Effects of Growth Options on Post-Earnings Announcement Drift

Transcription

1 Effects of Growth Options on Post-Earnings Announcement Drift Abstract As the longest anomaly in the finance literature, post-earnings announcement drift (PEAD) continues to exist and challenges the efficient markets hypothesis. This paper is the first to provide evidence that growth options have effects on the PEAD. I use different proxies to represent firms growth options, and find that PEAD is more likely to survive for firms with low growth options represented by ratio of market-to-book assets. A firm with low growth options has high risk because of high percentage of assets-in-place, resulting in an upward drift after the earnings announcement. On the other hand, a firm with high growth options is more likely to have a downward drift after the earnings announcement. This negative relation is independent on earnings surprise, which loses the power of predicting the direction of drifts for firms with high growth options, and has direct impact on calculation of abnormal returns for PEAD. JEL Classification: G14 Keywords: Anomaly, Post-earnings announcement drift, growth options 1

2 Effects of Growth Options on Post-Earnings Announcement Drifts Abstract As the longest anomaly in the finance literature, post-earnings announcement drift (PEAD) continues to exist and challenges the efficient markets hypothesis. This paper is the first to provide evidence that growth options have effects on the PEAD. I use different proxies to represent firms growth options, and find that PEAD is more likely to survive for firms with low growth options represented by ratio of market-to-book assets. A firm with low growth options has high risk because of high percentage of assets-in-place, resulting in an upward drift after the earnings announcement. On the other hand, a firm with high growth options is more likely to have a downward drift after the earnings announcement. This negative relation is independent on earnings surprise, which loses the power of predicting the direction of drifts for firms with high growth options, and has direct impact on calculation of abnormal returns for PEAD. 2

3 I Introduction Ball and Brown (1968) show that following a positive (negative) earnings surprise a firm has an upward (downward) cumulative abnormal returns, i.e. a post-earnings announcement drift (PEAD). Since this first description of the PEAD, it continues to exist as an anomaly evidenced by Konchitchki et al. (2010). Shivakumar (2007) addresses this as the longest standing anomaly in the finance and accounting literature, and Kothari (2001) concludes that the PEAD anomaly poses a serious challenge to the efficient markets hypothesis. There are different explanations for the PEAD. Bernard and Thomas (1989, 1990) show two main possibilities: the first is the result of information delay, which comes either from failure of market to integrate available information or from market friction where cost exceeds gain for traders to profit by exploiting the information. The second possibility is the consequence of failure of CAPM to capture the real risks used for calculating subsequent cumulative abnormal returns. This implies that companies with good news are inherently more risky than those with bad news. Further studies from Mendenhall (2004) suggest that there is a high cost of arbitrage in exploiting this phenomena, while Sadka (2006) raises the attention of liquidityrelated risks in arbitraging away the PEAD. Recently Da, Guo, and Jagannathan (2012) show that real options adjustment can eliminate several anomalies, indicating the existence of other risks that can t be captured by current asset pricing models. In this paper I exam the impact of growth options on the PEAD, and provide evidence that: 1) market-to-book assets (MABA) is one of the best proxies for growth options to 3

4 address their effects on PEAD; 2) there are downward drifts for firms with high growth options and upward drifts for firms with low growth options; 3) PEAD is more likely to survive for firms with low growth options, and that earnings surprise loses the power of predicting the direction of drifts for firms with high growth options. The results demonstrate that growth options have significant effects on the post-earnings announcement drifts. The remainder of the paper is structured as follows: first in section 2 there is a review of PEAD and growth options. Section 3 illuminates the data and methodology used in the study. Section 4 reports and discusses the empirical results, including robustness check. Section 5 summarizes the findings from the paper and provide directions for future research. II Literature review Post-earnings announcement drifts Consistent with one of the explanations of Bernard and Thomas (1989, 1990), Mendenhall (1991), Abarbanell and Bernard (1992), Bartov (1992), Ball and Bartov (1996) argue that investors underreact to earnings surprises and fail to correctly understand the implication of current earnings for future earnings, and propose that PEAD is a further adjustment of price to reflect the under-reaction of market participants at the beginning of the earnings announcement. Daniel (1998) and Fischer (2001) demonstrate that the under-reaction is due to investors overconfidence about their private and heterogeneous information. This overconfidence of the private information held by market participants, which is different from that reflected in public earnings announcements, results in the drift by overweighting the heterogeneous private 4

5 information and underweighting the public information. This implies that PEAD for firms with the most heterogeneous private information will be the strongest. Bartov et al. (2000) find a negative relationship between multitude of PEAD and percentage of institutional holdings, since naive investors are more likely to fail in recognizing the full implication for future earnings and they are the driving forces for PEAD. Bhushan (1994) also points out that transaction cost can play a role and it prevents market participants from making money on PEAD. On the other hand, Ball (1978), Foster, Olsen and Shevlin (1984), Ball, Kothari and Watts (1993), and Barnard and Seyhun (1997) argue that post-earnings announcement drifts are due to failure of CAPM to capture the real risks when calculating subsequent cumulative abnormal returns. Recently uncertainty is proposed by Zhang (2006), Francis, LaFond, Olsson and Schipper (2007), Anderson, Harris and So (2007), Angelini and Guazzarotti (2010) to explain the post-earnings announcement drifts. They shows that uncertainty of the quality of information signal can result in investor s under-reacttion to the earnings announcement. When there is more uncertainty, the magnitude of the post-earnings announcement drifts is larger. Furthermore, Bird and Yeung (2012) demonstrate that PEAD depends on both the level of uncertainty and investor s sentiment during the post-earnings announcement period. Growth options As Myers (1977) describes, a firm can be considered as a portfolio of assets already in place and future growth options. Berk, Green, and Naik (1999) indicate that expected returns depend on the average systematic risk of assets-in-place, the number and value of growth options, and interest rate. The risk exposure of a firm s equity depends on the weight of assets in place and growth 5

6 options, which dynamically adjust to optimal investment decisions over time. They demonstrate that expected return equation can be represented by size and book-to-market equity. Both factors change over time due to firms dynamically optimization of investment opportunities. Since growth options are unobservabl, the best way to capture them is to use proxies. Adam and Goyal (2008) show that market-to-book assets (MABA) ratio is one of the best proxies for investment opportunities and it is relatively independent from other factors. Here book value of assets represents assets in place and market value of assets represents total assets including assets in place and growth options. MABA ratio integrates the market value and captures the market s anticipation of a firm s growth opportunities in future. Specifically, MABA ratio integrates information about a firm s risk relative to the scale of total assets. A firm s market value is contingent on the firm s asset portfolio along its investment cycle. The market value of a firm increases when the firm invests in projects with low systematic risk. When the firm invest in these opportunities, its cash flow is reduced in the subsequent time horizon, on average resulting in a low realized returns. Carlson, Fischer and Giammarino (2004) demonstrate that a firm's systematic risks vary due to the exercise of growth options. Anderson and Garcia-Feijóo (2006) show that portfolios based on investment-growth rates have higher subsequently monthly returns when firms accelerates investment spending. 6

7 III Methodology and data Event study and the Fama-French-Momentum Model I use Eventus to study post-earnings announcement drifts (Prabhala 1997). The model to measure abnormal return is based on 4-factor model, which includes Fama French 3-factor model (Fama and French 1993) and momentum factor (Carhart 1997). The 4-factor model utilizes the risk premium on the market portfolio, size (SMB), book to market ratio (HML) and momentum (UMD) as determinants of asset returns. The return model has the form of R it R ft = α i + β1 i( Rmt R ft ) + β2ismbt + β3ihmli + β4iumdi + εit... (1) where R it is the return for common stock of firm j on day t. Three-month T-bill is used as a proxy for risk free rate of return on day t. R mt is the return on day t for the market proxy of equallyweighted S&P 500. SMB t is the average return on small market-capitalization portfolios minus the average return on large market-capitalization portfolios; HML t is the average return on high book-to-market equity portfolios minus the average return on low book-to-market equity portfolios; UMD t is Carhart s price-momentum factor based on 12-month momentum in returns, which is equal to the average return on high prior 12-month return portfolios minus the average return on low prior12-month return portfolios. α i is the intercept of the regression measured by ordinary least squares method. β 1i, β 2i, β 3i, and β 4i are coefficients to measure the sensitivity of R it to corresponding different factors. And ε it is the error term for firm i at time t with an expected value of zero and is uncorrelated with R mt. The data source for the 4-factor model and R it is from the Center for Research on Security Prices (CRSP) at the University of Chicago. 7

8 Event study and the abnormal returns Abnormal returns (AR it ) is computed as the difference between actual returns and the expected returns using above four-factor model: AR it = R it [( ˆ α + ˆ β ( R R ) + ˆ β SMB + ˆ β HML + ˆ β UMD ]... (2) E[ Rit ] = Rit R ft i i mt f 2i t 3i t 4 1 t i t where AR it, R it and E(R it ) are the abnormal, real, and expected returns, respectively, for announcement i and event window t. In this paper I mainly use buy-and-hold abnormal returns (BHAR) to measure abnormal returns during post earnings announcement periods, where BHAR is the difference between the actual compound return during the time period [t 1, t 2 ] and its corresponding expected return without the earnings announcement: BHAR i t 2 = (1+ R ) t= t1 t E i t= t 2 ( (1+ Ri t )) (3) t1 where t 1 and t 2 are, respectively, the first and the last trading day of the target window over which daily return, R it, is compounded. For example, t 1 = 0 means a window beginning at the earnings announcement day. E(.) is the expected return without an announcement event happened. The average abnormal returns for N firms on day t is calculated as : 1 AAR = N N t AR it i= 1... (4) 8

9 where AAR t is the average (abnormal) returns, and the cumulative average abnormal returns (CAR i ) in the window [t 1, t 2 ] is calculated as follows: CAR t 2 = = t i AR it i= t1... (5) Estimation of earnings surprise Following prior studies (Livnat and mendenhall 2006), I use a rolling seasonal random walk model. The time-series measure of standardized unexpected earnings (SUE1) is expressed as the following equation: SUE1 j, t = E j, t E P j, t j, t 4... (6) where E j,t is primary earnings per share (EPS) before extraordinary items for firm j in quarter t, and P i,t is the price per share for firm j at the end of quarter t from Compustat. In order to be consistent with E j,t, E j,t-4 is adjusted for any stock splits and stock dividends during the period (t 4, t ). EPS are based on Compustat s diluted EPS data. Analysts forecasts of earnings from I/B/E/S is used as the second measure of earnings surprise. The standardized unexpected earnings using analyst forecasts (SUE3) is defined as: SUE3 j, t = E IBES j, t P E j, t AF j, t... (7) where E, is the actual EPS reported in I/B/E/S and IBES j t E, is the mean of the analysts quarterly AF j t forecasts of EPS during the 90-day period prior to the disclosure of the actual earnings. The earnings surprise is also scaled by price per share for firm j at the end of quarter t. Base on formula 8, standardized unexpected sales (SUS) are calculated as: 9

10 Where IBES j t SUS S IBES AF j, t j, t j, t = AF S S S, is the actual sales reported in I/B/E/S and j, t... (8) S, is the mean of the analysts AF j t quarterly forecasts of sales during 90-day period prior to the disclosure of the actual earnings. The sales surprise is also scaled by the mean of the analysts quarterly forecasts of sales. Proxies of growthl options Following Cao and Zhao (2008), I use four proxies for growth options, which includes the ratio of the market value to book value of assets ( MABA), the ratio of capital expenditures to fixed assets (CAPEX), the ratio of debt to equity ratio (DTE), and return on asset (ROA). All these proxies are calculated using data from Compustat quarterly file. Appendix A provides more details about the calculation. The ratio of MABA integrates the market value and capture the market s anticipation of a firm s growth opportunities in future. Another popular proxy for growth options is Tobin s Q. These two proxies are widely used as growth options, such as in the articles of Collins Cothari (1989), chung charenowong (1991), Smith and Watts (1992), Berk, Green, and Naik (1999), Goyal, Lehn and Racic (2002), Carlson, Fisher, and Giammarino (2004), and Anderson and Garcia- Feijόo (2006). Here I only use MABA since my tests show that these two different proxies are quite similar. 10

11 DTE and CAPEX distinguish from above two proxies by just using data from the book value. Firms with high DTE have high financial leverage and there is a high possibility for these firms to have financial distress. CAPEX is used as a proxy for growth options because high capital expenditures imply that there are more on-going investment opportunities for a firm. Data selection I use Institutional-Brokers-Estimate-System summary statistics files (I/B/E/S) to gather information about quarterly earnings per share (EPS), announcement dates, and the mean analyst forecasts. I also use returns provided by the Center for Research on Security Prices (CRSP) at the University of Chicago. Stock splits and stock dividend have been adjusted so that all returns and earnings are aligned on a comparable basis. All the returns of benchmark directly come from Kenneth French s website. The sample period is between June 1998 and December 2011, since I also include data of sales surprises (SUS), which begins at 1998 and is calculated in a similar way as the standardized unexpected earnings using analyst forecasts (SUE3). In order to compare between time-series unexpected earnings and those measured by analyst forecasts, the company should at least has one analyst forecast from I/B/E/S database. And the company s shares are traded on the New York Stock Exchange, American Stock Exchange, or NASDAQ. The sample excludes companies with market value of equity less than $5 million at the end of quarter t-1, and also excludes companies with price less than $1 in Compustat. 11

12 IV Empirical results Descriptive statistics Descriptive statistics is summarized in table 1, which includes various proxies of grwoth options, earnings surprises, sales surprise, firm s characteristics such as book value and market value, and abnormal returns during earnings announcement period and post-earnings announcement periods. There are 81,805 firm quarter observations between the 4 th quarter of 1998 and the 2 nd quarter of Spearman correlations of relevant variables is displayed in table 2. First, the matrix shows that correlations between different proxies of growth options are significant. For example, there is a % correlation between MABA and DTE, and a 39.48% correlation between MABA and ROA. The matrix also includes MEBE, market-to-book equity ratio, to compare with MABA. The correlation between these two proxies is quite high (90.55%). Correlation between MEBE and DTE (-39.75%) is smaller than that between MABA and DTE (-63.58%). Correlations among SUE1, SUE3 and SUS are significant and positive. Compared with SUE1 (11.40%) and SUS (19.43%), SUE3 has a stronger correlation (31.69%) with the buy-and-hold abnormal returns during earnings announcement period (-1, +1). This shows that abnormal returns during the announcement window strongly correlate with earnings surprises measured by the method of analyst forecasts. 12

13 The magnitude of correlations between buy-and-hold abnormal returns and proxies of growth options (except for ROA) increases as time shifts from the earnings announcement period to post-earnings announcement periods. The correlation coefficient between MABA and BHAR changes from -2.29%, -9.62%, % to % during windows of (-1, +1), (+2, +22), (+2, +44), and (+2, +63), respectively. On the other hand, correlations between earnings (or sales) surprises and buy-and-hold abnormal returns decrease quickly as time shifts from earnings announcement period to post-earnings announcement periods. The correlation coefficient between SUE3 and BHAR changes, respectively, from 31.69%, 2.57%, 1.70% to 1.62% during windows of (-1,+1), (+2, +22), (+2, +44), and (+2,+63). These results indicate that post-earnings announcement abnormal returns are highly related to growth options of the firms during postearnings announcement periods. Univariate regressions I exam the relation between post earnings announcement drifts and growth options using monthly fama-macbeth cross-sectional regressions (1973). Panel A, B, C and D in Table 3 provide univariate regressions of buy-and-hold abnormal returns (dependent variable) on variables including growth options and earnings/sales surprises that may drive the abnormal returns during earnings announcement (Panel A) and post earnings-announcement periods (Panel B, C and D). The coefficients in table 3 are the average of N quarterly coefficients, and T- statistics in parenthesis are based on the standard deviations of these coefficients. 13

14 Models of (1) - (5) in Panel A of Table 3 show there are significant negative relations between buy-and-hold abnormal returns and proxies of growth options (except for DTE) during earnings announcement period. The results are consistent with findings from above correlation matrix, which shows that there is a negative correlation between growth options and post earnings announcement drifts. Models of (6) - (8) in Panel A of table 3 confirms that there are positive relations between earnings surprises (or sales surprises) and buy-and-hold abnormal returns. Compared with SUE1, SUE3 has more power (R 2 =4.8% VS 2.0%) in explaining the abnormal returns during earnings announcement. This is because the method of analyst forecasts has integrated more on-going information than the time-series random walk model, which just reflects information based on historical financial statements. The result is consistent with those from Chan (1996), and Livnat and Mendenhall (2006). Although relations between abnormal returns and growth options do not change when time shifts from earnings announcement period to post-earnings announcement periods (Panel B and in Table 3), relations between abnormal returns and earnings surprises (or sales surprises) are not as robust as those between abnormal returns and growth options. The coefficients of SUE1 and SUS change from positive to negative when time shifts from earnings announcement period to post-earnings announcement periods, and at the same time SUE3 loses its explanatory power for BHAR during periods of (+2, +44) and (+2, +63) (Panel C and D in Table 3). In summary, Table 3 shows that SUE3 has the most power in explaining abnormal returns during the earnings announcement period, but it is replaced by MABA ratio during the post-earnings announcement periods. 14

15 Multiple regressions With controls for earnings surprises (and/or sales surprises), I evaluate relations between growth options and buy-and-hold abnormal returns during earnings announcement period and postearnings announcement periods by using Fama- MacBeth regressions. Panels A, B, C and D in Table 4 present Fama-MacBeth coefficients for regression of firm s buy-and-hold abnormal returns, including BHAR (0,+1), BHAR (+2,+22), BHAR (+2,+44) and BHAR (+2,+63), on growth options and earnings / sales surprises. Model 10 and model 12 in Panel A of Table 4 show that buy-and-hold abnormal returns are significantly correlated with MABA, earnings surprise (SUE3), and sales surprise (SUS) during earnings announcement period. Model 1 and model 2 test which proxy of growth options is the best one for explaining buy-and-hold abnormal returns without controls for earnings surprises, and Model 2 tests factors of earnings surprise and sales surprise under no controls for growth options. Panel B in Table 4 shows that growth options (R 2 =0.038) have more power in explaining the abnormal returns compared with factors of earnings (sales) surprise (R 2 =0.027). And R 2 increases almost linearly to when integrating both factors of growth options and earnings surprise in model 10. This indicates that the factor of growth options does not subsume the factor of earnings surprise, but it has incremental ability in explaining the post-earnings announcement drifts. Model 10 in Panel B, C and D of Table 4 shows that MABA is negatively correlated with abnormal returns during earnings-announcement period and post-earnings announcement 15

16 periods. Although both SUE3 and SUS are significantly correlated with abnormal returns during earnings announcement period (-1,+1) in model 3, SUS loses its explanatory power for abnormal returns during post-earnings announcement period of (+2,+22) and changes the sign of coefficient from positive to negative when time is extended to (+2, +44) and (+2, +63). Above analyses show that MABA, representing growth options, has consistently negative effects on abnormal returns during post-earnings announcement periods. Why MABA ratio? Why does MABA ratio stand out as a proxy of growth options and have effects on abnormal returns during post earnings announcement periods? In the ratio of MABA, the market value of assets is a proxy for combination of assets in place and investment opportunity, and book value of assets is a proxy for assets in place. A low MABA ratio implies that there are relative few investment opportunities for a firm compared with its assets in place. Table 5 clearly shows that buy-and-hold abnormal returns for portfolios during post-earnings announcement period (+2, +22) decrease when their MABA ratios increase. This implies that firms with high MABA ratios (low growth options) have lower abnormal returns during post-earnings announcement periods. The result is consistent with those from using other proxies of growth options such as CAPEX and DTE, where firms with higher CAPEX (low debt to equity ratio) on average have more investing opportunities and on-going investing projects, and hence have more growth options and lower subsequent abnormal returns. As for DTE, the first two deciles of portfolios with low DTE ratios have higher returns compared with the third decile, which has a higher debt leverage. This can be due to penalty from market for not using any debt to take advantage of tax benefits. 16

17 Table 4 shows that CAPEX, a purely accounting-based proxy, is positively related to the value of investment opportunities, the relation is not as robust as that for MABA. This is also consistent with findings from Adam and Goyal (2008). They demonstrate that compared with other proxies, MABA has the highest information content of investment opportunities, including some information that can t be captured by the ratio of capital expenditures over the net book value of plant, property, and equipment (CAPEX), and the earnings price ratio. Adam and Goyal (2008) also show that it does not outperform MABA ratio even for extracting a common factor from different investment proxies. Yang and Zhao (2011) use the ratio of book-to-market equity to distinguish glamour stocks from value stocks, and show that they have asymmetry responses to positive and negative earnings surprises, i.e., glamour stocks are more sensitive to negative earnings surprises and value stocks are more sensitive for positive earnings surprises. Penman, Richardson, and Tuna (2007) indicate that there is a one to one link between book-to-market assets (BA/MA) and book-to-market equity (BE/ME) through the equation of: BE ME = BA MA + D ME BA ( MA -1)... (7) Where D is debt, and D/ME is the same as DTE which I use in this paper. Formula 7 indicates that high value premium (high BE/ME) can come from high operating leverage (low growth options), high financial leverage, or both. Compared with MEBE, MABA is a cleaner proxy for growth options. In this paper I use MABA as a proxy for growth options. Model 12 in Table 4 17

18 indicates that MABA has more power in explaining the abnormal returns for PEAD. Table 5 also shows that a hedge portfolio formed by MABA deciles has a higher abnormal return than that from a hedge portfolio formed by MEBE deciles. Table 6 shows that effects of growth options on abnormal returns during post-earnings announcement periods do not depend on earnings surprises. Abnormal returns in different investigated windows (panel A, B, C and D in Table 6) all decrease as MABA ratio increases in different deciles of SUE3. On the other hand, Table 6 shows that there are no PEAD for firms with highest decile of MABA ratios during post-earnings announcement periods of (+2, +22), (+2, +44) and (+2, +63). Panel D in table 6 shows that SUE3 completely loses its power in explaining drifts for different level of MABA ratios during post-earnings announcement period of (+2, +63). For better understanding of the effects of both MABA and SUE3 on magnitude of abnormal returns during post earnings announcement periods, Table 7 provides abnormal returns for zeroinvestment portfolios formed by MABA and SUE3. There is an average of 5.48% hypothetical abnormal return with standard deviation of 4.96% during period of (+2,+22) if going long in firms with low MABA ratios and at the same time selling short in firms with high MABA ratios. The average hedge return for portfolios formed by SUE3 is only 1.35% with standard deviation of 3.92% in the same period. The above hedge returns are hypothetical because periods of (+2, +22) are not perfectly overlapped in reality for firms with different announcement days. In order to test post-earnings 18

19 announcement abnormal returns by using practical trading strategy, where an anomaly is prone to be eliminated if there is no arbitrate risk, I divide the whole sample into a 5X5 matrix based on MABA and SUE3 scores in each month T when earnings are announced. Then portfolios are formed at the beginning of month T+1. Table 8 shows corresponding monthly buy-and-hold abnormal returns for these portfolios. For each SUE3 quintile, its abnormal return also decreases as MABA ratio increases. The table also demonstrates that PEAD is more likely to survive for firms with low MABA ratios. Table 8 indicates that most of negative abnormal returns come from the left bottom triangle. This implies that firms with both high MABA and low SUE3 tend to have downward post-earnings announcement drifts. The downward drifts can come from two parts: the first one is because companies with high growth options have low risks compared to those with low growth options; the second one is due to the change of market participants perception of the firms value of growth options. When there is a sudden change of investor s perception of firms value in an efficient market, there is a momentum of the change of stock prices if there is no contradictory information flowing in, resulting in PEAD. On the other hand, firms with low growth options have high risks compared with those with high growth options, and market participants are likely to upgrade their perceptions of these firms value of growth options, which is relatively low during the earnings announcement. Table 8 shows that most of positive abnormal returns come from the right top triangle. The quintile with the lowest MABA ratio and the highest SUE3 score has the largest positive abnormal return (1.69%). And the quintile with the highest MABA ratio but the lowest SUE3 score has negative 19

20 abnormal return of -0.46%, which is not the lowest for those quintiles with the highest MABA ratio. One possible explanation is that market participants are not sensitive to negative earnings surprises for firms with high growth options, because these firms have abilities to focus more on long term growth than short term earnings. Robustness checks Although most of previous papers use BHAR to measure PEAD, there are two problems for using BHAR, which include bad model (errors in measuring expected return is compounded), and biases of standard tests due to skewness (Fama 1998). Here I also use an alternative method, cumulative abnormal return, to measure the PEAD. Table 9 and Table 10 provide similar results and demonstrate that MABA ratio has significant effects on post-earnings announcement abnormal returns. These results also indicate that there is a negative relationship between MABA ratio and cumulative abnormal return, and that PEAD is more likely to survive for firms with low MABA ratios during periods of (+2, +30) and (+2, +60). V Conclusion and future research In this paper, I find that growth options, which are represented by MABA ratio, have significant effects on post-earnings announcement drift (PEAD). Evidence shows that there is a negative relationship between growth options and post-earnings announcement abnormal returns, and this negative relationship is independent of earnings surprise. Firms with high growth options are more likely to drift downward during the post-earnings announcement periods, regardless of the direction (positive or negative) of the earnings surprise. On the other hand, there are upward 20

21 post-earnings announcement drifts for firms with low growth options, because these firms have high percentage of assets-in-place and are risker than those with high growth options. PEAD is more likely to survive for firms with low growth options, and earnings surprise loses the power of predicting the direction of drifts for firms with high growth options. Further research on the following questions will be helpful for understanding why PEAD is more likely to survive for firms with low growth options: 1) whether there is any difference for market participants interpretation of earnings surprise for firms with high growth options and low growth options? 2) what are effects of distress from operating leverage and financial leverage on idiosyncratic volatility for firms with low growth options? 21

22 References Abarbanell, J., and V. Bernard., (1992). Tests of Analysts Overreaction/Underreaction to Earnings Information as an Explanation for Anomalous Stock Price Behavior. Journal of Finance 47, Adam, T., and V. Goyal., (2003). The Investment Opportunity Set and Its Proxy Variables. The Journal of Financial Research, 31(1), Anderson, C, and L. Garcia-Feijόo., (2006). Empirical Evidence on Capital Investment, Growth Options, and Security Returns. Journal of Finance 61: Anderson, K., J. Harris and E. So., (2007), Opinion Divergence and Post-Earnings Announcement Drift. McDonough School of Business Working Paper Angelini, P., and G. Guazzarotti., (2010), Information Uncertainty and the Reaction of Stock Prices to News. BancaD Italia Research Paper No Ball, R., Bartov, E., (1996). How Naïve is the Stock Market s Use of Earnings Information? Journal of Accounting and Economics, 21, (3), Ball, R.; S. P. Kothari, and R. L. Wattts., (1993). Economic Determinants of the Relation Between Earnings Changes and Stock Returns. The Accounting Review 68, Bartov, Eli, Suresh Radhakrishnan, and Itzhak Krinsky., (2000). Investor Sophistication And Patterns In Stock Returns After Earnings Announcements. The Accounting Review, v 75, Berk, Jonathan B., Richard C. Green, and Vasant Naik., (1999). Optimal investment, growth options and security returns, Journal of Finance 54,

23 Bernard, V.L, Thomas, J.K., (1989). Post-Earnings-Announcement Drift: Delayed Price Response of Risk Premium? Journal of Accounting Research, 27, (3), Bernard, V.L., & Seyhun, H.N., (1997). Does Post- Earnings Announcement drift in Stock Prices Reflect a Market Inefficiency? A Stochastic Dominance Approach. Review of Quantitative Finance and Accounting, Vol. 9, Issue 1, pp Bernard, V.L., Thomas, J.K., (1989). Evidence That Stock Prices Do Not Fully Reflect the Implications of Current Earnings for Future Earnings. Journal of Accounting and Economics, 13, (4), BHUSHAN, R., (1994). An Informational Efficiency Perspective on the Post-Earnings Announcement Drift. Journal of Accounting and Economics 18, Bird, R., Choi, D.F.S., & Yeung, D., (2013), Market uncertainty, market sentiment, and the postearnings announcement drift. Review of Quantitative Finance and Accounting, pp Brown, P., Kennelly, J.W., (1972). The Information Content of Quarterly Earnings: An Extension and Some Further Evidence. Journal of Business, 45, (3), Cao, C., Simin, T., Zhao, J., (2008). Can growth options explain the trend in idiosyncratic risk? Review of Financial Studies 21, Carhart, Mark M., (1997). On Persistence In Mutual Fund Performance. Journal of Finance, 52(1), Carlson, Murray, Adlai Fisher, and Ron Giammarino., (2004). Corporate investment and asset price dynamics: Implications for the cross-section of returns, Journal of Finance 59,

24 Chung, K. and C. Charoenwong., (1991). Investment options, assets in place, and the risk of stocks, Financial Management 20, Collins, D. W. and S. P. Kothari.,(1989). An analysis of intertemporal and cross-sectional determinants of earnings response coefficients, Journal of Accounting and Economics 11, Daniel, Kent, David Hirshleifer, and Avanidhar Subrahmanyam., (1998). Investo psychology and Security Market under- and Overreactions. The Journal of Finance 53, Fama, Eugene F. and James D. MacBeth., (1973). Risk, Return, And Equilibrium: Empirical Tests. Journal of Political Economy, v81(3), Fama, E.F., (1998). Market Efficiency, Long-Term Returns, and Behavioral Finance. Journal of Financial Economics, 49, Fischer, P., ( 2001). Drift as an evolutionary outcome. Working Paper, Penn State University. Foster, G., Olsen, C., Shevlin, R., (1984). Earnings Releases, Anomalies, and the Behavior of Security Returns. The Accounting Review, 59, (4), Freeman, R.N., Tse, S., (1989). The Multiperiod Information Content of Accounting Earnings: Confirmations and Contradictions of Previous Earnings Reports. Journal of Accounting Research, 27, (3), Goyal, V. K., K. Lehn, and S. Racic, (2002). Growth opportunities and corporate debt policy: The case of the U.S. defense industry, Journal of Financial Economics 64, Grullon, Gustavo, Evgeny Lyandres, and Alexei Zhdanov., (2012). Real options, volatility, and stock returns. Jounal of finance, 67, (4)

25 Livnat, Joshua, and Richard Mendenhall., (2006). Comparing the post-earnings announcement drift for surprises calculated from analyst and time series forecasts, Journal of Accounting Research 44, Mendenhall, R.R., (1991). Evidence on the Possible Under-Weighting of Earnings-Related Information. Journal of Accounting Research, 29, (1), Mendenhall, R., (2004). Arbitrage Risk and the Post-earnings-Announcement Drift. Journal of Business, Vol. 77, pp Myers, S., (1977). Determinants of corporate borrowing, Journal of Financial Economics 5, Penman, S. H., S. A. Richardson, and I. Tuna., (2007). The book-to-price effect in stock returns: accounting for leverage. Journal of Accounting Research 45: Prabhala, N. R., (1997). Conditional Methods in Event Studies and an Equilibrium Justification for Standard Event-Study Procedures. Review of Financial Studies, 10(1), 1 38 Sadka, R., (2006). Momentum and Post-Earnings-Announcement Drift: The Role of Liquidity Risk. Journal of Financial Economics, Vol. 80, pp Shivakumar, L., (2007). Discussion of Information Uncertainty and the Post-Earnings- Announcement-Drift. Journal of Business Finance & Accounting, 34, (3) & (4), Smith, C.W., Jr. and R. L.Watts., (1992) The investment opportunity set and corporate financing, dividends, and compensation policies, Journal of Financial Economics 32, Watts, R.L., (1978). Systematic Abnormal Returns after Quarterly Earnings Announcements. Journal of Financial Economics, 6, (2) & (3),

26 Zhang, X., (2006). Information Uncertainty and Stock Returns. Journal of Finance, Vol 61, No. 1, pp Yan Z, Zhao Y., (2011). When two anomalies meet: the post-earnings announcement drift and the value-glamour anomaly. F inanc. Anal. J. 67(6):

27 Appendix A Definition of variables: A. Proxies of growth options 1. CAPEX (or CAPEXPPE) = Capital Expenditures / Property, Plant, and Equipment 2. DTE = [Debt in Current Liabilities + Total Long-Term Debt + Preferred Stock ] /[Common Shares Outstanding Price ] 3. MABA = [Total Assets - Total Common Equity + Price Common Shares Outstanding ] / Total Assets 4. ROA = Earnings at Quarter T / Asset at Quarter (T-1) B. Proxies of surprises 5. SUE1: Standardized unexpected earnings measured by time-series random walk model. It is calculated as the actual earnings minus the average of former eight quarterly earnings, scaled by the quarterly closing price. 6. SUE3: Standardized unexpected earnings using analysts forecasts. It is calculated as the actual earnings minus the mean analyst forecast from I/B/E/S during 90-day period before the disclosure of financial statements, scaled by the quarterly closing price. 7. SUS: Sales surprises. It is calculated as the actual sales minus the mean of analyst forecast sales from I/B/E/S during the 90-day period before the disclosure of financial statements, scaled by the mean of analyst forecast of sales. C. Buy-and-hold abnormal returns and cumulative abnormal returns 8. BHAR (t 1, t 2 ): Buy-and-hold abnormal returns in window (t 1, t 2 ), based on Carhart-factor model. It is equally weighted for a portfolio. 9. CAR (t 1, t 2 ): Cumulative abnormal return in window (t 1, t 2 ), based on Carhart-4-factor model. It is equally weighted for a portfolio. 27

28 Table 1 Descriptive Statistics. The sample consists of 81,805 quarterly firm earnings announcement observations. CAPEX is the ratio of capital expenditures to Property, Plant, and Equipment. DTE is the ratio of debt to equity. MABA is the ratio of market value of asset to its book value. ROA is the return of total assets. SUE1 is the standardized unexpected earnings measured by time-series random walk model. SUE3 is the standardized unexpected earnings measured by analyst forecasts. SUS is the sales surprises. BHAR (-1, +1) is the buy-and-hold abnormal returns during a trading day before and a trading day after the earning announcement. BHAR (+2, +22), BHAR (+2, +44), and BHAR (+2, +63) are buy-and-hold abnormal returns for different post-earnings announcement periods. All buy-and-hold abnormal returns are calculated by using 4-factor model as the benchmark. N Mean Std Dev 5th Pctl 25th Pctl Median 75th Pctl 95th Pctl CAPEX DTE Q MABA ROA MEBE SUE SUE SUS BHAR(-1,+1) BHAR(+2,+22) BHAR(+2,+44) BHAR(+2,+63) Table 2 Correlation matrix. This table reports time-series average of Spearman correlations among listed variables, including growth options, earnings surprises, sales surprises, and buyand-hold abnormal returns. CAPEX DTE MABA ROA MEBE SUE1 SUE3 SUS BHAR (-1,+1) CAPEX 1 28 BHAR (+2,+22) BHAR (+2,+44) BHAR (+2,+63) DTE <.0001 MABA <.0001 <.0001 ROA <.0001 <.0001 <.0001 MEBE <.0001 <.0001 <.0001 <.0001 SUE <.0001 <.0001 <.0001 <.0001 <.0001 SUE <.0001 <.0001 <.0001 <.0001 <.0001 SUS <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 BHAR(-1,+1) < <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 BHAR(+2,+22) <.0001 <.0001 < <.0001 <.0001 < <.0001 BHAR(+2,+44) <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 BHAR(+2,+63) <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 < <.0001 <.0001 <.0001

29 Table 3 Univariate Fama-MacBeth regressions of buy-and-hold abnormal returns on growth options, earnings surprises and sales surprises. Coefficients of the regressions are averaged monthly across sample period from Oct 1998 to Dec T-Values in parentheses are calculated by using the Fama-MacBeth (1973) method and the parameter of lag in the Newey and West (1987) formula is equal to 3. More information about definition of variables is provided in Appendix A. ***, **, and * denote significance at the 1%, 5%, and 10% level, respectively. Panel A: Cross-sectional regressions of buy-and-hold abnormal returns in window (-1, +1) NAME MODEL1 MODEL2 MODEL3 MODEL4 MODEL5 MODEL6 MODEL7 MODEL8 INTERCEPT 0.006*** 0.003** 0.008*** 0.003*** 0.005*** 0.003*** 0.002** 0.003*** (4.99) (2.28) (5.05) (3.35) (4.46) (3.31) (2.39) (3.34) CAPEXPPE ** (-2.56) DTE (1.36) MABA *** (-3.63) ROA 0.007*** (2.64) MEBE ** (-2.41) SUE *** (3.65) SUE *** (9.19) SUS 0.063*** (8.05) RSQ 0.014* 0.009*** 0.015** 0.013* 0.012* 0.020*** 0.048*** 0.026*** (1.96) (5.29) (2.25) (1.85) (1.90) (3.03) (5.83) (3.93) N OBS Panel B: Cross-sectional regressions of buy-and-hold abnormal returns in window (+2, +22) NAME MODEL1 MODEL2 MODEL3 MODEL4 MODEL5 MODEL6 MODEL7 MODEL8 INTERCEPT *** 0.015*** *** *** *** *** (-0.20) (-5.36) (6.03) (-4.56) (-0.37) (-4.55) (-4.97) (-3.95) CAPEXPPE *** (-3.58) DTE 0.010*** (3.76) MABA *** (-12.07) ROA (-0.07) MEBE *** (-5.58) SUE * (-1.71) SUE (1.28) SUS (-0.99) RSQ 0.019*** 0.014*** 0.032*** 0.018*** 0.020*** 0.020*** 0.024*** 0.017** (2.90) (4.39) (4.80) (2.72) (3.07) (2.69) (3.59) (2.37) N OBS

30 Table 3 Univariate Fama-MacBeth regressions of buy-and-hold abnormal returns on growth options, earnings surprises and sales surprises. Coefficients of the regressions are averaged monthly across sample period from Oct 1998 to Dec T-Values in parentheses are calculated by using the Fama-MacBeth (1973) method and the parameter of lag in the Newey and West (1987) formula is equal to 3. More information about definition of variables is provided in Appendix A. ***, **, and * denote significance at the 1%, 5%, and 10% level, respectively. Panel C: Cross-sectional regressions of buy-and-hold abnormal returns in window (+2, +44) NAME MODEL1 MODEL2 MODEL3 MODEL4 MODEL5 MODEL6 MODEL7 MODEL8 INTERCEPT ** *** 0.015*** *** *** *** *** *** (-2.44) (-7.78) (3.17) (-7.38) (-4.16) (-7.48) (-7.68) (-7.38) CAPEXPPE *** (-4.61) DTE 0.019*** (4.41) MABA *** (-11.16) ROA (-0.87) MEBE *** (-5.54) SUE (-1.52) SUE (-0.91) SUS (-1.59) RSQ 0.021*** 0.014*** 0.040*** 0.020*** 0.023*** 0.020*** 0.022*** 0.017** (3.25) (5.83) (5.42) (2.63) (3.26) (2.65) (2.96) (2.38) N OBS Panel D: Cross-sectional regressions of buy-and-hold abnormal returns in window (+2, +63) NAME MODEL1 MODEL2 MODEL3 MODEL4 MODEL5 MODEL6 MODEL7 MODEL8 INTERCEPT ** *** 0.018** *** *** *** *** *** (-2.35) (-8.14) (2.53) (-6.94) (-4.85) (-8.26) (-7.90) (-8.17) CAPEXPPE *** (-4.69) DTE 0.026*** (3.99) MABA *** (-10.55) ROA (-1.63) MEBE *** (-5.20) SUE (-0.13) SUE (-0.98) SUS (-0.65) RSQ 0.023*** 0.017*** 0.045*** 0.017** 0.023*** 0.021*** 0.021*** 0.016** (3.39) (5.10) (5.80) (2.52) (3.34) (2.97) (3.07) (2.25) N OBS

31 Table 4 Multivariate Fama-MacBeth regressions of buy-and-hold abnormal returns on growth options, earnings surprises and sales surprises. Coefficients of the regressions are averaged monthly across sample period from Oct 1998 to Dec T-Values in parentheses are calculated by using the Fama-MacBeth (1973) method and the parameter of lag in the Newey and West (1987) formula is equal to 3. More information about definition of variables is provided in Appendix A. ***, **, and * denote significance at the 1%, 5%, and 10% level, respectively. Panel A: Cross-sectional regressions of buy-and-hold abnormal returns in window (-1, +1) NAME MODEL1 MODEL2 MODEL3 MODEL4 MODEL5 MODEL6 MODEL7 MODEL8 MODEL9 MODEL10 MODEL11 MODEL12 INTERCEP0.009*** 0.009*** 0.002** 0.009*** 0.008*** 0.009*** 0.006*** 0.005*** 0.007*** 0.007*** 0.005*** 0.007*** (4.64) (4.38) (2.19) (4.45) (3.88) (4.55) (4.07) (3.43) (4.35) (3.72) (3.31) (3.57) CAPEXPP * * ** ** ** (-1.33) (-1.41) (-0.98) (-1.65) (-1.51) (-1.74) (-2.41) (-2.26) (-1.53) (-2.30) (-1.60) DTE (-0.15) (-0.07) (0.31) (0.26) (-0.42) (0.74) (0.61) (-0.01) (0.09) (0.40) (0.09) MABA *** * *** *** *** *** * (-3.41) (-1.92) (-3.47) (-2.92) (-3.19) (-2.87) (-1.75) ROA ** * (1.61) (1.43) (1.98) (0.55) (1.71) (1.55) (0.30) (1.20) (0.99) (0.66) (0.77) MEBE ** ** ** * (-1.29) (-2.39) (-2.07) (-2.08) (-1.95) (-0.55) SUE *** 0.077*** (0.53) (3.14) (3.29) (0.57) (0.70) (0.78) SUE *** 2.150*** 2.172*** 2.019*** 2.036*** 2.034*** (8.60) (9.66) (9.38) (9.00) (8.78) (8.77) SUS 0.052*** 0.061*** 0.062*** 0.050*** 0.050*** 0.050*** (7.43) (7.66) (7.67) (7.05) (6.95) (6.95) ADJRSQ 0.010*** 0.009*** 0.051*** 0.018*** 0.047*** 0.024*** 0.016*** 0.044*** 0.021*** 0.060*** 0.057*** 0.059*** (3.41) (3.20) (11.14) (4.86) (10.86) (6.41) (4.11) (10.12) (5.68) (12.71) (11.81) (12.17) N OBS Panel B: Cross-sectional regressions of buy-and-hold abnormal returns in window (+2, +22) NAME MODEL1 MODEL2 MODEL3 MODEL4 MODEL5 MODEL6 MODEL7 MODEL8 MODEL9 MODEL10 MODEL11 MODEL12 INTERCEPT 0.014*** 0.014*** *** 0.013*** 0.013*** 0.014*** *** *** (5.04) (4.87) (-4.84) (5.07) (4.51) (5.23) (0.17) (-0.10) (0.43) (4.63) (-0.18) (4.49) CAPEXPPE ** ** ** ** (-0.79) (-0.74) (-0.83) (-0.70) (-0.85) (-2.35) (-2.18) (-2.36) (-0.80) (-2.30) (-0.78) DTE *** 0.006** 0.006** ** (1.05) (1.09) (1.27) (0.93) (0.88) (2.70) (2.45) (2.33) (1.07) (2.56) (1.11) MABA *** *** *** *** *** *** *** (-10.49) (-8.14) (-11.00) (-10.03) (-10.89) (-10.85) (-8.35) ROA (-0.20) (-0.22) (0.86) (-0.63) (-0.19) (-0.16) (-1.52) (-1.23) (0.15) (-0.69) (0.18) MEBE * *** *** *** *** (-1.87) (-5.62) (-5.36) (-5.68) (-5.54) (-1.58) SUE (-1.59) (-1.42) (-1.44) (-1.35) (-1.47) (-1.46) SUE *** 0.632*** 0.678*** 0.711*** 0.767*** 0.741*** (2.87) (2.62) (2.74) (3.02) (3.17) (3.05) SUS (-0.55) (-1.11) (-1.15) (-1.12) (-1.25) (-1.20) ADJRSQ 0.038*** 0.039*** 0.027*** 0.046*** 0.049*** 0.043*** 0.036*** 0.039*** 0.034*** 0.061*** 0.051*** 0.062*** (8.36) (8.47) (6.47) (9.57) (8.37) (8.66) (7.31) (6.78) (6.92) (9.66) (7.98) (9.88) N OBS