Research Support School of Business Administration October 1993 OPTION TRADING AND EARNINGS NEWS DISSEMINATION Working Paper #9405-09 Kaushik I. Amin and Charles M. C. Lee University of Michigan

I

Option Trading and Earnings News Dissemination Kaushik I. Amin* University of Michigan Charles M. C. Lee* University of Michigan August 22, 1993 Current Draft: October 12, 1993 Preliminary Work Comments welcomed Abstract: Option market activity increases by more than 10% in the four days before quarterly earnings announcements. We show that the direction of this preannouncement trading foreshadows the subsequent earnings news. Specifically, we find option traders initiate a greater proportion of long (short) positions immediately before good (bad) earnings news. Returns to "active-side" option trades executed immediately prior to earnings announcements are significantly higher than normal. Collectively, these findings suggest option trading facilitates the dissemination of earnings news by providing informed investors with a low cost trading alternative. * Both authors are from the School of Business Administration, University of Michigan, Ann Arbor, MI 48109-1234. We thank Charles Jones for providing research assistance. James Myers and Paul Seguin provided helpful comments. Lee is grateful for financial support from the KPMG Peat Marwick Foundation and the Sanford R. Robertson Assistant Professorship.

Abstract Option market activity increases by more than 10% in the four days before quarterly earnings announcements. We show that the direction of this pre-announcement trading foreshadows the subsequent earnings news. Specifically, we find option traders initiate a greater proportion of long (short) positions immediately before good (bad) earnings news. Returns to "active-side" option trades executed immediately prior to earnings announcements are significantly higher than normal. Collectively, these findings suggest option trading facilitates the dissemination of earnings news by providing informed investors with a low cost trading alternative.

1. Introduction Equity option contracts are redundant assets, in the sense that their returns can be dynamically replicated by a portfolio of riskless bonds and shares of the underlying stock. However, as Grossman (1988, p.275) observed, "the notion that a real security is redundant when it can be synthesized by a dynamic trading strategy ignores the informational role of real securities." If option markets provide a lower cost trading alternative for informed traders, it is likely that the availability of option markets enhances the informational efficiency of equity markets in general. This study investigates the informational role of option markets in the dissemination of earnings news. Prior studies have examined the effect of option markets by analyzing stock market reactions to earnings news conditional on the availability of option trading.1 A key finding from these studies is that option trading increases the informational efficiency of the stock market. When option trading is available, the stock price adjustment is faster [Jennings and Starks (1986)], market price reactions to earnings news are smaller [Skinner (1989), Ho (1993)], and post-announcement price drifts are less pronounced [Botosan and Skinner (1993)]. While this literature shows option trading contributes to overall price efficiency, it does not address how such contributions are made. By focusing solely on the stock market reaction, prior research offers few insights on the informational role of option trading. Our approach in this study differs in that we do not use the existence of the option market as a conditioning variable. Instead, we examine option trading activities directly. Specifically, we investigate the speed, direction and profitability of option trades immediately around the release of different earnings news surprises -- i.e., "good" and "bad" news earnings. Using transactional data from the Berkeley Options database, we document several new findings that suggest an informational role for option markets in news dissemination. First, we find that option market trading volume increases several days before earnings announcements. In contrast to equity markets, where abnormal trading does not begin until the announcement date, option volume is 10 to 15% higher up to four days before the announcement. As in equity markets, options activity remains higher than normal for 1 For example, Skinner (1990), Ho (1993), Jennings and Starks (1986), Botosan and Skinner (1993)]. Also related are general studies of the effect of derivative products on stock price behavior [e.g., Figlewski and Webb (1993), Damodaran and Subrahmanyam (1992), and Skinner (1989)].

several days after the announcement. However, controlling for contemporaneous increases in equity trading, we show option market activity is disproportionately high in the three to four days before the earnings news release. The timing of earnings news announcements is quite predictable,2 and our findings show that option traders adjust their positions in anticipation of these announcements. Our second result is that the increased preannouncement trading in the option market is not simply a volatility play. We find that option traders anticipate not only the magnitude of the price reaction [Patell and Wolfson (1981)], but also its direction. Specifically, we show that the direction of preannouncement option trades foreshadows the nature of the earnings news, such that a greater proportion of long (short) positions are initiated by active-side traders immediately before good (bad) earnings news. This finding suggests that some traders have superior information prior to earnings announcements, possibly obtained through information leaks [Seppi (1992), Seyhun (1992)], or increased information collection activities [Kim and Verrecchia (1991)]. The directional volume of preannouncement option trades contrasts with Lee (1992), who finds no evidence of "informed" trading in the buy/sell imbalance of equity trades prior to earnings announcements. Therefore, with Lee (1992), our findings show that "informed" traders may prefer to trade in the option (versus equity) market, prior to earnings announcements. Since options and equity are redundant assets in perfect markets, we posit that the differential cost structures across the two markets lead a greater proportion of informed traders to trade in options. Consistent with the conjectures of Black (1975) and Cox and Rubinstein (1985), our evidence shows that investors with certain private information (in this case, news about the forthcoming earnings release) gravitate toward the option market. Finally, we investigate gate the profitability of active-side trading in options. If option markets play a role in price discovery, then option (and stock) prices should respond to active-side option trades -- i.e., buyer-nitiated (seinfidati ed) option trades should precede price increases (decreases). This general phenomenon will result in short-run profits (ignoring bid-ask spreads) for active-side option traders. To assess the impact of earnings news, we compare profits (losses) to acdtive-side option trading during earnings announcements and 2 Using prior release dates, Kross and Schroeder (1984) show that over 80 percent of earnings announcements are within three days of the date predicted. Anecdotal evidence from discussions with market participants suggests some traders may have even more precise information about the timing of releases. 2

during a matched sample of "pseudo-announcements." Specifically, for each actual news announcement, we generate a pseudo-announcement using the same firm and time-of-day, but a random date. We then compare active-side profits on the 50 trades immediately before and after the actual announcements to the profits on the corresponding trades around pseudo-announcements. Using a simple trading strategy, we open positions based on the buy or sell direction of option trades and unwind them at the end of the second day of trading after the announcement (or pseudo-announcement). We find that active-side trades executed at mid-spread prices generate short-term returns of approximately 3% during non-announcement (i.e., pseudo-announcement) periods. This suggests that investors who initiate trades on the option market bring information to market during non-announcement periods. Moreover, we show that the profits to active-side trades are significantly higher during earnings announcements. Investors who initiated option trades immediately before earnings announcements earn a return of approximately 6% by the end of day +2 -- that is, 3% more than matched pseudo-announcement trades. If these option trades had been executed on the equity market, a lower (but still statistically significant) average return of 0.2% results. These findings show that the information brought to market by active-side option trades was not previously impounded in the stock price. In other words, option traders do participate in the price discovery process associated with the dissemination of earnings news. We find, however, that option trading around earnings news is on average unprofitable, because of wide bid-ask spreads. The quoted bid-ask spread on a typical option contract is 6-12% of the contract price, as compared to spreads of less than 1% for equities [e.g., see Phillips and Smith (1980) or Vijh (1990)]. John, Koticha, and Subrahmanyam (1991) posits that liquidity suppliers in the option market seek compensation for bearing greater information risk by charging more for dealer services. Consistent with this hypothesis, we find that after accounting for bid-ask spreads, option traders using the above strategy realize an average trading loss of almost 10%. This shows that, after transaction costs, investors who initiate option trades do not earn abnormal profits. In summary, extensive research has focused on the equity market reaction to earnings news. However, with the exception of Schachter (1988) and Anthony (1988a), this 3

research does not examine how earnings information affects option trading.3 Our study shows that when option trading is available, the option market is an important vehicle of response to earnings news. Not only does option trading increase around earnings announcements, but we find this trading anticipates the nature and timing of the news release. Our findings are also relevant to the literature on the intermarket linkage between stocks and options. Recent evidence on the speed of information transfer between option and equity markets has been mixed [e.g., Stephan and Whaley (1990), Chan, Chung and Johnson (1993)]. These studies employ Granger-Sims causality tests, without conditioning on an exogenous information signal. In contrast, we isolate abnormal trading activities that relate to a known information signal and show faster and more informed trading in the option market. Our evidence suggests that option markets play an informational role by facilitating price discovery in the underlying security during periods of news dissemination. The remainder of the paper is organized as follows. The next section formulates the hypotheses; Section 3 describes the sample and data; Section 4 reports results of the empirical tests; and Section 5 concludes. 2. Development of Hypotheses Stock options are derivative financial contracts that offer potentially reduced transaction costs and increased financial leverage, relative to trading the underlying stock. A call (put) option gives its owner the right to buy (sell) a fixed number of shares of a specified stock at a fixed price at any time on or before a given date. The return patterns of option contracts can be replicated by a dynamically adjusted portfolio of riskless bonds and shares of the underlying stock. Therefore, option contracts and shares of the underlying stock differ mainly in terms of their transaction costs. With concurrent trading in what are essentially redundant assets, we expect investors to select the market with lower overall costs.4 Many authors [e.g., Black (1975), Cox and Rubinstein (1985), Manaster and Rendleman (1982), and Skinner (1990)] have argued that informed traders may prefer the cost 3 Schachter (1988) shows that open interest declines prior to earnings announcements, Anthony (1988a) reports daily abnormal volume for a sample of ten firms. Neither study examines the direction or speed of the option market trading activity. 4 We define transaction costs broadly to include interest lost on margin accounts or short-sell proceeds etc. 4

structure in the option market.5 In brief, relative to trading in the underlying equity, option traders may find more favorable implicit borrowing rates, lower margin requirements, and less stringent restrictions on short-selling (both in terms of the "iUptick Rule" in equity markets and the interest on the proceeds of short-selling). Moreover, certain types of private information, such as information related to future volatility of stock prices, is much easier to exploit in the options market. For traders with private information and wealth constraints, these differences translate into greater leverage per dollar invested. Traders with short-lived private information about upcoming news events, in particular, should gravitate toward the option market. Thus, the option market may be preferred by informed traders around earnings releases. In this study, we investigate three issues related to the role of option trading in the dissemination of earnings news. First, we document the extent and speed of option volume increases around the release of earnings news. That is, we examine the extent to which traders use the option market as an alternative vehicle of response to earnings news. The abnormal trading volume across the two markets can be used to assess the relative speed of adjustment in the options and equity markets. Second, we use directional trading volume to examine whether option traders are better informed about earnings information than stock traders. Finally, we evaluate the economic incentives for trading in the option market by studying the returns to option trades initiated immediately around earnings announcements. 2.1 Do Option Traders React to Earnings News? Our first goal is to provide large sample evidence on the extent of option trading associated with earnings news releases. Schachter (1988) shows that open interest declines prior to earnings announcements, particularly for contracts whose values are most sensitive to volatility. But, since open interest measures only the total number of contracts outstanding, it does not capture investor reaction associated with earnings information.6 Anthony (1988a) studies option trading around earnings news, using a sample of only 10 firms. He does not examine directional volume and its relation to earnings news. We use a sample 5 By the same token, Vijh (1990) observed that noise traders who mistakenly believe they have pricerelevant information might also prefer the option market. 6 An option trader may initiate one of four transactions: open a purchase contract, open a sale contract, close a purchase contract, and close a sale contract In the first two instances, the trader buys an option he does not already hold as a writer, thus increasing open interest In the latter two instances, he cancels his position as a writer of some options, thereby reducing open interest Consequently, increased trading in options could either increase or lower open interest, depending on whether contracts are being opened or closed. 5

consisting of all firms cross-listed on the American or New York Stock Exchange (AMEX or NYSE) and the Chicago Board of Options Exchange (CBOE) to evaluate the extent of option trading around earnings news releases. Since the option market provides lower costs for some traders, we expect greater option activity around an earnings announcement. The extant evidence on the relative speed of price and volume adjustment across the two markets is mixed. Manaster and Rendleman (1982) and Battacharya (1987) suggest the option market leads the equity market, but Stephan and Whaley (1990) raise questions about their test design. Using daily data and the Granger (1969) and Sims (1972) causality test, Anthony (1988b) reports that trading volume in the option market leads trading volume in the equity market by one day. However, using intraday data, Stephan and Whaley (1990) report price changes and volume in the equity market lead the option market by fifteen minutes or more. Most recently, Chan, Chung and Johnson (1993) show that the Stephan and Whaley (1990) result on price changes is due largely to nonsynchronous trading problems in the option market. Using midspread quote prices, they show that price changes are approximately contemporaneous across the two markets. However, the volume results in Stephan and Whaley (1990) are not challenged. In contrast to these earlier studies, we examine the speed of adjustment by conditioning on a news event: the Broad Tape release of quarterly earnings. Using a specific event time and intraday methodology allows us to evaluate both the speed and trade direction of volume reaction in the two markets. The exogenous earnings signal, time-stamped to the nearest minute, provides an alternative context in which to examine the relative speed of adjustment in the two markets. We hypothesize that the volume reaction in the option market will lead the equity volume, for two reasons. First, option traders might take anticipative positions before earnings announcements to adjust their exposure to price volatility risk. Earnings news are known to increase price volatility, and traders may wish to manage this risk with options.7 If option volume increases prior to earnings news due to volatility plays, option volume will lead volume in the equity market Secondly, to capitalize on their private information, better informed traders may gravitate toward the option market If price discovery occurs 7 Some option traders may place volatility bets which cannot be implemented by trading in stocks. Others may close out positions to avoid the earnings announcement risk. Note that even investors who hold hedged positions in options and the underlying stock are subject to increased risk during earnings announcements. This risk arises from the non-linear relation between the stock price and the option price which, during a large price move, will expose the holder to risk. 6

first in the option market, option volume will again lead equity volume. We investigate both possibilities. 2.2 Are option traders better informed? To assess the information brought to market by option traders, we examine the proportion of long and short positions being taken by "active-side" option traders immediately before earnings news. The option market on the CBOE is a continuous auction, in which competing market makers post tradable quotes. In a continuous auction microstructure, it is generally possible to identify an "active-side" to each trade, defined as the side which initiated the trade. As explained in Section 4, we use an algorithm similar to Harris (1989) and Lee and Ready (1991) to identify each trade as either buyer- or seller-initiated. We then classify option trades as either long and short positions relative to the underlying stock. A long position is the purchase of a call or the sale of a put option. A short position is the purchase of a put or the sale of a call option. If initiators of option traders have knowledge of forthcoming news, we expect to observe a greater proportion of long (short) position trades immediately before good (bad) news announcements. Prior studies suggest that the buy/sell direction of block trades [Seppi (1992)] and insider trades [Seyhun (1992)] anticipate earnings news. However, Lee (1992) shows that, on average, the buy/sell imbalance of equity trades immediately before earnings announcements does not foreshadow upcoming news. Using Lee's approach, we conduct a similar test with option data. If option markets help enhance overall price efficiency by allowing informed traders to trade at lower costs, we expect option traders to better anticipate the direction of earnings news than equity traders. 2.3 Do option traders make systematic profits? We can also assess the informativeness of option trades through their profitability. Earlier studies based on the Black-Scholes model suggest options markets may not be perfectly price efficient [ e.g., Galai (1977, 1978)]. However, later studies suggest that after transaction costs, it is difficult to make systematic abnormal returns by trading options. In particular, Phillips and Smith (1980) identify the bid-ask spread as the largest cost facing option traders. Empirically, Phillips and Smith show that quoted spreads on options are typically 6 to 10% of the contract price, while quoted spreads on stocks are typically less 7

than 1% of the stock price. If market makers set sufficiently wide spreads in options, option traders will not, on average, make abnormal trading profits. The finding that spreads are wider for option contracts than stocks is consistent with the presence of more informed traders in the option market. If options market makers face a greater risk of trading with informed traders than do their equity counterparts, bid-ask spreads should be wider in the option market.8 Faced with generally wider spreads, it is no longer clear whether informed traders are always better off trading in options. This tension between increased leverage and wider spreads is modeled by John, Koticha and Subrahmanyam (1991). Their model examines the vehicle of choice for informed traders when concurrent trading is available in options and equities. In equilibrium, they find that both informed and liquidity traders divide their trades between the option and equity markets. However, their model does not address the pattern of informed trading immediately before an information event. In theory, market makers should respond to increased information asymmetry risk by widening spreads prior to an anticipated earnings release.9 However, unless the timing of the event is fully anticipated. market makers are unlikely to set sufficiently wide spreads to deter all informed trading. Consequently, informed traders may still find an advantage in preannouncement trading, despite generally wider spreads. Empirically, we evaluate this proposition by examining the profitability of option trades before and after accounting for the bid-ask spread. 3. Data and Sample Selection The firms for this study were selected from the Institute for the Study of Security Markets (ISSM) and the Berkeley Options (CBOE) databases for 1988 and 1989. The ISSM data contain trade and quote information for all NYSE and AMEX firms and the CBOE data contains similar information for option contracts on firms listed at the Chicago Board of 8 Relative spreads may differ across the two markets due to factors other than information asymmetry risk. The market microstructure literature identifies three components to the bid-ask spread, of which information asymmetry risk is only one [see Stoll (1989) for a good summary]. The other two factors are inventory costs and order processing or administrative costs. However, most market makers in options close the day "even" and bear little inventory risk. Moreover, the capital requirements and initiation fees for market making in options are generally much lower than those needed to become a specialist in equity trading. Therefore, these two factors are unlikely explanations for wider options spreads. 9 Using intraday data, Lee, Ready and Mucklow (1993) show that in equity markets spreads widen and depths drop immediately before earnings news. Using daily closing quotes, Anthony (1987) finds no evidence of this in the option market As pointed out by Lee, Mucklow and Ready (1993) in an equity context, the failure to detect this effect may be due to the coarseness of daily data. In this draft, we do not examine the effect of earnings announcements on the size of the bid-ask spread in the option market We plan to investigate this issue in later drafts. 8

Options Exchange. A total of 147 firms were listed for the two full years on both the CBOE and the ISSM tapes. Of these, three were utilities, with regulated earnings of limited interest to traders; two firms were off-shore ADR's (Hanson PLC and Hitachi Ltd. ADR); one was Student Loan Marketing, for which only the non-voting shares are traded. Eliminating these 6 firms resulted in a final sample of 141 firms, which represents essentially all the cross-listed firms on the two markets over the study period. These firms are listed in Appendix A. Except for three AMEX companies (AMH, FRX, PLL), all our sample firms are listed on the NYSE. The CBOE data contain all option trades and quote revisions, time-stamped to the nearest second. For each put or call traded, the database contains the time of execution, expiration date, strike price, trade price and volume, as well as the stock price of the last equity trade at the time of option execution. Each quote revision contains the execution time, expiration date, strike price, and bid and ask prices, as well as the concurrent stock price.10 The Chicago Board of Options Exchange begins trading each day at 8:30 a.m. CST (9:30 a.m. EST) and concludes at 3:15 p.m. CST (4:15 p.m. EST). Option trading commences at the same time equity trading begins on the New York Stock Exchange, and ends 15 minutes after the New York market closes. Our study period pre-dates the multiple-listing of options on several exchanges, so our dataset captures the entire options volume for the sample firms. We obtained the intraday time of earnings announcements for our sample firms from the Dow Jones News Retrieval service (DJNS, or Broad Tape). All quarterly earnings announcements were identified for our sample of 141 firms. The DJNS news release provides the time of the release to the nearest minute. In case of multiple releases about the same earnings news, we select the first release after the fiscal quarter end. Since the DJNS begins each day around 6:45 a.m. CST and continues until approximately 5:45 p.m. CST, an announcement may occur before the option market opens or after it closes. For our daily analysis, we categorize announcements made when the CBOE is closed (hereafter, overnight announcements) as arriving immediately before opening the next day. 10 Besides trade and quote information on calls and puts, the CBOE data also report trading in straddles and spreads. A spread is a simultaneous long and short position in two or more contracts that differ in either their strike price (a money spread) or expiration date (a time spread). A straddle is a simultaneous long or short position in a put and a call with the same expiration date and strike price. Since the trading in spreads and straddles is minimal, these trades are excluded. 9

In some of our tests, we use the Value Line Investment Survey forecast of quarterly earnings per share (EPS) from the most recent issue immediately preceding the announcement date to proxy for market expectations about earnings. Announcements in which the actual reported EPS exceed the Value Line forecast are deemed "good" news, while those where actual EPS is lower than the Value Line forecast are deemed "bad" news. To ensure EPS is calculated consistently, we checked the reported EPS from the DJNS against the actual EPS from the Value Line issue immediately after the announcement. In the few cases where differences were found, we use the Value Line figure. Although all 141 firms are cross-listed, several firms had very little actual option trading volume. To reduce outliers caused by skewed non-announcement reference distributions, in some of our tests we required firms to meet minimal levels of option trading activity. Specifically, in our daily analyses, we required that firms have an average of at least one option trade per day, eliminating 9 firms from the sample. 4. Results 4.1 Daily volume analyses To evaluate trading volume across the two markets around earnings announcements, we first document the daily abnormal trading in each market Figure 1 reports the daily abnormal trading volume in the market for the underlying stock, as well as in call and put options. For this analysis, the event window is the 21 days centered on the DJNS announcement date. Daily abnormal volume for each firmn is defined as the actual daily volume (in contracts or shares traded), minus the mean daily volume for the firm from nonevent days, expressed as a percentage deviation from the non-event period mean. The reported average abnormal trading volume is the daily abnormal volume averaged across all firms. We report results at 1, 5 and 10% significance levels based on two-tailed crosssectional t-tests. Figure la shows that the stock market volume reaction is focused on day 0 and day +1, relative to the DJNS announcement. On these two days, the trading volume in stocks is approximately 35% higher than normal. The magnitude of this increase is comparable to volume increases found for large firms in earlier studies [e.g., Lee (1992), Bamber (1986, 1987)]. Consistent with prior work, we find that trading volume in the stock market remains persistently high for over a week after the earnings news release. Also consistent 10

with prior results, we find little evidence of increased stock trading in the days immediately prior to the DJNS announcement. Figures lb and lc report the abnormal volume reaction to earnings news in call and put contracts. These figures show that option wtrading increases around the release of earnings information. As with the stock market reaction, the abnormal volume in options is most pronounced on day 0 and day +1, when abnormal option volume is also approximately 35% higher than normal. These findings support the hypothesis that option markets provide traders with an alternative vehicle for responding to earnings announcements. In Figure 1, the most striking difference between the stock and option volume reactions are in the days leading up to the announcement. For both call and put options, we find increased levels of trading at least 3 days before the announcement. This finding is consistent with Anthony (1988a), who finds that option trading volume increases as early as 10 days before earnings announcements. This increased preannouncement volume may reflect speculation based on expectations of higher volatility, investors closing out positions, or informed trading. We examine this issue in our intraday analyses. Since trading across the option and equity markets are contemporaneously correlated, we also examine the option market volume reaction after controlling for stock volume. We present the results of this analysis in Table 1. The dependent variables for these regressions are the daily number of trades in either calls (CTrd) or puts (PTrd). The independent variables are the daily number of stock trades (STrd) and stock shares (SVol) transacted, as well as indicator variables for the 13 trading days around quarterly earnings announcements. All volume measures are expressed as percentage deviations from nonevent period means. Models 1 and 2 regress option volume on the event indicator variables; Models 3 and 4 regress CTrd or PTrd on the stock volume variables; and Models 5 and 6 combine both event period indicators and stock volume variables. Regressing on stock volume provides adjusted R2s of 42.3% and 16.6% for call volume and put volume, respectively. These results indicate that stock volume accounts for a significant portion of the daily variations in option market volume. Interestingly, controlling for the increase in stock volume, we find that the abnormal volume in options occurs primarily in the days before the DJNS news release (Models 5 and 6). On days -4 to -2, we observe unusually high option trading relative to the stock trading. Again, the evidence shows elevated levels of option trading immediately before earnings news. 11

42 Intraday Analyses As discussed earlier, the increased activity in options before earnings releases may be due to either volatility concerns or informed trading. If option traders have information about the upcoming earnings news, the direction of their preannouncement trades should be correlated with the earnings news. As a first test of this hypothesis, we compute the proportion of long and short positions taken on the options traded immediately around the release of earnings news. We define long and short positions in terms of the underlying stock. Therefore, a long position is the purchase of a call or the sale of a put option, while a short position is the purchase of a put or the sale of a call option. If initiators of option trades have foreknowledge of earnings news, we would expect a greater proportion of long (short) position trades immediately before good (bad) news announcements. To implement this test, we need an algorithm for classifying option trades into buyer- and seller-initiated transactions. Similar to Harris (1989) and Lee and Ready (1991), we infer the buy/sell direction of each trade by comparing the trade price to the prevailing quote prices for the given option contract We classify each trade as buyer-initiated if it is closer to the ask price and seller-initiated if it is closer to the bid price. Trades that are exactly at the mid-point of the prevailing spread ("midspread" trades) are considered indeterminable in direction and are excluded. Lee and Ready (1991) use a "tick test", based on prior price changes, to classify midspread trades in their study of equity trading. Since trading in a given option contract is much less frequent than it is in a stock, the tick test is not as reliable for option markets. Moreover, our preliminary tests indicate that midspread trades are far less frequent in options, so the advantage of the "tick test" for this study is minimal. Table 2 reports the frequency and proportion of long and short positions taken in the option market in the 50 trades immediately before and after earnings releases. For this analysis, good (bad) news events are those in which the actual reported earnings are greater (less) than the most recent Value-Line forecast. Announcements with no Value-Line forecast or with actual earnings equal to forecasted earnings are excluded. Long positions are buyerinitiated calls or seller-initiated puts; short-positions are seller-initiated calls or buyerinitiated puts. To simplify the table, we report trades in groups of ten. We include all trades classifiable as buys and sells between the beginning of trading on day -2 and the end of trading on day +2.11 11 We do not distinguish between different option contracts. In theory, an informed trader with capital constraints will obtain the greatest leverage by trading options with the highest price elasticity. However, 12

Consistent with Vijh (1988), Table 2 shows a greater proportion of long positions for both good news and bad news announcements.12 More importantly, we find good news announcements are preceded by a greater percentage of long position option trades than bad news announcements. Before good news announcements, 55.1% of option trades are long positions, compared to 53.5% for bad news announcements. The pattern is similar in the postannouncement period, with a greater proportion of long positions in the good news sample. Statistically, the difference is significant at the 1% level in a two-tailed Chisquared test. This finding suggests the direction of option trades foreshadows the subsequent earnings news. Table 3 shows results for the same test, using the sign of the price change, rather than the sign of the Value-Line forecast error, as an indicator of good or bad news. The price change is the change in stock price from the time of the announcement to the end of trading on day +2. This analysis eliminates two potential problems with the test in Table 2. First, Value-Line forecasts are a "noisy"' proxy for market expectations of earnings at the time of the announcement Value-Line forecasts may be stale (issued up to 7 weeks earlier) and may not reflect market "consensus" about earnings expectations. Thus, conditioning on the Value-Line forecast error may introduce noise about the nature of the news event, making informed trading more difficult to detect. Secondly, the DJNS announcement time may not be precisely aligned with CBOE trading times. If earnings news is publicly available to option traders before the DJNS time stamp, Table 2 findings may not represent privately informed option trading. By using the price change after the DJNS announcement time to classify announcements, we ensure that preannouncement trades took place before the subsequent price move. Table 3 shows that, when announcements are partitioned on the basis of the actual price change, we find a greater difference between the good and bad news samples. Once again, a greater proportion of long positions are taken before good news than before bad news. Before good news announcements, 56.6% of the option trades represent long positions in in practice these options are typically out-of-the money, and have the highest relative bid-ask spreads. Given the trade-off between leverage and transaction costs, it is unclear which contracts will be most attractive to informed investors. As a practical matter, we find a vast majority of the volume is in at- or near-the-money options, so we did not separately analyze the different contracts. 12 Vijh (1988) shows that a greater proportion of option trades are executed at the ask price than at the bid price. Since call volume is much higher than put volume, our results on long and short positions are consistent with his findings. Vijh (1988) does not speculate on why there are, on average, more buys than sells in options. We suspect this result may be due to the fact that some contracts expire without being exercised. 13

the underlying stock. However, before bad news announcements, only 52.4% of the option trades represent long positions. The z-statistic for this difference is 6.00. Table 3 results suggest that active-side option trades anticipate the nature of earnings news, but these results may be subject to two problems. First, the Chi-squared statistics may be biased upwards if option trades are positively serially correlated -- that is, long (short) positions tend to follow long (short) positions. Secondly, active-side trades may anticipate future price moves even when no earnings news is announcement If option prices are generally sensitive to order imbalances, then Table 3 results may not be related to earnings information. We address these issues in the next section using a pseudo-announcement methodology. 4.3 Trading profits In addition to examining the direction of option trades, we also examine their profitability. To compute trading profits, we assume the active-side of each trade and assess the profits under three different trading strategies. In the first strategy, we compute the "midspread return" (MsRet). That is, we assess the information content of buys and sells by assuming that positions can be taken at the average of the averate bid and ask price (midspread price) at the time of the trade. Thus, we compute returns using the midspread price at the time of the trade and the midspread price at the end of day +2 for that particular contract. In effect, we compute the profit th ththe trade initiator would realize without paying the spread. If traders are uninformed about upcoming price changes, then the expected return computed from midspreads should be close to zero. Conversely, if trade initiators are informed about future price changes, this strategy should yield a positive return. In the second strategy, we initiate the position at the actual trade price and unwind at the bid (ask) price at the end of day +2 for buys (sells). This strategy mimics the trading profit of an option trader who closes her position at the end of day +2 at the quoted price. In effect, it represents the trading profit after costs due to bid-ask spreads. However, since trades often take place inside the quoted spread, our measure understates the actual trading profit from this strategy. We refer to this return as the "realizable return" (ReRet). In the third strategy, we assume the long or short position by trading in the underlying stock. Each option trade is accompanied by the corresponding stock price based on the last trade in the equity market. We assume a long (short) position in the stock for purchases (sales) of calls, and a short (long) position in the stock for purchases (sales) of puts. We 14

then unwind the position at the closing stock price at the end of day +2 in computing the "equity return" (EqRet). This trading rule mimics the returns the option trader would have received in the equity market. Since the last equity trade could be either a buy or a sell, the returns do not include the cost of the bid-ask spread. Thus, EqRet is comparable to MsRet in the option market. In our analysis of trading profits, 3228 trades (around 7%) involve contracts that expire on or before day +2. For these trades, we do not have closing quotes two days after the announcement, so we substitute the value of the contract at expiration for the closing quote price. Specifically, we compute trading profits for these trades using the following option value at expiration: Call Options: Max [S - X, 0] Put Options: Max [X - S,0] where X is the strike price of the contract and S is the stock price based on the last trade on day +2. This approach attributes to the contract approximately the value that the holder would have obtained on expiration. Table 4 reports the results of the three trading rule tests for the 50 option trades immediately before and after the earnings announcement. The MsRet results are consistently positive and significant, indicating that option traders are informed about near future price changes. The magnitude of MsRet suggests that, ignoring transaction costs, option traders gain approximately 6% on their contracts in the days around the earnings release. In contrast, the ReRet results show that although initiators of option trades may anticipate the price move, high bid-ask spreads in the option market make such trades generally unprofitable. This result merely demonstrates that option traders do not, on average, make abnormal profits. It does not mean that traders are irrational for choosing to trade options. To assess whether these traders behaved rationally (i.e., optimally), we need more information about costs and benefits for comparable trades in the stock market Moreover, it is worth noting that approximately one third of the option trades had positive ReRets. For these traders, this simple strategy proved profitable. The EqRet results show that, ignoring bid-ask spreads, trades executed in the equity market would also yield positive returns. The t-statistics on the EqRet results are lower, but still 15

statistically significant This suggests that the information brought to market by the activeside option traders was not yet impounded in the stock price at the time of the option trade. However, Table 4 shows that EqRets are much smaller than MsRets. The average EqRet for preannouncement option trades is only around 16 basis points, compared to 600 basis points for MsRet. This difference reflects the greater leverage available in the option market. It also provides a potential explanation for why these trades are executed in the option market rather than in the equity market. As mentioned earlier, one possible concern with Table 4 is that the MsRet results are not attributable to the release of earnings news. Prior studies [e.g., Hasbrouck (1988), Lee and Ready (1991), Petersen and Umlauf (1991)] have shown that the active-side of each trade foreshadows subsequent price moves. That is, buyer (seller) initiated trades tend to be followed by increases (decreases) in stock prices. This pattern reflects the adverse selection problem faced by market makers. Since market makers are relatively uninformed, they infer the arrival of information from incoming active-side orders. That is, they respond to buyer-initiated trades by moving the price up and seller-initiated trades by moving the price down. This phenomenon may result in a positive MsRet even when no earnings news is being released. To examine the relative profitability of active-side trades across event and non-event periods, we generate a sample of "pseudo-announcements." For each announcement in our sample, we create a pseudo-announcement by retaining the same time-of-day and randomly drawing a date distribution of non-announcement dates for the same firm. For each pseudo-announcement, we also draw the nearest 50 trades immediately before and after the announcement time (provided these trades take place after the opening of trading on day -2 and before the close of trading on day +2). We then replicate the three trading profit tests using this random sample of pseudo-announcements. Table 5 reports the results for our pseudo-announcements. This table shows that the average MsRet during non-announcement periods is approximately 2.5%. Consistent with findings in the market microstructure literature on equity trading, the active-side of option trades do contain some ability to forecast future price changes even without an earnings announcement. Nevertheless, the information advantage of the active option trades is much higher immediately before the earnings news release. The midspread-to-midspread return on option trades (MsRet) is over 5%, or approximately twice as large, for option positions established immediately before earnings announcements. Evidently active-side option 16

trades are more profitable around the earnings release date. This finding is consistent with option traders establishing positions with foreknowledge of earnings news. Finally, to evaluate the incremental trading profit accruing to active-side option trades around earnings news, we subtract the returns on pseudo-announcement trades from the returns on trades around actual earnings announcements. Table 6 reports the average abnormal trading profit of option trades after controlling for corresponding trades from pseudo-announcements. The midspread returns in this table shows that even after controlling for the general information advantage of active-side trading, option traders make abnormal profits immediately before and after earnings announcements. For trades -30 to +20, active option trades earn abnormal profits of around 3 to 4 percent. Interestingly, the results for the realizable returns (ReRet) are also statistically significant and just slightly lower in magnitude, at approximately 3 percent. This suggests that spreads may have increased slightly during the announcement period, but even net of spreads gains to active trading increases around the earnings release. The equity return (EqRet) shows that the information advantage in option trades can also be detected using equity trade prices. 5. Conclusion This study investigates the abnormal trading volume in option markets around the announcement of earnings news. We show that option trading increases in the days immediately before and after earnings reports. As in the equity markets, this abnormal volume is most pronounced on day 0, relative to the DJNS report, and persists for several days after the earnings announcement On day 0, call option volume is 35% and put option volume is 55% higher than normal. However, after controlling for contemporaneous trading in the stock market, abnormal option trading is concentrated in the three to four days before the quarterly earnings announcement. We also investigate the extent to which the differential cost structures across the two markets result in an information asymmetry between traders in each market. We hypothesize that informed traders gravitate to the option market, where their short-lived proprietary news can be leveraged to yield higher returns. Consistent with this hypothesis, we find that option traders take a greater proportion of long (short) positions immediately before good (bad) earnings news. After controlling for a variety of transaction costs, we show that the information advantage that the option traders bring to market seems to be earnings-related. 17

Our findings suggest that for firms with listed options, a significant portion of the overall reaction in trading volume takes place in the option market. Extensive research has focused on the information content of earnings news. However, with the exception of Schachter (1988) and Anthony (1988a), this research has focused solely on the equity market reaction. Thus, our study provides the first large sample evidence on the effect of earnings news on option trading. We also provide new insights on the dissemination of earnings news. Prior studies have suggested that the availability of option markets enhances the overall information efficiency of the market for the underlying stock. Our results on the direction of option trades suggest that this enhancement arises, at least partially, from the role of option markets in the dissemination of earnings news. In particular, we show some option traders have advance information about earnings news. Finally, we provide new evidence on the economic role of the option market in the price discovery process. If relative cost structures lead informed traders first to the option market, then the option market may provide a cost-effective means for new information to be impounded in prices. As suggested by Grossman (1988), options are not merely redundant securities which can be substituted by dynamic'trading strategies in stocks and bonds. Rather, they provide a cost efficient mechanism for price discovery. Our findings on the speed and direction of option trades support this view. 18

References Anthony, Joseph H., 1988a, "Joint Option Equity Analysis: An Investigation of Information Trading In the Call Options Market," Proceedings: AAA Western Regional Meetings. Anthony, Joseph H., 1988b, "The Interrelation of Stock and Options Market TradingVolume Data," Journal of Finance. 43, 949-964. Anthony, Joseph H., 1987, "The Effect of Information Announcements on Bid/ask Spreads in the Call Options Market," Contemporary Accounting Research, 3,460-476. Bamber, Linda S., 1986, "The Information Content of Annual Earnings Releases: A Trading Volume Approach," Journal of Accounting Research, 24,40-56. Bamber, Linda S., 1987, "Unexpected Earnings, Firm Size, and Trading Volume Around Quarterly Earnings Announcements," Accounting Review, 25, 510-532. Battacharya, M., 1987, "Price Changes of Related Securities: The Case of Call Options and Stocks," Journal of Financial and Quantitative Analysis, 22, 1-15. Black, Fischer, 1975, "Fact and Fantasy in the Use of Options," Financial Analysts Journal, 31, 36-41 and 61-72. Botosan, Christine and Douglas J. Skinner, 1993, "The Relation Between Post-EamingsAnnouncement Drift and Options Listing," University of Michigan working paper, May. Chan, K., Y. Peter Chung and H. Johnson, 1993, "Why Option Prices Lag Stock Prices: A Trading-based Explanation," forthcoming, Journal of Finance. Cox, John C. and Mark Rubinstein, 1985, Options Markets, Prentice-Hall Inc., NJ. Damodaran, Aswath and Marti Subrahmanyam, 1992, "The Effects of Derivative Securities on the Markets for the Underlying Assets in the United States: A Survey," New York University working paper series, S-92-24. Figlewski, S. and G. Webb, 1993, "Options, Short Sales, and Market Completeness," Journal of Finance, 48, 761-778. Galai, Dan, 1977, "Tests of Market Efficiency and the Chicago Board Options Exchange," Journal of Business, 50, 167-197. Galai, Dan, 1978, "Empirical Tests of Boundary Conditions for CBOE Options," Journal of Financial Economics, 6, 187-211. Granger, Clive W. J., 1969, "Investigating Causal Relations by Econometric Models and Cross-Spectral Models," Econometrica, 37, 424-438. Grossman, Sanford, 1988, "An Analysis of the Implications for Stock and Futures Price Volatility of Program Trading and Dynamic Hedging Strategies," Journal of Business, 61, 275-298. 19

Harris, Lawrence, 1989, "A Day-end Transaction Price Anomaly," Journal of Financial and Quantitative Analysis, 24, 29-45. Hasbrouck, Joel, 1988, "Trades, Quotes, Inventories, and Information," Journal of Financial Economics, 22, 229-252. Ho, Li-Chin Jennifer, 1993, "Option Trading and the Relation Between Price and Earnings: A Cross-Sectional Analysis," The Accounting Review, 68,368-384. Jennings, Robert and Laura Starks, 1986, "Earnings Announcements, Stock Price Adjustment, and the Existence of Option Markets," Journal of Finance. 41, 107-125. John, Kose, Apoorva Koticha and Marti Subrahmanyam, 1991, "The Micro-Structure of Options Markets: Informed Trading, Liquidity, Volatility and Efficiency," New York University working paper. Kim, Oliver and Ro Verrecchia, 1991, "Market Reaction to Anticipated Announcements," Journal of Financial Economics, 30, 273-309. Kross, W., and D. Schroeder, 1984, "An Empirical Investigation of the Effect of Quarterly Earnings Announcement Timing on Stock Returns," Journal of Accounting Research, 22, 153-176. Lee, Charles M. C., 1992, "Earnings News and Small Traders: An Intraday Analysis," Journal of Accounting and Economics, 15, 265-302. and Mark J. Ready, 1991, "Inferring Trade Direction from Intraday Data," Journal of Finance, 46, 733-746., Belinda Mucklow and Mark J. Ready, 1993, "Spreads, Depths and the Impact of Earnings Information: An Intraday Analysis," Review of Financial Studies, 6,345 -374. Manaster, Steven and Richard J. Rendleman, Jr., 1982, "Option Prices as Predictors of Equilibrium Stock Prices," Journal of Finance, 37, 1043-1057. Patell, James M. and Mark A. Wolfson, 1981, "The Ex Ante and Ex Post Price Effects of Quarterly Earnings Announcements Reflected in Option and Stock Prices," Journal of Accounting Research 19,434-458. Patell, James M. and Mark A. Wolfson, 1979, "Anticipated Information Releases Reflected in Call Option Prices," Journal of Accounting and Economics. 1, 117-140. Petersen, M., and S. Umlauf, 1991, "An Empirical Examination of the Intraday Behavior of the NYSE Specialist," working paper, University of Chicago and London Business School. Phillips, Susan M. and Clifford W. Smith, 1980, "Trading Costs for Listed Options," Journal of Financial Economics, 8, 179-210. Schachter, Barry, 1988, "Open Interest in Stock Options Around Quarterly Earnings Announcements," Journal of Accounting Research. 26, 353-372. 20

Seppi, Duane J., 1992, "Block Trading and Information Revelation around Quarterly Earnings Announcements," Review of Financial Studies, 5, 281-305. Seyhun, Nejat H., 1992, "Effectiveness of Insider Trading Sanctions," Journal of Law and Economics, 35, 149-182. Sims, C. A., 1972, "Money, Income, and Causality," American Economic Review, 62, 540-552. Skinner, Douglas J., 1989, "Options Markets and Stock Return Volatility," Journal of Financial Economics. 23, 61-78. Skinner, Douglas J., 1990, "Option Markets and the Information Content of Accounting Earnings Releases," Journal of Accounting and Economics. 13, 191-211. Stephan, Jens A. and Robert E. Whaley, 1990, "Intraday Price Change and Trading Volume Relations in the Stock and Stock Option Markets," Journal of Finance, 45, 191-220. Stoll, Hans R., 1989, "Inferring the Components of the Bid-Ask Spread: Theory and Empirical Tests," Journal of Finance, 44, 115-134. Vijh, Anand M., 1988, "Potential Biases from Using Only Trade Prices of Related Securities on Different Exchanges: A Comment," Journal of Finance, 43, 1049-1055. Vijh, Anand M., 1990, "Liquidity of the CBOE Equity Options," Journal of Finance, 45, 1157-1179. Whaley, R. and J. Cheung, 1982, "Anticipation of Quarterly Earnings Announcements: A Test of Option Market Efficiency," Journal of Accounting and Economics, 57-83. 21

Figure 1 Abnormal trading volume around the release of earnings These graphs depict the abnormal trading volume around earnings news releases in the equity market, and in the market for call and put options. All firms cross-listed on the NYSE and the CBOE for both 1988 and 1989 are included. Abnormal trading is expressed as percentage deviation from the daily mean for each firm and averaged across all firms. Two-tailed significance levels: - 1%; M 5%; E3 10%. Fig la: Abnormal Trading in Equity Market q) cg S C~ "0 *5 4) *iI 7:$ 0 I C C 4) "0 Cs S I C C 4) 0 (4 4) p4 Cn 30 -20 10 -10 --20- " Fig lb: Abnormal Trading in Call Options 50 -40 -30 -20 -I I I I I I I I 1- Fig Ic: Abnormal Trading in Put Options 60 -40 0,-10- 1 --- —--— r --- —-20-_ o I I o I I I I I,-4 Date relative to Announcement of Earnings News

Table 1 Volume Reaction in Calls and Puts Around Earnings Releases This table reports the results of time-series regressions of call and put option trading volume on event period indicator variables and on trading volume in the equity market. All firms cross-listed on the NYSE and the CBOE for both 1988 and 1989 are included. The dependent variable is the daily number of trades in either calls (CTrd) or puts (PTrd). The independent variables are the daily number of trades (STrd) and shares (SVol) in the underlying security and indicator variables for the 13 trading days around quarterly earnings announcements as reported by the Dow Jones News Service (DUMi, i = -6 to +6). The trading volume in stocks, calls and puts are all expressed in terms of percentage deviation from the mean daily volume of each firm. Table values represent estimated coefficients from a regression of the form: +6 (CTrd or PTrd) = a + i=-6 pi DUMi + y1STrd + y2SVol + e Models 1 and 2 include only the indicator variables; Models 3 and 4 include only the stock volume variables; Models 5 and 6 combine both indicator and stock volume variables. Two-tailed significance levels at 1% (**) and 5% (*) are marked. Stock Event Period Indicator Variables Volume Model Dep. a Day Day Day Day Day Day Day Day Day Day Day Day Day STrd SVol Ad. Var. -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 R (%) 1 CTrd -2.91 3.94 5.32 8.32 13.2 15.6 14.2 33.0 29.6 20.5 15.4 9.1 7.6 8.1 - - 0.3 ** * ** ** ** ** ** ** ** * * *.. 2 PTrd -2.97 3.57 8.11 20.2 10.8 17.6 5.6 39.6 29.5 8.7 13.1 12.0 10.1 9.0 - - 0.2 3 CTrd 0.00 - - - - - - - - 1.47 0.11 42.3 4 PTrd 0.00 - - - - - - - 1.09 0.13 16.6 5 CTrd -0.12 -0.9 0.8 0.13 7.3 8.5 4.5 -4.3 -6.6 0.85 1.7 -2.0 -1.5 -1.1 1.47 0.11 42.3 6 * *r _.8 *._.. * * 4 *. I I ** ** 1. 6 PTrd -0.80 -0.1 4.8 14.1 6.1' 12. -1.8 0.2 1.0 -6.6 2.4 3.4 3.0 1.9 1.09 0.13 16.7 IItit I I III in I I

Table 2 Long and short positions taken in option contracts immediately around earnings news that either exceeds ("good news") or is less than ("bad news") the most recent Value-Line earnings forecast This table reports the frequency and proportion of long and short positions taken in the option market in the 50 trades immediately before and after "good" and "bad" news earnings releases. Long positions are buyer-initiated calls or seller-initiated puts; short-positions are seller-initiated calls or buyer-initiated puts. Good (bad) news events are those in which the actual reported earning is greater (less) than the most recent Value-Line forecast. The times of the announcements are obtained from the Dow Jones News Service (DJNS). All transactions classifiable as buys or sells are included provided they are executed within two trading days of the announcement time. The zstatistics are based on a Chi-squared test of the null hypothesis that the long and short positions taken are uncorrelated with the earnings signal. Good News Earnings Bad News Earnings Timing of trade Long Short Percent Long Short Percent Difference relative to DJNS positions positions long positions positions long (z-stat) announcement -50 to-41 802 699 53.4 737 619 54.4 -0.49 -40 to-31 945 713 57.0 880 598 59.5 -1.44 -30 to-21 1106 843 56.7 885 814 52.1 2.82 -20 to-11 1239 996 55.4 1043 924 53.0 1.57 -10 to -1 1449 1273 53.2 1339 1295 50.8 1.76 Total Pre- 5541 4524 55.1 4884 4250 53.5 2.20 announcement +1 to +10 1611 1357 54.3 1360 1353 50.1 3.13 +11 to+20 1372 1176 53.8 1311 1039 55.8 -1.36 +21 to +30 1253 1030 54.9 1144 893 56.2 -0.84 +31 to +40 1177 850 58.1 978 896 52.2 3.69 +41 to +50 1050 819 56.2 892 851 51.2 3.01 Total Post- 6463 5232 55.3 5685 5032 53.1 3.33 announcement

Table 3 Long and short positions taken in option contracts immediately around earnings news that resulted in either a positive return ("good news") or. negative return ("bad news") This table reports the frequency and proportion of long and short positions taken in the option market in the 50 trades immediately before and after "good" and "bad" news earnings releases. Long positions are buyer-initiated calls or seller-initiated puts; short-positions are seller-initiated calls or buyer-initiated puts. Good (bad) news events are those in which the actual equity market return from the announcement time to the end of day +2 is positive (negative). The times of the announcements are obtained from the Dow Jones News Service (DJNS). All transactions classifiable as buys or sells are included provided they are executed within two trading days of the announcement time. The z-statistics are based on a Chi-squared test of the null hypothesis that the long and short positions taken are uncorrelated with sign returns associated with the announcement. Good News Earnings Bad News Earnings Timing of trade Long Short Percent Long Short Percent Difference relative to DJNS positions positions long positions positions long (z-stat) announcement_ _ -50 to-41 882 682 56.4 748 678 52.5 2.16 -40 to-31 1029 690 59.9 925 674 57.8 1.18 -30 to-21 1099 881 55.4 1017 843 54.7 0.52 -20 to -11 1281 980 56.7 1113 1066 51.1 3.73 -10 to -1 1590 1272 55.6 1367 1431 48.9 5.04 Total Pre- 5881 4505 56.6 5170 4692 52.4 6.00 announcement +1 to +10 1649 1405 54.0 1472 1456 50.3 2.88 +11 to+20 1505 1181 56.0 1327 1109 54.5 1.12 +21 to +30 1338 1086 55.2 1127 988 53.3 1.29 +31 to +40 1184 986 54.6 1032 861 54.5 0.03 +41 to +50 1106 958 53.6 913 818 52.7 0.52 Total Post- 6782 5616 54.7 5871 5232 52.9 2.80 announcement___

Table 4 The Profitability of Option Trades Around Earnings Announcements This table reports the average returns obtained from taking the active side of the 50 option trades immediately before and after the release of earnings news. Results using three different trading rules are reported. MsRet is the average "mid-spread return" per trade, assuming trades are made at the middle of the bid-ask spread. For option buys (or sells), a long (or short) contract position is established at the middle of the prevailing spread at the time of the actual trade and unwound at the closing mid-spread price on day +2. The average "realizable return" (ReRet) per trade incorporates the bid-ask spread, so that buys (sells) are established at the actual trade price and unwound at the closing bid (ask) price on day +2. The average "equity return" (EqRet) is computed using the stock price in the equity market at the time of the option trade. For each option trade, a corresponding long (or short) position is taken in the equity market using the last equity trade price and unwound at the last equity trade price on day +2. All returns are expressed in percents. The times of the announcements are obtained from the Dow Jones News Service (DJNS). All transactions classifiable as buys or sells are included provided they are executed within two trading days of the announcement time. The t-statistics are based on a test of the null hypothesis that the average return is not significantly different from zero. Midspread Realizable Equity Returns Returns Returns Timing of trade No. of relative to DJNS Obser- MsRet t-stat ReRet t-stat EqRet t-stat event time vations (%) (%) (%) -50 to -41 3189 3.13 3.39 -9.74 -10.6 0.133 2.06 -40 to -31 3530 6.96 6.66 -7.04 -7.10 0.224 3.78 -30 to -21 4109 6.28 5.35 -9.22 -10.0 0.136 1.96 -20 to -11 4705 6.08 6.57 -9.69 -11.0 0.135 2.75 -10to-1 6000 5.02 7.30 -10.45 -15.1 0.207 4.69 +1 to +10 6436 7.42 9.49 -11.41 -16.8 0.102 2.70 +11 to +20 5477 5.61 7.74 -11.31 -16.9 0.092 2.13 +21 to +30 4834 4.96 6.10 -11.45 -15.6 0.063 1.22 +31 to +40 4340 4.20 4.60 -11.80 -13.4 -0.019 -0.38 +41 to +50 4023 3.30 4.01 -11.75 -14.8 0.017 0.26

Table 5 The Profitability of Option Trades Around Randomly Selected "Pseudo-Announcement" Dates This table reports the average returns obtained from taking the active side of the 50 option trades immediately before and after "pseudo-announcements". For each actual announcement, a "pseudoannouncement" is created using the same firm and time-of-day, but a randomly selected nonannouncement date. Results using three different trading rules are reported. MsRet is the average "mid-spread return" per trade, assuming trades are made at the middle of the bid-ask spread. For option buys (or sells), a long (or short) contract position is established at the middle of the prevailing spread at the time of the actual trade and unwound at the closing mid-spread price on day +2. The average "realizable return" (ReRet) per trade incorporates the bid-ask spread, so that buys (sells) are established at the actual trade price and unwound at the closing bid (ask) price on day +2. The average "equity return" (EqRet) is computed using the stock price in the equity market at the time of the option trade. For each option trade, a corresponding long (or short) position is taken in the equity market using the last equity trade price and unwound at the last equity trade price on day +2. The times of the announcements are obtained from the Dow Jones News Service (DJNS). All transactions classifiable as buys or sells are included provided they are executed within two trading days of the announcement time. The t-statistics are based on a test of the null hypothesis that the average return is not significantly different from zero. Midspread Realizable Equity Returns Returns Returns Timing of trade No. of relative to DJNS Obser- MsRet t-stat ReRet t-stat EqRet t-stat event time vations (%) (%) (%) -50 to-41 2913 1.926 2.26 -10.856 -13.15 -0.018 -0.25 -40 to-31 3236 2.624 2.96 -10.678 -12.99 0.065 1.33 -30 to-21 3775 -0.106 -0.14 -13.511 -18.44 -0.092 -1.75 -20 to-11 4724 3.075 3.39 -11.890 -15.00 -0.011 -0.24 -10Oto-l1 5975 3.149 4.57 -12.869 -19.42 -0.037 -1.07 +1 to +10 6260 2.980 5.37 -15.037 -27.63 -0.092 -2.70 +11 to+20 5167 2.200 4.31 -13.813 -27.22 -0.019 -0.58 +21 to +30 4418 3.613 5.61 -12.277 -19.92 0.017 0.48 +31 to +40 3891 2.955 4.12 -12.067 -17.63 -0.380 -1.70 +41 to +50 3450 1.849 2.43 -12.224 -17.08 -0.390 -1.54

Table 6 Abnormal Profitability of Option Trades Around Earnings Releases After Controlling for Matched-sample "Pseudo-announcments" This table reports the average abnormal returns obtained from taking the active side of the 50 option trades immediately around earnings releases. For each announcement, a "pseudo-announcement" is created using the same firm and time-of-day, but a random non-announcement date. Each trade is matched with a corresponding trade from the pseudo-announcement. Abnormal return is defined as the actual return per trade, minus the return on the matching trade, averaged across all paired trades. Three different trading rules are used. MsRet is the average "mid-spread return", assuming trades are made at the middle of the bid-ask spread. For option buys (or sells), a long (or short) contract position is established at the middle of the prevailing spread at the time of the trade and unwound at the closing mid-spread price on day +2. The average "realizable return" (ReRet) incorporates the bid-ask spread, so that buys (sells) are established at the actual trade price and unwound at the closing bid (ask) price on day +2. The average "equity return" (EqRet) is computed using the stock price in the equity market at the time of the option trade. For each option trade, a corresponding long (or short) position is taken in the equity market using the last equity trade price and unwound at the last equity trade price on day +2. All transactions classifiable as buys or sells are included provided they are executed within two trading days of the announcement time. The t-statistics are based on a test of the null hypothesis that the average return is not significantly different from zero. Midspread Realizable Equity Returns Returns Returns Timing of trade No. of relative to DJNS Obser- MsRet t-stat ReRet t-stat EqRet t-stat time of actual vations (%) (%) (%) announcemnt -50 to -41 1537 0.22 0.13 1.01 0.63 0.178 1.23 -40 to -31 2467 1.68 1.06 0.99 0.66 0.007 0.08 -30 to -21 3272 3.24 2.14 2.93 2.09 0.167 2.00 -20 to-11 4002 4.20 3.04 3.27 2.59 0.138 1.88 -10 to -1 5468 2.30 2.37 2.83 3.01 0.207 3.51 +1 to +10 5906 4.19 4.38 3.75 4.21 0.236 4.40 +11 to+20 4740 4.08 4.12 2.91 3.13 0.230 1.63 +21 to +30 4067 0.96 0.82 0.15 0.14 0.244 1.53 +31 to +40 3086 2.03 1.42 1.31 0.97 0.467 1.63 +41 to +50 1996 -3.59 -2.45 -4.05 -2.87 -0.077 -0.65

APPENDIX A - List of 141 Sample Firms 02224910 01447610 02635110 28176010 02687410 02680410 02390510 03189710 03190510 03251110 04882510 05430310 02581610 09702310 06605010 07181310 11704310 09738310 09179710 07785310 05873210 07159310 11009710 12189710 08750910 07390210 17119610 20491210 12484510 13985910 19121910 17303410 17052010 21236310 19044110 15852510 12550910 19416210 20536310 AA AAL AGC AGE AIG AIT AMH AMP AN APC ARC AVP AXP BA BAC BAX BC BCC BDK BEL BLY BMG BMY BNI BS BSC C CA CBS CCB CCE CCI CCN CDA CGP CHA CI CL CSC ALUMINUM CO AMERICA ALEXANDER ALEXANDER SER AMER GEN CORP A.G.EDWARDS INC AMERICAN INTL GROUP INC AMERICAN INFORMATION TECH CO AMDAHL CORP AMP INC AMOCO CORP ANADARKO PETROLEUM CORP ATLANTIC RICHFIELD CO AVON PRODUCTS INC AMERICAN EXPRESS COMPANY BOEING CO BANKAMERICA CORPORATION BAXTER INTERNATIONAL INC BRUNSWICK CORP BOISE CASCADE CORP BLACK DECKER CORP BELL ATLANTIC BALLY MFG CORP BATTLE MOUNTAIN GOLD CO CLA BRISTOL MYERS CO BURLINGTON NORTHERN INC BETHLEHEM STEEL CORP BEAR STEARNS COMPANIES INC CHRYSLER CORP COMPUTER ASSOCIATES INTL INC CBS INC CAPITAL CITIES /ABC INC COCA-COLA ENTERPRISES INC CITICORP CHRIS-CRAFT INDUSTRIES INC CONTROL DATA CORP COASTAL CORP CHAMPION INTERNATIONAL CORP CIGNA CORP COLGATE-PALMOLIVE CO COMPUTER SCIENCES CORP 24736110 25365110 26353410 25384910 25468710 26054310 26188510 29284510 27746110 27805810 34537010 31330910 31818110 34386110 31945510 34583810 35671410 36935210 36955010 36960410 21932710 37044210 39109010 36476010 40018110 39144210 36868210 40621610 43739D10 43507110 43761410 42307410 43850610 41387510 44485910 42823610 32054810 45920010 45950610 DAL DBD DD DEC DIS DOW DRY EC EK ETN F FDX FFC FLR FNB FRX FTX GCN GD GE GLW GM GNN GPS GQ GWF GY HAL HFD HIA HM HNZ HON HRS HUM HWP I IBM IFF DELTA AIR LINES INC DEL DIEBOLD INC DU PONT DE NEMOURS E I CO DIGITAL EQUIPMENT CORP DISNEY WALT COMPANY DOW CHEMICAL CO DREYFUS CORP ENGELHARD CORP EASTMAN KODAK CO EATON CORP FORD MOTOR CO FEDERAL EXPRESS CORP FIREMANS FUND CORP FLUOR CORP FIRST CHICAGO CORP FOREST LABS INC FREEPORT MCMORAN INC GENERAL CINEMA CORP GENERAL DYNAMICS CORP GENERAL ELECTRIC CO CORNING GLASS WORKS GENERAL MOTORS CORP GREAT NORTH NEKOOSA CORP THE GAP INC GRUMMAN CORP GREAT WESTERN FINANCIAL CO GENCORP INC HALLIBURTON CO HOMEFED CORP HOLIDAY CORP HOMESTAKE MINING CO HEINZ H.J CO HONEYWELL INC HARRIS CORP HUMANA INC HEWLETT-PACKARD CO FIRST INTERSTATE BANCORP INTL BUSINESS MACHINES CORP INTL FLAVORS FRAGRANCES

APPENDIX A (continued) 45988410 46014610 45067910 47816010 48258410 49238610 19121610 50545810 53802110 54385910 53271610 54042410 84474110 57777810 58013510 58505510 58283410 59018810 60405910 60705910 58933110 61166210 62886210 66680710 65584410 63764010 66581510 67634610 67459910 71344810 69642930 73109510 69562910 70990310 75127710 77434710 75511110 IGL IP ITT JNJ KM KMG KO LAC LIT LOR LTD LTR LUV MA MCD MDT MEA MER MMM MOB MRK MTC NCR NOC NSC NSM NT OG OXY PEP PLL PRD PWJ PZL RAL ROK RTN INTL MINERALS CHEM CORP INTERNATIONAL PAPER CO ITT CORP JOHNSON AND JOHNSON K MART CORP KERR-MCGEE CORP COCA-COLA CO LAC MINERALS LTD LITTON INDUSTRIES INC LORAL CORP THE LIMITED INC LOEWS CORP SOUTHWEST AIRLINES CO MAY DEPARTMENT STORES CO MCDONALDS CORP MEDTRONIC INC MEAD CORP MERRILL LYNCH CO INC MINNESOTA MNG MFG CO MOBIL CORPORATION MERCK CO INC MONSANTO CO NCR CORP NORTHROP CORP NORFOLK SOUTHERN CORP NATIONAL SEMICONDUCTOR CORP NORTHERN TELECOM LTD OGDEN CORP OCCIDENTAL PETROLEUM CORP PEPSICO INC PALL CORP POLAROID CORP PAINE WEBBER GROUP INC PENNZOIL CO RALSTON PURINA CO ROCKWELL INTL CORP 1 PAR RAYTHEON CO 81238710 82270360 82434810 83083010 80685710 87161610 03017710 87538210 87933510 87913110 89233510 89604710 88250810 90553010 90921410 91530210 90291110 91301710 93443610 97463710 96332010 96945710 93114210 96216610 30229010 98412110 S SC SHW SKY SLB SYN T TAN TDY TEK TOY TRB TXN UCC UIS UPJ UST UTX WCI WGO WHR WMB WMT WY XON XRX SEARS ROEBUCK CO SHELL TRANS TDG PUB LTD DS SHERWIN-WILLIAMS CO SKYLINE CORP SCHLUMBERGER LTD SYNTEX CORPORATION AMERICAN TEL TEL CO TANDY CORP TELEDYNE INC TEKTRONIX INC TOYS R US INC TRIBUNE COMPANY TEXAS INSTRUMENTS UNION CAMP CORPORATION UNISYS CORP UPJOHN COMPANY UST INC UNITED TECHNOLOGIES CORP WARNER COMMUNICATIONS INC WINNEBAGO INDS INC WHIRLPOOL CORPORATION WILLIAMS COMPANIES WAL-MART STORES INC WEYERHAEUSER COMPANY EXXON CORP XEROX CORPORATION it