Ebook Management Forecasts, Disclosure Quality, and Market Efficiency
Management earnings forecasts are increasingly common voluntary public disclosures through which managers can influence price formation in the equity markets. Many papers have examined the short-term returns in response to management forecasts to draw inferences from investors’ immediate reaction to the forecasts (e.g., Ajinkya and Gift, 1984; Waymire 1984; Jennings, 1987; Rogers and Stocken, 2005). Although these papers show that management forecasts are informative, they do not address whether the market underreacts or overreacts to management forecasts. We address this question by examining the future long-term abnormal returns following management forecasts. In addition, we study the influence of disclosure quality on the magnitude of these returns.
The post-earnings-announcement drift (PEAD) literature provides substantial evidence that investors underreact to earnings news in a mandatory disclosure setting (e.g., Bernard and Thomas, 1989, 1990; Doyle, Lundholm, and Soliman, 2006; Livnat and Mendenhall, 2006). Management forecasts is an interesting setting to investigate whether and the extent to which the market underreacts to earnings news when the disclosure is voluntary. The existing literature on management forecasts emphasizes that there are likely to be credibility concerns arising from the voluntary and non-audited nature of the forecasts. For example, prior research have concluded that bad forecast news are more credible than bad forecast news from evidence that the magnitude of the abnormal returns around management forecasts is larger for bad news forecasts than for good news forecasts (e.g., Jennings, 1987; Skinner, 1994; Hutton, Miller, and Skinner, 2003; Rogers and Stocken, 2005, among others). Rogers and Stocken (2005), however, do not find any significant difference in the forecast bias (i.e., the difference between actual earnings and forecasted earnings) between bad news and good news forecasts. This suggests that an underreaction, if any, could be asymmetric between good news forecasts and bad news forecasts. Management forecasts also offer a unique setting to examine whether disclosure quality related to the forecasts affects any underreaction to news. This is because we can measure disclosure quality for management forecasts given that management forecasts vary in terms of precision, prior accuracy, and horizon.
In our paper, we first examine the price reaction to management forecasts using concurrent short-term (three days around the forecast) and future long-term (12 months after the month of the forecast) return windows. In the short-term, the abnormal returns per dollar of bad news are at least six times larger than the abnormal returns per dollar of good news, supporting the argument that investors perceive good news as being less credible than bad news. The future long-term abnormal returns, however, are muted for each dollar of bad news but positive and significant for each dollar of good news. This suggests that the market underreacts more to good news than to bad news around the forecasts and subsequent market correction occurs in the 12 months following the month of the forecasts.
We then investigate the magnitude of the market underreaction to management forecasts by examining the abnormal returns to hedge portfolio strategies. A hedge portfolio buying (selling) firms in the top (bottom) quintile of the distribution of management forecast surprises earns annual size-adjusted returns of 23.77%. Such economically significant returns are robust to alternative asset pricing models. In our profitability analyses, we find that the deduction of transaction costs, estimated using bidask spreads, does not reduce the net returns from the trading strategy significantly.
Further, the hedge returns are mainly driven by the buy-side of the trading strategy; firms reporting extreme good news (the top quintile) earn size-adjusted returns of 19%, whereas firms reporting extreme bad news (the bottom quintile) earn size-adjusted returns of -5% over the next year.
Finally, we investigate whether disclosure quality, measured as forecast precision, prior forecast accuracy, and forecast horizon, affects the efficiency with which investors respond to management forecasts. We find that the hedge portfolio returns are smaller for firms issuing more precise forecast and for firms with higher prior forecast accuracy. This result contributes to the literature on market efficiency by showing that higher disclosure quality is associated with a smaller underrreaction to news. Our finding on the association between disclosure quality and the magnitude of the post-management-forecast drift is conceptually related to the post-earnings announcement-drift being smaller among firms with high quality accruals (Francis, LaFond, Olsson, and Schipper, 2005) and among firms that host conference calls (Kimbrough, 2005). It is also related to the accrual anomaly being smaller for more reliable accruals (Richardson, Sloan, Soliman, and Tuna, 2005).
We also investigate whether the hedge portfolio returns covary with the determinants of PEAD such as firm size (Bernard and Thomas, 1989, 1990), investor sophistication (Bartov, Radhakrishnan, and Krinsky, 2000), arbitrage risk (Mendenhall, 2004), and arbitrage cost (Bhushan, 1994; Ng, Rusticus, and Verdi, 2006). In multivariate analyses using Fama-MacBeth regressions, we find that firm size, arbitrage risk, and prior forecast accuracy are the main determinants of the cross-sectional variation in the PMFD.
Our results are consistent with the behavioral finance theory of conservatism, which assumes that investors with complete information are irrational due to intrinsic psychological biases (Barberis, Shleifer, and Vishny, 1998). This theory is based on the earlier work in psychology by Edwards (1968) who documents that people, when faced with new information, are conservative in updating their beliefs. That is, they update their beliefs in the right direction but not in full magnitude compared to a rational Bayesian. To the extent that investors’ conservatism is a function of the credibility of news, the underreaction to management forecasts can be attributed to the lack of credibility of voluntary disclosures. Alternatively, the results are also consistent with the rational structural uncertainty theory which predicts that rational investors with incomplete information place less weight on signals in which there is larger structural uncertainty (Brav and Heaton, 2002). Due to the similar mathematical properties and empirical implications of the conservatism theory and the rational structural uncertainty theory, these theories are hard to distinguish empirically (we discuss these theories in more detail in Section 2). We note, however, that the asymmetric underreaction to good versus bad news appears to be more consistent with the argument of exacerbated conservatism for less credible good news disclosures.
Though theories on asset pricing generally do not predict the magnitude of the abnormal returns or appropriate risk premia, the economically large abnormal returns, coupled with evidence of low transaction costs, make our results interesting yet puzzling. One plausible explanation based on the conservatism theory is that investors suffer from significant conservatism bias due to the credibility concerns related to voluntary nature of management forecasts. Alternatively, our results could be explained by omitted risk factors that we have not controlled for. To address this concern, we show that our abnormal returns are robust to alternative empirical asset pricing models that are used in the literature. It is possible that this economically significant underreaction has not been arbitraged away in the past due to the lack of prior systematic evidence of an underreaction related to management forecasts. As emphasized in Merton (1987), Lee (2001), Brav and Heaton (2002), and many others, knowledge of market inefficiency drives arbitrage activities and even when market inefficiency is uncovered, it may take time for the information to diffuse and be acted upon.
The remainder of the paper proceeds as follows. Section 2 develops the hypotheses. Section 3 describes our data and variable measurement and Section 4 discusses the results. Section 5 concludes.
Download
PDF Ebook Management Forecasts, Disclosure Quality, and Market Efficiency
Posted in :
