The post earnings announcement drift (henceforth PEAD) has been the most resilient and widely researched phenomenon since it was first noted over four decades ago by Ball and Brown (1968). This delay in prices assimilating earnings information poses a challenge to the semi-strong form of market efficiency which assumes efficient incorporation of new publicly available information into stock prices (Lui et al 2003). A number of explanations have been put forward to explain the PEAD phenomenon. Bernard and Thomas (1989) explained the phenomenon by two competing theories; 1) failure to adjust abnormal returns fully for risk (inadequacy of the CAPM) and 2) delay in response to earnings reports either due to failure by the investors to assimilate available information and/or that costs involved limits immediate exploitation of earnings information. Earlier academic research, on this phenomenon was based on the US market where PEAD was first observed, but eventually research was extended to other markets such as UK (Lui et al 2003), Spain( Forner et al, 2009) and New Zealand ( Troung, 2010). The findings across markets variedmarkets varied due to methodological issues, risk pricing models, behavioural models as well as country specific factors. This essay seeks to summarise and critically discuss the findings and explanations given by researchers on different markets about the anomaly drawing particular attention to methodological issues, risk pricings models and behavioural aspects.
2 Methodological Issues (International Comparisons)
The way in which data is collected matters because it may lead to methodological problems and variance in results. Table 1 summarises the methodology employed in academic papers across different market. US and New Zealand papers used event time to track stock and construct portfolios, this method is subject to look-ahead bias and the requirement of cutting off values using deciles in early research design could be misleading. UK and Spain studies applied a calendar time method to fix these problems and provided an easy way of implementing investment strategies. Table 1 here? Earnings measurement variables are decided after the data collection. Though SUE probably is the most commonly used measurement for earnings, it can be calculated in very different ways in different circumstances. The selected US and UK papers standardize the unexpected earnings by its prior forecast errors, this adjustment allowed comparison of firms with different volatilities. Spanish study scaled the unexpected earnings with book value of equity at beginning of the year and New Zealand paper scale earnings with stock price prior the announcement to address the mispricing problem. Analyst forecast based measurement is also widely used, though this approach gives an investor a more timely comparison, biased analyst forecasts is always a serious drawback. Traditional method is to track the revision of analyst forecasts over time. In contrast, the New Zealand study tracks the gap between actual earnings and forecasted earnings presenting a new way to partially solve this problem. Studies across four countries all tried to explain the post earning drift with various risk factors. Original CAPM, Fama French three factor models and conditional versions of them are commonly used to address systematic risk’s impact on earnings. US study tested the significance of the overall economic influence using Chen Roll and Ross 1986 five factor model. Market macrostructure factors such as market value, firm size and book to market ratio are widely used and implied ways to track firm’s specific characteristics. As evident from the US and New Zealand papers, factors and models which are irrelevant to risk such as investors’ behaviour, momentum effect and transaction costs are continuously studied by researches. In Spain the investor conservatism bias model was applied along with the overreaction, the underreaction and the gradual-information diffusion model in order to explain post earnings drift from the behavioural finance perspective. Although researches conducted in these four countries have some similarities, different variables, procedures, and tests were applied to the area which is important from a researcher’s point of view. Due to differences in culture, institutional characteristic and market efficiency, it’s hard to standardize methodology and a claim cannot be made on which measurement is the most accurate..
3 Test results for risk related models and factors
Risk related models tried to regard post earnings drift as compensation for risk in the efficient market. The inadequacy of CAPM and other asset pricing models is widely tested worldwide. Differences in market depth, data limitation, errors of methodology and risk misspecification can result in inconsistent outcome across countries. Moreover, results can differ with the risk model prediction. Forner (2008) found a negative relationship between CAPM beta and PEAD profits in Spain while other countries showed a positive sign. In Forner’s study, arbitrage risk is positively related to the earning profits, which is contradicting to Mendenhall’s (2004) conclusion based on US sample. Though results vary worldwide, most researchers conclude that risk is insignificant on return difference, and neither the systematic risk nor other risk factors are able to explain the PEAD phenomenon separately or jointly. PEAD effectively remains after the risk adjustment and in order to fully understand and explain the PEAD phenomenon, factors other than risks must be considered.
4 Behavioural Factors
Based on US data Bernard and Thomas (1989) found that a large proportion of the drift was due to investors’ adopting a naive assumption that earnings follow seasonal random walk and failing to understand the correlation between current earnings and future quarterly earnings. The claim of naivety was refuted by Bartov et al (2000) who observed that the drift is smaller in firms that were largely held by institutional investors indicating more sophisticated investors than claimed by Bernard and Thomas. However, almost all investors are guided to some degree by psychological biases regardless of whether they are individual or institutional investors. Barberies et al (1998) posit that investors exhibit conservative behaviour and representative bias that leads to underreaction to earnings announcement. This suggests that investors seem to be conservative and do not act immediately to earnings news on the premise that the positive earnings growth is temporary and that it will eventually revert. For institutional investor, the delayed response probably can be explained by rule which put constrains on arbitrage and insider trading. Daniel et al (1998) departing from the underreaction based explanation hypothesizes that investors experience a combination of overconfidence and self attribution bias and tend to exhibit patterns of overreaction to privately sourced information which generate short-term momentum. On the other hand, underreaction to publicly obtained information will produce a long-term reversion in stock return. However, when this model was tested in Spain where the momentum effect is weak, there was no significant evidence showing the relationship between overconfidence and overreaction. This is because Spanish culture is regarded as less individualistic and people give more weight on public information thus exhibiting a low level of self attribution and overconfidence (Forner, 2009). Hong et al (1999) theory on the other hand presented a significantly different theory based on interactions between heterogeneous agents namely the “newswatchers” and “momentum traders” and observed gradual diffusion of firm specific information which lead to initial underreaction and subsequent long term overreaction. Unlike in the US, the undereaction in Spain is not significant. This is due to the code law legal system and low level of investor protection, insiders have less limits to trade on the information they obtained and the price adjust quickly which weaken the momentum effect (Forner, 2009)
Post earnings announcement drift phenomenon was observed across different markets since 1968 and although many academic studies were conducted and lots of trading strategies were applied, the anomaly persists. Different methodology is adapted across countries to measure the magnitude of post earning drift. Though more and more proxies are used, each measurement has its own shortcomings. Differences in market efficiency, market depth and transparency and accounting regimes may lead to biased measurement starting from the data collection stage. Risk factor models try to explain the post earnings announcement drift under efficient market hypothesis and the effectiveness of model is largely dependent on market specific features. Difference in media coverage could influence the information dispersion and delay investors’ response to share price. Return is also affected by country specific macroeconomic indicators, market size, regulation on insider trading and speculative activity. Even in US and UK where the markets are more efficient compared to other countries, the risk models failed to fully explain the drift. Behavioural approaches address the persistence of drift from a different perspective based on investors’ sentiments, heuristics bias and limits to arbitrage. Nevertheless, investors’ behaviour varies due to cultural differences and their sophistication level. Although most investors may suffer from overconfidence bias, investors in highly individualistic cultures may react to earning announcement differently than those from low individualistic culture. Limits to arbitrage put constrain on investors’ early exploration of earnings announcement and increase the difficulties in correcting the anomaly. Because of the differences in market and behavioural factors, no standardize approach can be used. Therefore, strategies aimed to exploit benefit from the post earnings announcement drift should take into account countries’ specific features.
Ball, R. and Brown, P. (1968) An empirical evaluation of accounting income numbers, Journal of Accounting Research, 6, pp. 159-178 Barberis, N., Shleifer, A. and Vishny, R. (1998) A model of investor sentiment, Journal of Financial Economics, 49, pp. 307-343 Bartov, E., Radhakrishnan, S., Krinsky, I., 2000. Investor sophistication and patterns in stock returns after earnings announcements. The Accounting Review 75, 43-63. Bernard,V., K.Thomas., (1989), “Post Earnings Announcement Drift: Delayed Price Response or Risk Premium?” Journal of Accounting Research, Vol.27, pp.1-36 Cameron Truong, Post earnings announcement drift and the roles of drift-enhanced factors in New Zealand, Pacific-Basin Finance Journal, Volume 18, Issue 2, April 2010, Pages 139-157 Chan, L., Jegadeesh, N. and Lakonishok, J. (1996) Momentum strategies, Journal of Finance, 51(5), pp. 1681-1713. Daniel, K., Hirshleifer, D. and Subrahmanyam, A. (1998) Investor psychology and security market under- and overreactions, Journal of Finance, 53(6), pp. 1839-1885. Forner, C., Sanabria, S. and Marhuenda, J. (2009) Post-earnings announcement drift: Spanish evidence, Spanish Economic Review, 11(3), pp. 207-241. Forner, Carlos and Sanabria, Sonia ((November 1, 2009): Post-Earnings Announcement Drift in Spain and Behavioural Finance Models European Accounting Review Forthcoming. Available at SSRN: https://ssrn.com/abstract=1498029 Foster, George, Olsen, Chris, Shevlin, Terry, 1984. Earnings releases, anomalies, and the behaviour of security returns. The Accounting Review 59 (4), 574-603 Hong, H. and Stein, C. (1999) A unified theory of underreaction, momentum trading, and overreaction in asset markets, Journal of Finance, 54(6), pp. 2143-2184 Liu, Weimin, Strong, Norman, Xu, Xinzhong, 2003. Post-earnings-announcement drift in the UK. European Financial Management vol 9(1), 89-116. Richard R. Mendenhall, 2004. Arbitrage Risk and Post-Earnings-Announcement Drift,” Journal of Business, pp. 875-894 ——————————————————————————–
Foster et al (1984) also concurs with this reason
Ball & Brown(1968), Foster et et al(1984), Bernard & Thomas(1989)