Investor Trading and Return Patterns around Earnings Announcements

Transcription

1 Investor Trading and Return Patterns around Earnings Announcements Ron Kaniel, Shuming Liu, Gideon Saar, and Sheridan Titman This version: September 2007 Ron Kaniel is from the Fuqua School of Business, One Towerview Drive, Duke University, Durham, NC (Tel: , Shuming Liu is from the College of Business, San Francisco State University, 1600 Holloway Avenue, San Francisco, CA (Tel: , Gideon Saar is from the Johnson Graduate School of Management, Cornell University, 455 Sage Hall, Ithaca, NY (Tel: , Sheridan Titman is from the McCombs School of Business, University of Texas at Austin, Austin, TX (Tel: , This research began while Saar was on leave from New York University and held the position of Visiting Research Economist at the New York Stock Exchange. The opinions expressed in this paper do not necessarily reflect those of the members or directors of the NYSE.

2 Investor Trading and Return Patterns around Earnings Announcements Abstract This paper investigates the behavior of individual and institutional investors around earnings announcements using a unique dataset of NYSE stocks. We find that intense individual buying (selling) prior to the announcement is associated with significant positive (negative) abnormal returns in the three months following the event, with most of the abnormal returns generated by stocks that experience extreme earnings surprises. A similar strategy that follows the trading of institutional investors does not yield significant payoffs. We also examine the behavior of individuals after the earnings announcement and find that they trade in the opposite direction to both pre-event returns (i.e., exhibit contrarian behavior), as well as the earnings surprise (i.e., exhibit newscontrarian behavior). The latter behavior may contribute to the post-earnings announcement drift, if it slows down the adjustment of prices. While trading against the drift may seem like a dubious strategy, the combination of patterns we observe before and after the events could suggest that individuals are profitably reversing positions which they entered before the announcements.

3 I. Introduction There is a growing body of empirical research that documents the extent to which stock returns are predictable around earnings announcement dates. This research finds that unconditional expected returns are higher around earnings announcements, 1 that characteristics that predict returns unconditionally tend to more accurately predict returns around earnings announcements, 2 and that returns are abnormally high (low) in the months following good (bad) earnings announcements surprises (the drift ). 3 Although the higher risk around earnings announcements is likely to generate some abnormal returns, the degree of return predictability and the magnitude of the post-earnings announcement drift are likely to be due to trading frictions or other inefficiencies. Some researchers suggest that since earnings announcements are quite visible, they are likely to attract the attention of individual investors who are more likely to be influenced by behavioral biases. As a result, we may see more uninformed or irrational trading around earnings announcements, which may generate the above mentioned biases. For example, Hirshleifer, Myres, Myres, and Teoh (2002) posit that if the drift reflects a profit opportunity due to mispricing, more sophisticated investors should buy after positive surprises (before an upward drift) and sell after negative surprises (before a downward drift). Naïve investors would trade in the opposite direction, and their trading would affect the supply of shares in the market and therefore would slow down the adjustment of prices to the information, creating the drift. Investigating the trades of clients of a discount broker, Hirshleifer et al. did not find support for this hypothesis. Our paper contributes to the literature in a number of ways. First, we examine a comprehensive dataset that aggregates the trades of all individual or institutional investors who trade on the NYSE over a four-year period. On this dimension, our 1 See Chari, Jagannathan and Ofer (1988) and Lamont and Frazzini (2006). 2 Jegadeesh and Titman (1993), for example, document high returns around future earnings announcements for stocks that have performed well in the previous six months. 3 The drift was first described in Ball and Brown (1968). See also Foster, Olsen, and Shevlin (1984) and Bernard and Thomas (1989, 1990). 1

4 research extends prior research that either indirectly infers the trades of individuals (or institutions) based on trade size, or looks at a small subset of the market (either a sample of individual trades from one discount brokerage house or the TORQ sample of 144 NYSE stocks over a three-month period). Second, we also examine the trading of institutions around earnings announcement to contrast the individuals behavior with that of (presumably) more sophisticated investors. Third, we investigate the relation between individual trading prior to earnings announcements and future returns. This analysis relates to Kaniel, Saar and Titman (2007), who suggest that by acting as contrarians, individual investors implicitly provide liquidity to institutions. Kaniel, Saar, and Titman find that unconditionally, the liquidity provision of individuals is profitable; however, it is possible that conditional on a future information event, individuals will lose money to the more informed institutions. Alternatively, it may be the case that because of the uncertainty associated with future earnings announcements, institutions will have a greater need for liquidity, providing individuals with greater profit opportunities. Our results are consistent with the latter hypothesis. Specifically, we find that stocks that individuals bought in the ten days prior to the event realize abnormal returns that exceed the abnormal returns of stocks they sold by 1.45% in the two-day event window and 5.45% in the first three months after the event. Double sorting on net individual trading prior to the event and earnings surprise (relative to analysts forecasts) reveals that these abnormal returns in the post-event period are strongest for stocks that experience the most extreme news, both positive and negative. Regression analysis confirms that the positive relation between net individual trading before the earnings announcement and returns following the announcement are robust to controlling for past returns and the contemporaneous relation between returns and individual trading. If institutions are motivated to sell prior to earnings announcements in order to reduce the risk of their holdings, then we would expect to see more selling by institutions 2

5 and buying by individuals in stocks with higher earnings uncertainty. Our evidence suggests that this is indeed the case. When we sort on earnings uncertainty (measured by the dispersion of analysts forecasts) prior to the event, we find that institutions sell and individuals buy stocks with high earnings uncertainty two and three months before the announcements. We also find that the post announcement returns associated with net individual trading before the event are greater for stocks with higher earnings uncertainty. An additional research question we investigate is whether the trading of individuals slows down the adjustment of prices and hence contributes to the creation of the post-earnings announcement drift. The most robust pattern we observe for the individuals in the post-event period is that they trade in the opposite direction to preevent returns (i.e., exhibit contrarian behavior), and that their trading is also affected by the direction of the earnings surprise in a news-contrarian fashion. The news-contrarian pattern means that they buy more (or sell less) of the stocks that experience more negative earnings surprises. Institutional investors exhibit the opposite patterns in the post-event period, showing a momentum tendency (trading in the direction of pre-event returns) that combines with a news-momentum pattern (more pronounced trading in the direction of the earnings surprise). The news-contrarian behavior of individuals is consistent with the hypothesis in Hirshleifer et al. that individuals are responsible (at least in part) for the post-earnings announcement drift phenomenon. While trading in the opposite direction of the drift may in fact slow down the price adjustment process and may not, in isolation, be a good strategy, combining our findings on individual trading before and after the events may suggest that individuals are profitably reversing positions to which they have entered before the announcements. The rest of this paper proceeds as follows. The next section describes the sample and the comprehensive dataset we use. Section III investigates the relation between net imbalances of individuals and institutions prior to the earnings announcements and 3

6 subsequent returns. Section IV examines the behavior of individuals and institutions after the announcements. Section V contains a detailed discussion of the most related papers in the literature, comparing and contrasting our results with prior evidence. Section VI concludes. II. Sample and Data II.A. Individual and Institutional Trading Data We study the trading of individuals and institutions around earnings announcements using a comprehensive dataset that contains four years of daily buy and sell volume of executed orders for a large cross section of NYSE stocks. The dataset was constructed from the NYSE's Consolidated Equity Audit Trail Data (CAUD) files that contain all orders that execute on the exchange. The CAUD files include a field called Account Type that specifies for each order whether it originates from an institution or an individual investor. Account Type is a mandatory field a broker has to fill for each order that is sent to the NYSE. The Account Type field is not audited by the NYSE on an order-by-order basis, but NYSE officials monitor the use of this field by brokers. In particular, any abnormal use of the individual investor designation in the Account Type field by a brokerage firm is likely to draw attention, which prevents abuse of the reporting system. We therefore believe that the Account Type designation of orders is fairly accurate. 4 An important advantage of our dataset is that the information about daily buy and sell volume of individual and institutional investors was created by aggregating executed orders, rather than trades. In other words, the classification into buy and sell volume in our dataset is exact, and we do not have to rely on classification algorithms such as the one proposed by Lee and Ready (1991). 4 Additional information on the Account Type field (and the reporting of individual investor trading) can be found in Lee and Radhakrishna (2000) and Kaniel, Saar, and Titman (2007). 4

7 Most previous work on U.S. markets examined the behavior of institutions and individuals around earnings announcements using either indirect measures of investor trading or a very limited sample. For example, researchers have used small and large trades as proxies for the trades of individual and institutional investors (e.g., Lee (1992), Shanthikumar (2004), and Frazzini and Lamont (2006)). While such a classification was shown to be reasonable for a 1990 sample of NYSE stocks (Lee and Radhakrishna (2000)), recent research casts doubt on its usefulness. For example, Campbell, Ramadorai, and Schwartz (2007) look at how trades of different sizes relate to changes in institutional holdings from 1993 through 2000 and conclude that the smallest trades (below $2,000) are more likely to come from institutions rather than individuals. 5 To examine the behavior of institutional investors, some papers have used changes in quarterly institutional holdings around earnings announcements (e.g., Ali, Durtschi, Lev, and Trombley (2004) and Baker, Litov, Wachter, and Wurgler (2005)). This, however, limits the ability to examine trading over shorter intervals, and has the added complication that the mapping between institutional trading and changes in institutional holdings is not one-to-one. To examine trading by individuals, some researchers (e.g., Hirshleifer, Myers, Myers, and Teoh (2002)) have used a sample of individuals who traded through one discount broker from 1991 through While measuring directly individual trading, this sample is much smaller than the one we are using, and looks at a subset of individuals that may or may not be representative of the overall population. Finally, while Welker and Sparks (2001) and Nofsinger (2001) have used the NYSE s TORQ database to look at individual and institutional trading around public announcements, they do not observe the main results we report (especially the relation between individual trading prior to the event and subsequent returns). While TORQ also contains Account Type data (like our dataset), it includes only 144 stocks for a three month period between November 5 Hvidkjaer (2005), who investigates the relation between small trade volume and stock returns, also notes that small trade volume increases markedly in the final years of his sample (that ends in 2004), and it no longer seems to be negatively related to changes in institutional holdings. The bulk of the increase in small trading is probably coming from institutions that split orders into small trades. 5

8 1990 and January Hence, our different results could be due to the small sample size in the TORQ database or because it contains much older data. We start our construction of a daily abnormal net individual trading series by computing an imbalance measure: subtracting the value of the shares sold by individuals from the value of shares bought and dividing by the average daily dollar volume (from CRSP) in the calendar year. 6 We then subtract the daily average of that imbalance measure over the sample period to get an abnormal net individual trading measure, which we believe is more suitable for examining the patterns of trading around earnings announcements. Specifically, we define IndNT i,t for stock i on day t as: IndNT where, = Individual Imbalance it, it, it, all days in Individual Imbalance it, Individual Imbalance Individual buy dollar volume Individual sell dollar volume = Average daily dollar volume in the calendar year it, it, it, We define abnormal net institutional trading for stock i on day t, InsNT i,t, in an analogous fashion (i.e., we compute the daily imbalance of institutional orders, buys minus sells divided by volume, and subtract from it the average of the institutional imbalance over the sample period). 7 We define cumulative abnormal net individual or institutional trading over a certain period, [t,t], as: T T i i [, tt] = ik, and [, tt] = ik, k= t k= t IndNT IndNT InsNT InsNT 6 Kaniel, Saar, and Titman (2007) note that some trading in NYSE-listed stocks does not take place on the NYSE. For example, some brokers either sell some of their retail order flow to wholesalers for execution or internalize a certain portion of their clients orders by trading as principal against them. These trades take place on one of the regional exchanges (or alternatively reported to the NASD) and are therefore not in our sample of NYSE executions. However, these brokers still send a certain portion of their retail order flow to the NYSE, and are more likely to send those orders that create an imbalance not easily matched internally. Therefore, Kaniel, Saar, and Titman argue that net individual trading (i.e., imbalances in individuals executed orders on the NYSE) probably reflects, even if not perfectly, the individuals imbalances in the market as a whole. 7 We ignore orders marked as index arbitrage when constructing our net trading measures because they presumably do not contain information about firm specific events such as earnings announcements. Also, we apply a filter to remove potential reporting errors in the dataset: we take out days where the magnitude of the imbalance in the trading of an investor clientele is more than five times the average daily dollar volume of the stock. 6

9 where the periods are defined relative to the earnings announcement date (day zero). For example, IndNT [-10,-1] is cumulative abnormal net individual trading from ten days prior to the earnings announcement to one day prior to the announcement. Similarly, InsNT [2,61] is cumulative abnormal net institutional trading from two days after the announcement to (and including) day 61. Note that on any given period our measures of net individual trading and net institutional trading need not exhibit mirror image behavior because of the presence of dealers on the NYSE. More specifically, each NYSE stock is associated with a single specialist who makes a market in the stock, buying and selling for his own account. Specialists are supposed to ensure price continuity and stability for the stocks, and hence could absorb to some extent temporary imbalances that a certain investor clientele may exhibit. II.B. Sample Our sample contains all common, domestic stocks that were traded on the NYSE any time between January 1, 2000 and December 31, We use the CRSP database to construct the sample, and match the stocks to the NYSE dataset of individual and institutional trading by means of ticker symbol and CUSIP. This procedure results in a sample of 2,034 stocks. We then use IBES and COMPUSTAT to identify all the dates where stocks in our sample had earnings announcements, and impose two restrictions on the sample. 8 First, we require 60 days of data prior to and after the announcements, which eliminates most announcements from the first (and last) three months of the sample period. Second, in order to compute our analysts earnings surprise measure we require that there is an observation in the IBES database for the mean analysts forecast in the month prior to the earnings announcement (i.e., at least one analyst with an earnings forecast), and also information about the actual earnings number. 8 For each stock on each quarter, we compare the announcement dates from IBES (the REPDATS field) and COMPUSTAT (the RDQE filed) and choose the earlier one if they are different. 7

10 Our screens result in a final sample of 1,821 stocks with 17,564 earnings announcement events. 9 Panel A of Table 1 presents summary statistics from CRSP on the sample stocks (for the entire sample and for three size groups). Panel B of Table 1 reports the number of events in each month of the sample period. Table 2 looks at net individual and institutional trading around earnings announcements. In Panel A we observe that individuals buy stocks in the month prior to earnings announcement. At the time of the event itself (days [0,1]) individuals sell (mostly large stocks), and we observe continued selling in the two weeks after the event. Panel B of Table 2 shows that institutional investors buy stocks during the event-window, and continue buying in the week after the event mainly small and mid-cap stocks. II.C. Abnormal Returns and Earnings Surprises Throughout the paper we define abnormal returns as market-adjusted returns, and use the equal-weighted portfolio of all stocks in the sample as a proxy for the market portfolio. To create the cumulative returns of the market portfolio, say over a 60-day period, we first compute for each stock the cumulative (raw) return over the relevant 60-day period. The average of these returns across the stocks in the sample is what we define as the return on the equal-weighted market portfolio. Our definition of cumulative abnormal returns for stock i in period [t, T], CAR i,[t,t], is the cumulative return on stock i minus the cumulative return on the market proxy (for period [t,t]). Our results are robust to the use of size-adjusted returns as an alternative definition of abnormal returns. Our investigation focuses not on the unconditional relation between investor trading and returns around earnings announcements but rather on whether investors anticipate news and how they react to good and bad news. Therefore, we require a measure of earnings surprise (or the news component of the earnings announcement), and use analysts forecasts to define that surprise. More specifically, we define the 9 This is the number of events we use for analysis of individual investor trading. Due to the filter we discuss in footnote 7 that removes potential reporting errors from our NYSE dataset, the sample we use for analysis of institutional investor trading consists of 17,419 events. 8

11 normalized earnings surprise, ES, as the actual earnings minus the earnings forecast, divided by the price on the forecast day. The earnings forecast is the mean of analysts forecast one month before the earnings announcements. An earnings surprise measure using analysts forecasts is rather standard in the literature, but we certainly acknowledge that it is just a proxy for the surprise. There are also papers that use the abnormal return at the time of the earnings announcement as a proxy for the surprise, and each measure has its own advantages and disadvantages. 10 In our regression analysis explaining post-event investor trading we include, in addition to the analysts earnings surprise measure, the abnormal return at the time of the announcement as an additional proxy for the news content of the announcement. III. Investor Trading before the Event and Return Predictability Earnings announcements represent the arrival of news to the market. We are interested in investigating whether a certain investor clientele, individuals or institutions, seems to have the ability to predict the news or its consequence (in terms of abnormal return). Therefore, we examine the trading of individuals and institutions in the 10 trading days (two weeks) prior to the earnings announcement and look at the stocks they intensely buy or sell. Then, we relate their actions prior to the announcement to the cumulative abnormal returns on and after the announcement date. In the first exercise, we sort all stocks each quarter according to net investor trading prior to the announcement, either by individuals or by institutions, and put the stocks in five categories (quintile 1 contains the stocks that investors sold the most and quintile 5 contains the stocks investors bought the most). We compute for each stock the 10 The analysts earnings surprise measure presumably reflects the surprise relative to the opinion of wellinformed, sophisticated investors. It has the advantage that it does not involve the price level or return at the time of the event that can be affected by liquidity shocks unrelated to the actual updating of beliefs about the stock. On the other hand, it is perfectly conceivable that investors other than sell-side analysts (e.g., skillful individuals, hedge funds, and proprietary trading desks) have information that is relevant to the pricing of the stock that sell-side analysts do not possess. As such, the return at the time of the announcement would aggregate everyone s opinion, leading to a better measure of surprise than the one that solely considers the information set of the sell-side analysts. 9

12 cumulative market-adjusted return over a certain period, and present in Table 3 the average of the market-adjusted abnormal returns of all the stock-quarters in each of the different quintiles (with t-statistics testing the hypothesis of zero abnormal returns). We use sixty days starting two days after the announcement as the length of our post-event period to be consistent with the literature that examined the post-earnings announcement drift. Panel A of Table 3 presents the results on trading by individuals. We observe a very strong pattern: the stocks that individuals intensely bought in the two weeks before the announcements outperform those that they intensely sold, on average, by 1.45% during the event window (days [0,1]), and an additional 4.13% payoff in the three-month post-event period (days [2,61]). These excess returns are of comparable magnitudes for stocks with intense buying and selling. Stocks that individuals intensely sell (quintile 1) experience a negative abnormal return of 0.7% on the event, and 2.8% in the postevent period, while those they intensely buy (quintile 5) have 0.80% abnormal return in the event window, and 1.35% after the event. 11 We also sorted the stocks according to size and repeated the analysis separately for three market-capitalization groups: small, mid-cap, and large stocks. 12 For this analysis, we computed abnormal returns for a stock by subtracting from it the return of the equal-weighted portfolio of all stocks in its group. In all three size groups the strategy of buying the stocks that individuals bought before the event and selling those they sold produces significant profits: 8.03% after three months ([0,61]) for small stocks, 3.51% for mid-cap stocks, and 3.03% for large stocks. In Panel B of Table 3 we conduct a similar analysis for institutions. Sorting on InsNT [-10,-1], we find that the stocks institutions intensely sell realize abnormal returns of 11 We find a similar pattern when we sort on net individual trading in the 20 days prior to the announcement. 12 We sort stocks into deciles by market capitalization and define small stocks as those in deciles 1, 2, 3, and 4, mid-cap stocks as those in deciles 5, 6, and 7, and large stocks are those comprising deciles 8, 9, and

13 1.8% from day zero until day 61, which is statistically different from zero. However, stocks that institutions intensely buy also realize negative abnormal returns, of 2.3% in the [0, 61] period. We also conducted a separate analysis of stocks by size categories and the result was similar to the one we present for the entire sample in the bottom row of the panel: profit on the zero-investment strategy that is indistinguishable from zero. To summarize the results in Table 3, we observe that pre-event trading by individuals is significantly related to abnormal returns at the time of the event and in the post-event period. On the other hand, there is not a significant difference in the abnormal returns of stocks that institutions intensely buy and sell. 13 In Table 4 we sort stocks independently along two dimensions: the analysts earnings surprise measure (ES) and IndNT [-10,-1]. We put the stocks into 25 categories: five groups of earnings surprise (quintile 1 is the most negative surprise and quintile 5 the most positive surprise), and five groups of net individual trading (quintile 1 are those stocks individuals sold the most in the 10 days prior to the announcement and quintile 5 are those they bought the most over that period). We then examine the cumulative market-adjusted returns in three periods: the event window ([0, 1]), the post-event period [2, 61]), and the combined return on and after the event ([0, 61]). Panel A of Table 4 reports the abnormal return during the event window for the 25 categories. Consistent with previous research, stocks that experience a positive (negative) surprise with respect to the analysts forecast have a positive (negative) abnormal return at the time of the event. As in Table 3, we also observe that stocks individuals bought have higher abnormal returns than those they sold in each earnings 13 The pattern of significant positive abnormal returns for the strategy that follows individual trading is mirrored by significant abnormal negative returns for the strategy that follows institutional trading during the event window ([0,1]); however, the negative returns for the institutional strategy are insignificant in the post-event period. While many patterns we observe in the trading of institutions are mirror image of those exhibited by individuals, this need not always be the case. As we mention in Section II.A., NYSE specialists have a substantial presence in the trading of stocks, and our analysis of institutional trading also intentionally excludes index arbitrage. The more heterogeneous nature of institutional trading strategies coupled with the presence in the market of the specialist (in addition to individual investors) means that we are likely to have less power detecting patterns that are associated with the trading of institutions. 11

14 announcement quintile. It seems as if the performance of the zero-investment strategy that follows individual trading (in the bottom row of the panel) is better for stocks that experience extreme earnings surprises (1.91% for bad news in quintile 1 and 1.85% for good news in quintile 5) than for no-news stocks (1.17% in quintile 3), but the differences are not statistically significant. The performance of this zero-investment strategy differs much more between news and no-news events when we examine the post-event period ([2, 61]) in the last row of Panel B. The most prominent pattern we observe is that buying the stocks individuals bought and selling those they sold is profitable mainly in the extreme surprise quintiles (e.g., quintile 1 and quintile 5). For example, the payoff to this strategy for good-news stocks is 5.50% while the payoff for the no-news stocks (quintile 3) is only 2.13%. 14 Panel C of Table 4 provides evidence on returns in the combined period that includes the event window and the post-event period. We observe that the abnormal return in the three months following the announcement is the most negative for stocks that experience both individual selling prior to the event and negative news (-6.80% abnormal return over [0,61]), while the abnormal return is most positive for those stocks that experience both individual buying prior to the event and positive news (7.57% over that period). One interpretation of the results in Table 4 is that individuals have useful information about the forthcoming earnings announcements. Alternatively, as discussed in Kaniel, Saar, and Titman (2007), individuals may profit by providing liquidity to institutions that may have an incentive to exit positions prior to earnings announcement dates. However, the magnitude of the returns observed in this study suggests that liquidity provision is unlikely to provide the sole explanation. In comparison to Kaniel, Saar, and Titman (2007), who find cumulative abnormal returns of about -0.5% (0.8%) in the two to three months after a week of intense selling (buying) by individuals, the three 14 The difference between them is marginally statistically significant (p-value = ) using an F-test. 12

15 month returns associated with individual trading prior to earnings announcements are four to five times larger. In Table 5 we present the same analysis for institutional investors. Panel A provides no further insights relative to Table 3: we observe that the strategy that mimics institutional buying and selling is unprofitable. Similarly, conditioning on news does not help us observe any predictability with respect to post-event abnormal returns: Panel B shows that abnormal returns in the post-event period are generally negative for both the stocks institutions intensely sell and those that they buy. A similar picture is observed when we consider abnormal returns over the combined event window and post-event period. Overall, it does not seem as if conditioning on the earnings surprise enables us to better uncover any predictive ability by institutions with respect to returns on and after the event. We proceed to investigate the ability of net individual trading prior to announcements to predict cumulative abnormal returns on and after the announcements (CAR [0,61] ) using multivariate regressions that control for mean-reversion in returns and the contemporaneous relation between net individual trading and returns. For robustness, we use models where pre-event abnormal returns and net individual trading are measured over either 10 days or 60 days before the announcements. 15 In order to overcome potential econometric problems associated with contemporaneously correlated errors for earnings announcements that are clustered in time (i.e., in the same quarter), we use a methodology in the spirit of Fama and MacBeth (1973). We first run a cross-sectional regression separately for each quarter. We then use time-series variability in the 16 estimates we have for each coefficient (one from each quarterly regression) to get the standard error for the mean coefficient. 15 The reason we consider both specifications is that while in Table 4 and Table 5 we focus on net investor trading in the 10 days before the event, our choice for a three-month post-event period follows other papers in the drift literature, and therefore we also look at a pre-event period of 60 days to have equal periods before and after the announcements. 13

16 In Panel A of Table 6 we observe that the coefficient on pre-event net individual trading is positive and statistically significant in the regressions using the entire sample. This means that more intense individual buying before the earnings announcement is associated with higher market-adjusted abnormal returns on and after the event. 16 We also ran separate regressions predicting the abnormal return at the time of the event (CAR [0,1] ) and in the post-event period (CAR [2,61] ). The results were similar: the coefficient on net individual trading before the event was statistically significant in all specifications. Panel A of Table 6, which reports separate regressions for small, mid-cap and large stocks, reveals similar results across the size groups. In contrast, the coefficient on pre-event net institutional trading in Panel B of Table 6 is statistically significant in only two of the eight regression specifications, and it has the wrong sign (i.e., it is negative). In other words, there is no evidence of institutional ability to predict excess return associated with the event. 17 III.A. Earnings Uncertainty Up to this point we have shown that net individual investor trading prior to earnings announcements is associated with abnormal returns on and after the event. The evidence of a relation between excess returns and individual trading is stronger when following extreme earnings surprises. Since more extreme surprises (either good or bad) could be the result of greater uncertainty about earnings prior to the announcement, we proceed by examining the relation between the returns on the individuals strategy and earnings uncertainty. To measure earnings uncertainty (EU) prior to the announcement we use the dispersion in analysts forecasts: the high forecast minus the low forecast divided by the price on the forecast day. The high and low forecasts are from one month prior to the 16 We include IndNT [0,61] in the regression to see whether the apparent ability to predict return is simply due to a contemporaneous relation between net individual trading and returns after the event coupled with serial correlation in individual trading. We observe that the predictive ability of net individual investor trading in the pre-event period is not subsumed by the inclusion of this variable. 17 Similar results are obtained when we use either CAR [0,1] or CAR [2,61] as the dependent variable. 14

17 announcement, and we only use stocks with at least three analysts providing forecasts. 18 Stocks are sorted every quarter according to EU and put into five groups: quintile 1 is the one with the least uncertainty (smallest forecast range) and quintile 5 is the one with the most uncertainty (largest forecast range). Table 7 presents the returns associated with net individual and institutional trading prior to the earnings announcement conditional on the level of uncertainty. We observe an interesting pattern: institutions sell stocks with high uncertainty of earnings prior to the announcement, and their intensity of selling is significantly different from zero in the three months and one month periods prior to the event. Individuals, on the other hand, buy stocks with high earnings uncertainty prior to the announcement, and their buying is significantly different from zero in the three months, one month, and two weeks periods prior to the event. They also buy stocks with low uncertainty about earnings, but at least over the entire three-month period prior to the announcement, their buying of high uncertainty stocks is greater than their accumulation of low uncertainty stocks (i.e., the number in the last row of the table, Q5 Q1, is statistically different from zero). The evidence in Table 7 points to a potential explanation for the abnormal return we document for the strategy that follows individual buying. It is possible that when institutions dispose of high earnings uncertainty stocks prior to the announcements, individuals end up buying the high uncertainty stocks, and are compensated by realizing higher returns on average during the event window and in the post-event period. In Panel A of Table 8 we double sort independently on earnings uncertainty and net individual investor trading in the 60 days prior to the event. 19 We then look at abnormal returns in 18 This reduces the number of earnings announcement events in the analysis of individual (institutional) trading to 14,149 (14,118). 19 We use 60 days in this case rather then 10 days as in the rest of the analysis because the evidence in Table 8 is that individual buying of stocks with high uncertainty is significantly greater than their selling of these stocks when we consider a three-month period prior to the event, but not in the two weeks prior to the event. 15

18 the three months after the event (CAR [0,61] ). We observe that the excess returns associated with individual accumulation are much stronger for high earnings uncertainty stocks: the difference in returns is 7.39% in EU quintile 5 versus 2.04% in EU quintile 1 (and the difference is statistically significant). 20 It is interesting to note, though that we cannot detect the opposite pattern for the institutional investors in Panel B of Table 8. In other words, it does not seem as if their selling of high uncertainty stocks prior to the event means that they earn more or less return after the event. 21 IV. Investor Trading after the Event While the previous section focused on the behavior of individuals and institutions prior to earnings announcements, in this section we look at their trading during and after the event. The reason to focus on their behavior after the event is that one of the puzzles associated with earnings announcements is the drift, which is the phenomenon that stocks with negative earnings surprises experience negative abnormal returns in the postevent period and stocks with positive earnings surprises experience positive abnormal returns in the post-event period. Some authors have conjectured that the behavior of individuals is responsible for the slow adjustment of prices to information in earnings announcements, which manifests itself as the drift. Indirect evidence for this effect is found in Bartov, Krinsky, and Radhakrishnan (2000), who document that the drift is negatively related to the extent of 20 We use an F-test to statistically verify this effect. 21 Varian (1985) provides a theoretical model where a greater dispersion of beliefs is associated with lower asset prices, which would imply that when earnings uncertainty is reduced (after the announcement) we should observe higher prices. In Miller (1977), on the other hand, divergence of opinions along with short selling restrictions lead to overpricing (because prices would reflect the more optimistic valuation of the assets), and the resolution of uncertainty should therefore lead to a downward revision in prices. Garfinkel and Sokobin (2006) use unexpected volume at the earnings announcement as a measure of investors opinion divergence and find that higher opinion divergence is positively correlated with future returns in the three months after the announcements, which they interpret as supporting the Varian (1985) risk story. Our finding of higher abnormal returns after individuals buy stocks with higher earnings uncertainty seems consistent with the results of Garfinkel and Sokobin. Diether, Malloy, and Scherbina (2002) and Dimitrov, Jain, and Tice (2006) provide evidence consistent with the Miller prediction. 16

19 institutional holdings. So far, however, there has been no direct evidence using individual trading data in the U.S. that is consistent with this idea. In particular, Hirshleifer, Myers, Myers, and Teoh (2002) hypothesize that if the drift reflects mispricing, then more sophisticated investors (i.e., institutions) should buy immediately after good news (before an upward drift) and vice versa after bad news. They conjecture that naïve individual investors would take the opposite side of these transactions, and their trading would slow down the adjustment of prices to the information. Hirshleifer et al. investigate this idea using a sample of retail clients of a discount broker, but conclude that their data does not support it. In Table 9 we sort all stocks each quarter according to the analysts earnings surprise measure and put the stocks in five categories (quintile 1 is the most negative surprise and quintile 5 the most positive surprise). We look at net individual trading for each stock over a certain period, and present in Panel A the average of the net individual trading measure of all the stock-quarters in each of the different quintiles (with t-statistics testing the hypothesis of zero net individual trading). We observe a news-contrarian pattern in the post-event period: individuals buy those that experience bad news (quintile 1) and sell those that experience good news (quintile 5). In Panel B we observe an opposite pattern for institutions in that they behave in a news-momentum manner: they sell after the event those stocks that experience bad news and buy those stocks that experience good news. The results in Table 9 seem consistent with the idea that individuals trade in the direction that would slow down the adjustment of prices to the news after earnings announcements. However, Kaniel, Saar, and Titman (2007) show that individual investors generally trade in a contrarian fashion. If prices move prior to the earnings announcements to reflect information that would only later be announced publicly, it is possible that the results in Table 9 simply show the tendency of individuals to trade in response to price patterns as opposed to trading in response to the public release of news. 17

20 To differentiate between these two potential effects, we look at net trading by individuals and institutions during the event window and in the post-event period conditional on two variables: earnings surprise and abnormal return prior to the earnings announcement. Conditioning on return before the event is meant to enable us to discern whether investor behavior is motivated by past prices (e.g., the contrarian tendency of individual investors documented in Kaniel, Saar, and Titman (2007)) or whether it is motivated by the news itself, where we use the earnings surprise measure as a proxy for the news content of the announcement. We sort all events according to earnings surprise and put them in five groups: quintile 1 is the most negative surprise and quintile 5 is the most positive surprise. We also independently sort on cumulative abnormal return in the three months prior to the event. 22 Panel A of Table 10 shows a very clear picture: individuals trade during the event window predominantly in response to prior price patterns, not the earnings surprise. IndNT [0,1] is positive and significant (i.e., individuals buy) across the first line of the panel that corresponds to the quintile of stocks that experienced the most negative return before the event, but there is not much difference between net individual trading of bad news and good news stocks. Similarly, individuals sell irrespective of the earnings surprise if the return before the event was positive (i.e., abnormal return quintile 5). In other words, individuals simply behave as contrarians. Panel B of Table 10 looks at IndNT [2,61], or the trading of individuals during the post-event period. Here we observe a more complex behavior. It is still the case the individuals behave as contrarians: sell (buy) stocks that went up (down) in price before the event. However, there is also a news-contrarian effect whereby individuals buy more of the stocks that went down in price and had bad news than stocks that went down 22 The period over which we consider return prior to the event is somewhat arbitrary, but we present the analysis using three months of return before the event because we are also using three months of return after the event. We repeated the analysis conditioning on 20-day and 10-day returns prior to the events, and our conclusions did not change: the same (statistically significant) patterns were found conditioning on these two shorter periods. 18

21 in price but had good news. Similarly, for stocks that had the highest return before the event, individuals sell less of those stocks with bad news than those with good news. The statistical significance of the news-contrarian effect is demonstrated in the last column of the table (Q5 Q1). It is interesting to note that individuals seem to be much more active in the cells (Q1,Q1) and (Q5,Q5) of the table: the dogs and angels cells. The dogs had both the most negative return before the event and bad news, and individuals buy them almost twice as much as they buy stocks in other cells of the table (that either experience negative return or bad news). The angels had both the most positive return before the event and good news, in which case individual sell them almost twice as much as they sell the stocks in any other cell in the table. 23 Intense individual buying or selling therefore seems to be shaped by both past return and news in a contrarian fashion. The trading of institutions is reported in Table 11. Panel B shows two significant patterns in the behavior of institutions in the post-event period. The first is a momentum effect, where they buy stocks that experience positive returns before the event and sell stocks that experience negative returns. The second effect is a news-momentum effect: they buy (sell) much more of the stocks that both went up (down) in price prior to the event and had good (bad) news than those that went up (down) in price and had bad (good) news. The analysis in the Q5 Q1 column shows that the news-momentum pattern is statistically significant. The behavior of institutions in the post-event period therefore seems to mirror that of the individuals. As we mention in Section II.C., the magnitude of the change in prices at the time of the event could be viewed as an alternative measure of earnings surprise. In particular, the sell-side analysts forecasts that we use to construct the ES measure can be nonrepresentative of the entire market s expectations, biasing the surprise measure. 23 We have verified that the differences between these cells two cells and other cells in the table are statistically significant. 19

22 Therefore, the news-contrarian or news-momentum patterns we observe in the post-event period could also be responses to price changes during the event window itself. 24 We believe that conceptually, however, a post-event trading pattern that goes in opposite direction to returns at the time of the announcement should not be simply labeled as contrarian or momentum (i.e., as a response to prices rather than news) because at the time of the announcement both the price adjustment and the analysts earnings surprise measure are proxies for the same thing the change in beliefs of market participants. To use both proxies for earnings surprise in a single framework we regress net investor trading in the post-event period ([2,61]) on abnormal returns before the event (to control for contrarian tendencies), and both the ES measure and abnormal returns during the event window (as proxies for surprise). As in Table 6, we use a methodology in the spirit of Fama and MacBeth (1973) whereby we run a cross-sectional regression separately for each quarter, and then use time-series variability in the 16 estimates we have for each coefficient to compute standard errors. We present models where pre-event abnormal returns and net individual trading are measured over either 10 days or 60 days before the announcements, and show the results of the regressions for the entire sample and separately for three size groups (small, mid-cap, and large stocks). The results for the entire sample in Panel A of Table 12 indicate that net individual trading after the event is negatively related to both proxies for earnings surprise (ES and CAR [0,1] ) in the specification with a 60-day pre-event period. Looking at separate regressions by size groups, the analysts earnings surprise is not significantly related to post-event net individual trading, while the other proxy for surprise, the abnormal return at the time of the event, is significant in all size groups. 25 The contrarian 24 We find similar patterns of trading in the post-event period when we double-sort on CAR [0,1] (as an alternative measure of earnings surprise) and CAR [-60,-1]. 25 The results of the regressions suggest that the while individual investor behavior in the post-event period seems to respond to news released during the event, it is conceivable that part of the reaction is to the 20

23 pattern (negative relation between post-event net individual trading and pre-event returns) is observed in mid-cap and large stocks. The pattern we observe whereby individuals trade after the event in opposite direction to the news has the potential to impede the adjustment of prices to the news. In fact, combining the results in sections III and IV could suggest that individual investors prior to the event buy (sell) the stocks that will experience high (low) abnormal returns following the event, and then reverse their positions in the post-event period. Such a trading strategy could potentially be profitable, and at the same time it could also slow down the adjustment of prices after the event and give rise to (or sustain) the drift. Unfortunately, our data do not enable us to identify specific individual investors and observe their strategies. Our net individual trading measure represents a fictitious aggregate or representative individual investor, and therefore we cannot say for sure that the profitable strategy above is actually pursued by certain traders. It is, however, consistent with the relationships between return and trading that we observe. While our results suggest that individual investor trading could in fact contribute to the formation of a drift, the interpretation of the results seems very different from the hypothesis discussed by Hirshleifer et al (2002). They propose that individuals are naïve investors who lose money by trading in opposite direction to the drift, while institutions trade as if they know a drift exists and are taking advantage of it. We, on the other hand, show evidence consistent with the idea that while individuals on the NYSE trade in a direction that slows down the adjustment of prices after earnings announcements, they could be reversing potentially profitable positions to which they have entered before the events, a behavior that could hardly be characterized as naïve. pattern of price adjustment to the news at the time of the event. We do not make an attempt here to try to separate these two (potentially inseparable) effects. 21