables. Section 4 presents the empirical results and Section 5 concludes.
Previous research documents the persistence of the accrual anomaly, the negative correlation between firms' accrual levels and subsequent abnormal returns, over several decades. This association suggests that most investors did not fully appreciate the importance of accruals levels for future operating performance (e.g., Sloan, 1996). Lev and Nissim (2006) point out that “the timeliness of institutional response to accrual information is an important issue both for assessing market efficiency and explaining the persistence of the accrual anomaly” (pp. 196–197). If accrual information is valuable to investors, as most of the accrual literature and our own tests suggest, then sophisticated investors should react at the time the accrual signal first becomes available. Given that prior research indicates investors who initiate large trades act in a much more sophisticated manner around earnings announcements than those initiating small or medium size trades, we hypothesize that large traders might, on average, react to accruals information at the time of the 10-K/Q filing date, when it first becomes available. Based on the fact that even low accrual firms fail to exhibit positive abnormal returns following negative earnings surprises, we hypothesize that, if institutions cannot or will not go short, trading in response to accruals will be stronger following non-negative earnings surprises. Results support each of these hypotheses.
Initially, we obtained results suggesting that investors who initiate the smallest trades, those less than 500 shares, respond in the opposite direction to the accrual information. Follow-up tests strongly suggest that, consistent with Barber and Odean (2008), these investors are attracted to “attention grabbing” stocks and are not acting in response to accruals.
We believe this paper represents the first direct evidence that any subset of investors responds to the information in accrual information when it first becomes available. Prior tests rely on differences in share holdings measured at least 13 weeks apart, which, we argue, cannot isolate the effects of accruals levels from the other potential explanations. Our results are consistent with those of Battalio and Mendenhall (2005) in that the same group that exhibits the most sophisticated response to earnings announcements appears to also exhibit the most sophisticated response to accruals.
Finally, although when large traders act correctly, they tend to reduce the accrual anomaly, we do not find evidence that their behavior in aggregate has a significant effect. The limited extent to which large traders act on accruals combined with the fact that less sophisticated traders do not respond at all, contributes to the puzzling persistence of the accrual anomaly.
