Do option open-interest changes foreshadow future equity returns?

Transcription

1 Do option open-interest changes foreshadow future equity returns? Andy Fodor* Finance Department Ohio University Kevin Krieger Department of Finance and Operations Management University of Tulsa James Doran Bank of America Professor of Finance Department of Finance Florida State University January 13, 2010 *Corresponding Author: Copeland Hall 234 Athens, OH

2 Abstract Recent work has considered whether information is simultaneously reflected in both option and equity markets. We provide new evidence supporting Black s (1975) conjecture that information is first revealed in option markets. Specifically, changes in call and put open interest levels have predictive power for future equity returns. Large increases in put open interest are followed by poor equity returns. Call open interest increases precede relatively strong future returns but the relationship is considerably less pronounced. The recent change in the call-to-put open interest ratio most strongly predicts equity returns over the following few weeks, even after controlling for traditional factors.

3 1. Introduction The information flow between equity and option markets has received increased attention in recent years. Findings by Bali (2008), Cremers and Weinbaum (2008) and Doran and Krieger (2009) show that information flows from option markets to equity markets, reflected in volatility or option prices changes. These changes can forecast the direction of underlying equity price movements, depending on the location and degree of option information gathered from across the volatility skew. Several questions still remain. What are the mechanisms that drive changes in option markets and the subsequent change in equity markets? What characteristics are manifested by these mechanisms? Finally, what is driving the change? The development of literature relating the forecastability of equity returns to the trading characteristics of markets has proceeded along multiple fronts, both theoretical and empirical. Equity volume was initially investigated as a potential precursor to the movement of equity prices, but the potential leverage provided by option markets as first noted by Black (1975) allowed researchers an additional avenue to consider. Additional trading properties, including adjusted volumes, open interest, and put/call differentials have grown in visibility in recent years, and numerous researchers have utilized special data sets and investigated trading characteristics surrounding specific events. Easley and O Hara (1988) develop a theoretical model which demonstrates the importance of equity trade size in suggesting further movements in equity markets. Trade size is correlated with private information, and thus, those with superior information desire to execute notably larger trades. The model determines that transaction prices rise after block buys and decline after block sales as market makers anticipate large trades to be indicative of superior information. Along with Black s (1975) leverage argument for the sustainability of option markets, other authors have noted advantages to derivative trading. Cox and Rubenstein (1985), for example, describe potential savings in trading costs, and Diamond and Verecchia (1987) show the ability of option trading to overcome short-sale restrictions. Such developments encouraged Easley et al. (1998) to further extend a model which describes the importance of option volumes in forecasting equity prices. This model considers the impact of increased volume to be reflective of good news or bad news, and considers both calls and puts; thus, it is seen as an improvement beyond the initial work of Stephan and Whaley (1990). If option markets provide

4 a more favorable environment for trading, investors may first reveal their preferences there, and Easley et al. (1998) conclude that option volumes do, in fact, portend information about future stock prices. Informed traders who buy (sell) calls or sell (buy) puts carry positive (negative) information about future stock prices. In their empirical work, Easley et al. (1998) reject the hypothesis that option volumes carry no information about future stock price changes when they consider the aggregate of good news vs. bad news option trades. They do not expect overall option volume to be indicative of stock price movements as there are many motivations for option trades, but when they classify trades executing below (above) the halfway point in the bid-ask spread as sells (buys), the aggregate volume significantly Granger causes (see Granger, 1969 and Granger and Newbold, 1977) movements of stock prices over the following, short intraday periods using data from October and November of Black (1975) and Manaster and Rendleman (1982) initially theorized and demonstrated the use of option markets as a place for information trading based on increased available leverage. In recent years, using Hasbrouck s (1995) methodology for determining information share, Chakravarty et al. (2004) determine that option markets provide an average of 17% of information discovery in equity prices based on volume and various spreads. In order to do so, they describe the importance of focusing on the permanent component of stock price changes as this movement is indicative of the impact of real information. Using data from 60 firms over the period, Chakravarty et al. (2004) also find that out-of-the-money (OTM) options, which provide greater leverage to traders, contain the most information regarding future stockprice movements. Findings regarding the importance of option volume for future equity prices are not uniformly supportive of a linkage, even over very short intervals. Chan, Chung and Fong (2002), for example, find no lead effect of option volume for stock returns (though they do document a strong link of returns to stock-net-trade volume). Vijh (1990) determines that large option trades have a very small effect on equity prices, though Srinivas (1993) describes a considerably stronger link after taking exception to Vijh s sample selection. In an effort to reconcile previous efforts, Pan and Poteshman (2006) utilize data from and consider long-short stock portfolios formed based on put-call volume ratio. Investing in those firms with low put-call ratios while shorting those with high ratios yields highly statistically significant adjusted returns of 40 basis points the following day and over 100

5 basis points through the following week. These results, however, are based on a unique data set that notes whether the initiator of a trade is the buyer or seller and, furthermore, whether she is opening or closing a position. Given the private nature of the data, no market inefficiency need be in place. Pan and Poteshman (2006) find that option signals from full-service brokerage houses provide considerably stronger signals, which they find plausible given the concentration of hedge funds under the full-service umbrella. Alternatively, publicly visible option volume data, based on an algorithm similar to Easley et al. (1998) also holds predictive power for stock returns, but only for a day, and in regressions of future equity returns on both forms of information, the public portion of the information is subsumed by the impact of the private portion. Pan and Poteshman (2006) also determine that option signals from deep OTM options exhibit the greatest level of predictability while those generated from lower leverage contracts exhibit very little. This is verified by Blasco et al. (2009) who find informed options trading taking place in OTM options in the Spanish market. A further branch of the literature detailing the link between options market characteristics and equity returns considers the market characteristics surrounding certain events. Cao and Yang (2009) develop a theoretical model which demonstrates increased utility for investors with the introduction of options. They hypothesize that option trading volume should increase near public events, like mergers and acquisitions, earnings announcements, and credit rating changes. Furthermore, trading volume should be higher for optionable stocks as investors may use the equity to hedge their positions in options markets. A number of empirical papers have sought to explain the interaction of option market characteristics and equity returns surrounding such events. Amin and Lee (1997) find that more long (short) positions are undertaken in the options market immediately before positive (negative) earnings reports. Cao, Chen and Griffin (2005) find higher pre-announcement volume of call options portends increased higher takeover premiums for M&A targets, but they do not detect much information in option volumes at non-event times. Launois and Van Oppens (2003) find similar results and further detect that informed traders prefer to buy OTM options and sell ITM options prior to takeovers. Arnold et al. (2005) describe how, in the absence of an option market for an underlying stock, abnormal stock volume exists for target firms prior to cash tender offer announcements. However, when options markets are present for a firm, the volume effect of stock markets dissipates and the increased option volume emerges at an earlier point

6 than the would-be uptick in stock volume (13 days prior to the tender offer rather than 10 days in stock markets when no option markets exist). The lead-volume effect of option markets has been linked to other informational events as well, including the terrorist attacks of September 11 th, Poteshman (2006) documents an abnormally high level of put buying in the days preceding the attacks, consistent with informed trading. Put-call ratios greater than six in the case of American Airlines and over 25 in the case of United Airlines were present on September 6 th, These ratios are shown to be statistically significant and indicative of early trading in options markets. While the importance of both option and equity volume as lead indicators of equity prices have been discussed throughout the literature, the impact of open interest has been a more recent and less developed topic. Lakonishok et al. (2007) seek to describe the distribution of option open interest among investors by utilizing a unique data set in order to describe the open-interest characteristics of contracts by investor class. They obtain classifications for firm proprietary traders, public customers of full-service brokers, public customers of discount brokers, and other public customers, and analyze daily open-interest levels from 1990 through They determine full-service customers, who provide the majority of nonmarket maker transactions, have more written than purchased open interest. Firm proprietary traders, discount customers and other public customers have greater purchased than written open interest. Finally, nonmarket-maker investors have four times as much purchased call as purchased put open interest. Early efforts to specifically link option open interest to equity returns centered on the dynamic surrounding certain events. Results, however, have been quite varied. Schachter (1988) believes option open interest might provide insight on equity returns and finds a significant drop in abnormal option open interest prior to earnings announcements. This effect was particularly prevalent in options with short time to maturity and those most sensitive to volatility changes. Open interest was significantly above normal in certain days following announcements when further distinctions were drawn by moneyness. In studying a different corporate event, merger announcements, Jayaraman et al. (2001) find an increase in trading activity of both calls and puts for firms before public announcements or publicized rumors. This activity precedes abnormal trading activity in equity markets. Studying a sample of 33 announcements, the authors note increased abnormal open interest and volume preceding the announcements. Open-interest increases are concentrated in out-of-the-

7 money options with short time (less than 60 days) to maturity. This supports the leverage hypothesis of options trading emphasized by Black (1975) and others. Launois and Van Oppens (2003), in their study of takeover announcements, note the advantages of utilizing open interest as indicative or market activity, rather than volume. First, open interest is less volatile than volume, and second, open interest is not affected by very shortterm, intraday speculation. Chesney et al. (2009) focus on puts and stock market crashes. From the authors consider 14 companies and study the importance of increased daily open interest. The results are further verified with a sample of European firms. The authors next consider put open interest and find that when large, day-to-day increases occur, not indicative of hedging activity, that most significantly positive subsequent equity returns are due to the announcement effects of M&A announcements, earnings announcements, quarterly financial statements, and the September 11 th, 2001 terrorist attacks. Additional work has recently developed exploring the connection between option open interest and equity markets outside the periods surrounding major corporate events. Bhuyan and Chaudhury (2005) study 30 firms from February to July of They use open interest of nearmaturity equity options to form a number of portfolios based on call and put characteristics and in so doing outperform the S&P 500, passive covered call, and buy-and-hold strategies based on the underlying stocks. However, the period of measured performance is notably short, the impact of option open-interest characteristics on equity returns are not directly tested, and the theory linking the two markets is not developed. Bhuyan and Yan (2002) develop stock price predictors based on option open interest and volume characteristics. Srivastava (2004) verifies these findings in the context of the Indian stock and option markets and notes that option open interest is a superior predictor of equity prices than option volume while utilizing logarithmic regressions; however, the sample sizes of these studies are small and cover only brief periods while failing to control for typical factors believed to affect asset prices. We hypothesize that option traders will demand relatively more (fewer) call (put) options when they believe the underlying asset will perform well in the near future and conversely that option traders demand relatively more (fewer) put (call) options when they believe the underlying asset will perform poorly in the near future. We expect these demand changes will

8 lead to changes in aggregate open interest for call and put options that will have power to predict future equity returns. The leverage argument first described by Black (1975) suggests the long side of option contracts denotes the holdings of informed investors. In this paper, we demonstrate the informational content of option open-interest changes for near-term equity price movements. Those firms in the highest quintile of recent change in put open interest underperform those firms in the lowest quintile by a highly significant 31.4 basis points per month. The predictive impact of changes in call open interest is less demonstrable, though some evidence of superior performance for equities with increased call open interest exists. When the effects are combined, we find those firms in the highest Call/ Put open-interest quintile outperform those firms in the lowest quintile by 52 basis points per month. This underlying result remains after considering numerous controls and methodological approaches. The remainder of the paper develops as follows. Section 2 describes the data sources and empirical methodology. Section 3 presents results. Section 4 concludes. 2. Data and Methodology Our sample consists of all firms in the CRSP database with options data available from Optionmetrics. Our sample period is January, 1996 through September, Firms must have CRSP share codes of either 10 or 11 and be traded on the NYSE, NASDAQ or AMEX to be included in the sample. Data for calculating size and momentum are from CRSP. Option openinterest levels and implied volatilities are from Optionmetrics. Data used to calculate book value of equity is from Compustat. To form our open-interest measures we separately calculate the aggregate changes in open interest for call and put options for each firm over approximately the past month. We compare open-interest levels two trading days before each option expiration date to those on the trading day following the previous option expiration date. We also measure the change in the ratio of call open interest to put open interest over this period. For example, for the May of 2000 expiration date of Friday, May 19 th, we measure the changes in call open interest, put open interest, and the ratio of call-to-put open interest from the next trading day, Monday, May 22 nd, 2000 to two trading days preceding the following expiration, Wednesday, June 14 th, For

9 constructing open-interest measures, only options expiring from month t+1 through month t+11 are used. Firms with beginning aggregate open-interest levels of less than 50 call or put contracts are excluded from the sample. 1 All changes are measured in percentages. For each expiration date, we sort firms into quintiles based on call and put open-interest changes, call-to-put open-interest ratio changes, and the current call-to-put open-interest ratio. We then calculate the mean buy-and-hold equity return for all firms in each quintile over the next month and test for differences in returns across quintiles. We also track the buy-and-hold performance of high and low quintile portfolios through time, rebalancing after each option expiration date. Additionally, we evaluate the high-low portfolio returns for the open-interest characteristic quintiles separately for each quintile of market equity and the book-to-market ratio. We next utilize a four-factor calendar-time regression approach in order to evaluate quintile portfolio returns while controlling for common factors shown to have power to predict future equity returns. At the end of each open-interest change period, firms are placed into quintiles based on open-interest measures over the previous approximate month. These rankings remain in place until the day following the next expiration date, open-interest measure calculation. The high-low open-interest quintile returns are then regressed on the three daily Fama-French (1993) factors and the momentum factor taken from Ken French s website. 2 The intercepts of these regressions are interpreted as abnormal returns associated with the high-low portfolios. Similar analyses are also performed for each size and book-to-market quintile. We lastly implement the Fama-Macbeth (1973) regression procedure to further test the ability of open-interest change measures to predict future equity returns in a framework allowing for the easy inclusion of additional control measures. Fama-Macbeth regressions are performed with expiration-to-expiration returns as the dependent variable and open-interest measures, as well as size, book-to-market, momentum and implied volatility as dependent variables. The means and significance levels of dependent variable coefficients are then presented. 3. Results Our initial analysis describes the differences in the mean one-period-ahead stock returns across quintiles formed on changes and levels of open-interest variables. We separately consider 1 This eliminates approximately the lowest 1% of open interest firms from the sample. 2

10 the percentage changes in aggregate-call open interest and aggregate-put open interest from the day following an option expiration date to the second-to-last day preceding the following expiration date. These statistics are denoted Call and Put, respectively. The ratio of aggregate call open interest to aggregate put open interest is denoted C/P. The change in this ratio from the trading day following the latest option expiration date to the second-to-last day preceding the next option expiration is denoted C/P. 3.1 Initial Results We sort the sample into quintiles based on the open-interest variables and compare the mean equally-weighted one period ahead buy-and-hold returns across the quintiles. These results are presented in Table 1. [Insert Table 1] The highest Call quintile portfolio outperforms the lowest Call portfolio by a significant (at the 5% level) 24.2 basis points per month. We also track the growth of buy-and-hold equity portfolios which are long (short) firms in the highest (lowes)t Call quintiles. These portfolios are rebalanced two days prior to each expiration date. The results are seen in Figure 1 and demonstrate the generally superior performance of firms with higher call open-interest changes over the course of the previous month relative to those with lower call open-interest changes. [Insert Figure 1] The lowest Put quintile outperforms the highest Put quintile portfolio by a significant (at the 1% level) 31.4 basis points per month. Figure 2 shows that the buy-and-hold equity portfolio returns of the lowest quintile of Put firms strongly outperforms the buy-and-hold portfolio returns of the highest Put quintile, a relationship more pronounced than that seen for the Call portfolios.

11 [Insert Figure 2] The most striking initial result is the vastly superior performance of the highest C/P quintile portfolio relative to the lowest C/P quintile portfolio. The mean difference of 52 basis points per month is strongly significant with a t-statistic of In Figure 3 we track the buyand-hold returns of the highest and lowest quintile portfolios formed based on C/P. [Insert Figure 3] The portfolio of firms in the lowest C/P quintile doubles in value over the January 1996 to September 2008 period, but the portfolio of firms in the highest C/P quintile increases in value seven-fold. Initial evidence shows that movement towards more (fewer) existing call (put) contracts indicates superior performance for the underlying equity in the near future. We also test whether levels of aggregate call and put open interest have power to predict future equity returns. Separating the sample by the ratio C/P results in significantly higher mean returns for the high C/P quintile compared to the low C/P quintile by 31.0 basis points per month. The buy-and-hold performance of these two portfolios in presented in Figure 4. [Insert Figure 4] The degree to which the relationships discussed above can be explained by the risk of the respective portfolios or characteristics of firms comprising these portfolios is the focus of Sections 3.2 and Double Sorts Double-sorted portfolio returns are presented by first sorting firms into quintiles each period based on size or book-to-market ratio then further sorting within these quintiles based on Call, Put, C/P and C/P. Mean high-low open-interest variable quintile returns are then presented for each size and book-to-market quintile.

12 [Insert Table 2] Mean return differences for size quintiles are shown in Table 2. The superiority of the high Call portfolios as compared to the low Call portfolios is driven by the quintile of smallest firms in our sample. The mean high-low portfolio difference is a strongly significant 95 basis points per month for this quintile while differences are small and insignificant for the other quintiles. The higher returns of low Put open-interest firms relative to high Put firms is more persistent after controlling for firm size. The high-low Put open-interest returns are a significant basis points per month in the second size quintile and basis points in the middle size quintile. High-low Put open-interest returns are basis points in the low-size quintile and basis points in the high-size quintile, respectively, but these results are not statistically significant. Combining the effects of the call and put open-interest changes in the C/P measure demonstrates a strong open-interest effect that spans the smallest three size quintiles. Mean zero-cost portfolio returns of 132 basis points per month for size quintile 1, 62 basis points for size quintile 2, and 45 basis points for size quintile 3 are all statistically significant (at the 1% level for quintiles 1 and 2 and at the 5% level for quintile 3). The high-low C/P portfolio returns are insignificant in size quintiles 4 and 5, though at 21 basis points and 16 basis points, respectively, these differences are of the expected sign. Double-sort zero-cost portfolio returns that begin with stratification via book-to-market, before sorting by option open interest characteristics, do not exhibit an overall concentration in a certain book-to-market quintile. Results can be seen in Table 3. High-low Call returns are significantly positive in book-to-market quintiles 4 and 5, with mean monthly portfolio returns of 53.2 and 58.2 basis points, respectively, but significance of the Put high-low returns is seen only in book-to-market quintile 1, at basis points per month. [Insert Table 3] Classification by the C/P measure results in 5% significant zero-cost portfolio returns in four of the five book-to-market quintiles of the double sorts (59 basis points per month in book-to-

13 market quintile 1, 39 basis points in quintile 3, 59 basis points in quintile 4, and 57 basis points in quintile 5). As in the market equity case, the strongest option open-interest indications of future equity returns consider both call and put open interest. 3.3 Regression Results To consider the impact of the option open-interest measures in the presence of a multitude of controls, we first conduct four-factor calendar time regressions by placing firms into portfolios each period based on the Call, Put, C/P and C/P measures. For each measure, quintiles are determined on the trading day following an option expiration date, and these quintile rankings are maintained until the day preceding the next option expiration date. The daily high-low quintile portfolio returns for each of the measures are regressed on the three Fama-French (1992) factors and the momentum factor provided by Ken French with the alphas of the regressions serving as measures of abnormal performance. Results are seen in Table 4. [Insert Table 4] While the daily Call-based high-low portfolio exhibits an insignificant 1.0 basis point per day abnormal return, the Put-based high-low portfolio returns a highly significant -3.4 basis points per day, and the C/P-based high-low portfolio returns a highly significant 3.8 basis points per day. The impact of put open-interest changes appears to drive the effect of future equity returns, though there is evidence of an additional contributory effect from the Call factor. In the simultaneous presence of the traditional factors, recent option open-interest movements are significant predictors of future equity returns. To further consider the predictive ability of option open-interest measures, we first stratify the sample into the size and book-to-market quintiles earlier discussed before implementing the calendar-time procedure. This multi-step procedure allows for further analysis regarding the impact of the control measures. In Table 5, we present the intercepts of the various calendar-time regressions after first sorting firms into quintiles each period based on size. [Insert Table 5]

14 From Table 5, we note the tendency of the option open-interest impact to manifest itself most strongly amongst small firms. The strongest result is the 65 basis point return to the zero-cost C/P portfolio in the smallest firm quintile. All results for the three smallest size quintile zerocost portfolios are significant at the 5% level or better for both Put and C/P, again indicating the relative importance of the put open interest changes. In Table 6, calendar-time regression results are presented after first sorting firms into quintiles each period based on book-to-market ratio. [Insert Table 6] The separation of the sample into book-to-market quintiles prior to the implementation of the calendar-time procedure yields somewhat more significant option open-interest results amongst low book-to-market firms. Alphas for Put portfolios are significant for the three lowest bookto-market quintiles while C/P is again most powerful for predicting future equity returns. Alphas from each of the size quintiles are significant at the 10% level or better when this variable is used to form portfolios. The abnormal returns range from 2.0 basis points per day for book-to-market quintile 3 (significant at the 10% level) to 3.1 basis points per day for book-tomarket quintile 1 and 3.3 basis points per day for book-to-market quintile 5 (both significant at the 1% level). The Fama-Macbeth (1973) procedure is next utilized in order to include two additional control regressors, momentum and the current implied volatility level of firms. Results are presented in Table 7. [Insert Table 7] We confirm the importance of the change in put open interest. The coefficient on the Put regressor is negative and significant at the 1% level while the Call coefficient is not significant. The C/P predictor is also significant at the 1% level with the expected positive sign. The added

15 implied volatility control measure, IV, is not significant while the one-year momentum control is. Overall, the evidence suggests an important, significant predictability of future equity returns when the change in put open interest is considered, even in the presence of control measures. With greater recent increases in put open interest, the one-month future equity returns worsen. The contribution of the call open-interest change is positive and significant in some instances, but this effect is not nearly as consistent as that seen from put open-interest changes. The combination of the call and put open interest changes, however, as seen in the C/P measure, serves as the strongest and most consistent indicator of future equity returns. 4. Conclusions In this paper, we further expand the wealth of literature linking option market information and future underlying equity returns. Specifically, we examine the relationship between option open-interest changes and future returns. We show that option traders demand relatively more (fewer) call (put) options when they believe the underlying asset will perform well in the near future. Conversely, option traders demand relatively more (fewer) put (call) options when they believe the underlying asset will perform poorly in the near future. These demand changes lead to changes in aggregate open interest that have power to predict future equity returns. The leverage available to informed investors in options markets results in the link of increased long call (put) positions to equity upswings (downswings). In our empirical investigation we demonstrate a strong link between recent changes in aggregate put open-interest levels and the future underlying equity price movement. We sort firms into quintiles based on our open-interest variables and find strong relationships between these measures and future equity returns. Firms with increases in recent put open interest greatly underperform those firms that have seen declines in put open interest. Call open-interest changes portend the opposite effect, but the relationship is much less pronounced. The most effective open-interest predictor of future equity returns is the ratio of the recent changes in call open interest to put open interest. Large increases in the ratio are followed by relatively strong future stock returns. We further demonstrate that our findings are robust by performing additional tests. We show that in general, these relationships persist after controlling for firm size and book-to-

16 market. Also, by employing four-factor calendar time and Fama-Macbeth regression analysis, we show our results hold when controlling for factors shown to have power to predict future equity returns. After controlling for typical asset pricing factors, a strong option open interest change effect remains. We demonstrate the documented preference of investors with superior information, as first discussed by Black (1975), to hold options rather than equities due to their superior leverage. The ability to greatly profit from long option positions with relatively small initial outlays encourages informed investors to purchase puts (calls) of firms they suspects will exhibit negative (positive) performance in the near future. While market efficiency suggests that such information would be immediately and completely reflected into both option and equity markets, we present further evidence that real-world informational differences between the two markets result in different speeds for the incorporation of information.

17 References Amin, K and C Lee, 1997, Option trading, price discovery, and earnings news dissemination, Contemporary Accounting Research, 14, Arnold, T, Erwin, G, Nail, L, and T Nixon, 2006, Do option markets substitute for stock markets? Evidence from trading on anticipated tender offer announcements, International Review of Financial Analysis 15, Back, K, 1993, Asymmetric information and options, Review of Financial Studies 6, Bali, T, 2008, The intertemporal relation between expected returns and risk, Journal of Financial Economics, 87, Bhuyan, R and M Chaudhury, 2005, Trading on the information content of open interest: evidence from the US equity options market, Journal of Derivatives and Hedge Funds 11, Bhuyan, R and Y Yan, 2002, Informational role of open interests and volumes: Evidence from option markets, Paper presented at twelfth annual Asia-Pacific futures research symposium held in Bangkok. Black, F, 1975, Fact and fantasy in the use of options, Financial Analysts Journal 31, Blasco, N, Corredor, P, and R Santamaria, 2009, Does informed trading occur in the options market? Some revealing clues, Working Paper. Carhart, M, 1997, On persistence in mutual fund performance, Journal of Finance 52, Cao, C, Chen, Z, and J Griffin, 2005, Informational content of option volume prior to takeovers, Journal of Business 78, Cao, H and H Yang, 2009, Differences of opinion of public information and speculative trading in stocks and options, Review of Financial Studies 22, Chan, K, Chung, P, and W Fong, 2002, The informational role of stock and option volume, Review of Financial Studies 15, Chesney, M, Crameri, R, and L Mancini, 2009, Detecting informed trading activities in options markets, Working Paper. Cox, J and M Rubenstein, Options Markets, Englewood Cliffs, NJ: Prentice-Hall, Cremers, M and D Weinbaum, 2008, Deviations from put-call parity and stock return predictability, Journal of Financial and Quantitative Analysis, Forthcoming. Diamond, D and R Verrecchia, 1987, Constraints on short-selling and asset price adjustment to private information, Journal of Financial Economics 18, Doran, J and K Krieger, 2009, Information and implications for equity returns in the implied volatility skew, Financial Analysts Journal, Forthcoming.

18 Easley, D and M O Hara, 1987, Price, trade size and information in securities markets, Journal of Financial Economics 19, Fama, E and K French, 1992, The cross-section of expected stock returns, Journal of Finance 47, Fama, E and J Macbeth, 1973, Risk, return and equilibrium: Empirical tests, Journal of Political Economy 81, Granger, C, 1969, Investigating causal relations by econometric models and cross-spectral models, Econometrica 37, Granger, C and P Newbold, 1977, Forecasting economic time series, (Academic Press, New York). Hasbrouck, J, 1995, One security, many markets: Determining the location of price discovery, Journal of Finance 50, Jayaraman, N, Frye, M, and S Sabherwal, 2001, Informed trading around merger announcements: an empirical test using transaction volume and open interest in options market, The Financial Review 37, Lakonishok, J, Lee, I, Pearson, N, and A Poteshman, 2007, Option market activity, Review of Financial Studies 20, Launois, T, and H Oppens, 2003, Informed trading around corporate event announcements: Stocks vs. options, Working Paper. Manaster, S and R Rendleman, 1982, Option prices as predictors of equilibrium stock prices, Journal of Finance 37, Pan, J, and A Poteshman, 2006, The information in option volume for stock prices, Review of Financial Studies 19, Poteshman, A, 2006, Unusual option market activity and the terrorist attacks of September 11, 2001, Journal of Business 79, Schachter, B, 1988, Open interest in stock options around quarterly earnings announcements, Journal of Accounting Research 26, Srinivas, P, 1993, Trade size and the information content of option trades, Working Paper. Srivastava, S, 2004, Informational content of trading volume and open interest, an empirical study of stock option markets in India, Indian Journal of Finance and Research 14, Stephan, J and R Whaley, 1990, Intraday price change and trading volume relations in the stock and stock option markets, Journal of Finance 45, Vijh, A, 1990, Liquidity of the CBOE equity options, Journal of Finance 45,

19 Table 1: Firm Stock Returns by Open Interest Quintiles This table presents mean firm stock returns after dividing the sample into quintiles each period based on open interest variables. Call and Put are calculated by measuring open interest the first trading day after an option expiration date and again two trading days prior to the next option expiration date. C/P Ratio is the change in the ratio of call open interest to put open interest with both characteristics measured as described above. Changes are calculated on a percentage basis. C/P Ratio is the ratio of call open interest to put open interest, measured two trading days prior to the option expiration dates. Returns are buy-and-hold returns from the first trading day after expiration through the last trading day prior to the next option expiration. Open interest variables from period t-1 and used to divide firm prior to calculating mean period t returns for each quintile. The sample period is from January 1996 through September Options with between 2 and 11 months to expiration on the initial measurement day are included. Mean quintile returns are presented in percentages with standard deviations in parentheses. Differences between high and low quintile mean returns are also presented with t-statistics in parentheses. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. Quintile Low High High-Low Open Interest Measure Call Put C/P Ratio C/P Ratio (0.166) (0.177) (0.158) (0.152) (0.156) (0.155) (0.156) (0.157) (0.158) (0.155) (0.163) (0.158) (0.163) (0.159) (0.165) (0.168) (0.172) (0.167) (0.173) (0.178) ** *** *** *** (2.26) (2.88) (4.96) (2.95)

20 Table 2: Firm Stock Returns by Market Equity and Open Interest Quintiles This table presents mean firm stock returns after first dividing the sample into quintiles each period based on market equity (ME) and then into quintiles based on open interest variables. Call and Put are calculated by measuring open interest the first trading day after an option expiration date and again two trading days prior to the next option expiration date. C/P Ratio is the change in the ratio of call open interest to put open interest and is measured in the same manner. Changes are calculated on a percentage basis. C/P Ratio is the ratio of call open interest to put open interest, measured two trading days prior to the option expiration dates. ME is measured at the end of month t-1 where the return period begins in month t. Returns are buy-and-hold returns from the first trading day after expiration through the last trading day prior to the next option expiration. The sample period is from January 1996 through September Options with between 2 and 11 months to expiration on the initial measurement day are included. Differences between high and low open interest quintile mean returns within each ME quintile are presented with t-statistics in parentheses. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. ME Quintile Low High Open Interest Measure Call Put C/P Ratio C/P Ratio *** *** (2.96) (1.33) (4.18) (0.64) ** *** * (0.13) (2.22) (2.48) (1.65) ** ** *** (0.00) (2.25) (2.00) (3.90) (0.75) (0.50) (1.20) (0.75) (0.14) (1.63) (1.06) (0.37)

21 Table 3: Firm Stock Returns by Book-to-Market Equity and Open Interest Quintiles This table presents mean firm stock returns after first dividing the sample into quintiles each period based on bookto-market equity (BM) and then into quintiles based on open interest variables. Call and Put are calculated by measuring open interest the first trading day after an option expiration date and again two trading days prior to the next option expiration date. C/P Ratio is the change in the ratio of call open interest to put open interest and is measured in the same manner. Changes are calculated on a percentage basis. C/P Ratio is the ratio of call open interest to put open interest, measured two trading days prior to the option expiration dates. BM is calculated using month t-1 market equity and the last reported book equity (month t-1 or prior). Returns are buy-and-hold returns from the first trading day after expiration through the last trading day prior to the next option expiration. The sample period is from January 1996 through September Options with between 2 and 11 months to expiration on the initial measurement day are included. Differences between high and low open interest quintile mean returns within each BM quintile are presented with t-statistics in parentheses. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. BM Quintile Low High Open Interest Measure Call Put C/P Ratio C/P Ratio *** ** (1.33) (2.82) (2.17) (0.73) (0.44) (1.32) (1.62) (0.21) * * (1.41) (0.66) (1.78) (1.65) *** *** *** (2.50) (0.75) (2.73) (2.67) *** *** (2.29) (0.46) (2.32) (1.27)

22 Table 4: Calendar-Time Regressions This table presents results of regressing daily high-low open interest quintiles returns on the three factors of Fama and French (1993) and the momentum factor provided on Ken French s website. Call and Put are calculated by measuring open interest the first trading day after an option expiration date and again two trading days prior to the next option expiration date. C/P Ratio is the change in the ratio of call open interest to put open interest and is measured in the same manner. Changes are calculated on a percentage basis. C/P Ratio is the ratio of call open interest to put open interest, measured two trading days prior to the option expiration dates. Options with between 2 and 11 months to expiration on the initial measurement day are included in measurement of open interest variables. Firms are placed into portfolios based on these measures and remain until the second trading day after the next measurement. Each day mean portfolio returns are calculated for high and low quintiles. The dependent variable in regressions is the difference between the mean daily high and low quintile returns, measured on a percentage basis. The sample period is from January 1996 through September Coefficients are presented with standard errors in parentheses. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. Open Interest Measure Call Put C/P Ratio C/P Ratio α *** *** * (0.01) (0.01) (0.01) (0.01) MKT *** *** * (0.01) (0.01) (0.01) (0.01) SMB *** *** *** (0.02) (0.02) (0.02) (0.02) HML *** *** *** (0.02) (0.03) (0.02) (0.02) MOM *** *** *** *** (0.01) (0.01) (0.01) (0.01)

23 Table 5: Calendar-Time Regressions by Market Equity Quintile This table presents alphas from regressing daily high-low open interest quintile returns on the three factors of Fama and French (1993) and the momentum factor provided by Ken French s website after first dividing the sample into quintiles each period based on market equity (ME). ME is measured at the end of month t-1 where the return period begins in month t. Call and Put are calculated by measuring open interest the first trading day after an option expiration date and again two trading days prior to the next option expiration date. C/P Ratio is the change in the ratio of call open interest to put open interest and is measured in the same manner. Changes are calculated on a percentage basis. C/P Ratio is the ratio of call open interest to put open interest, measured two trading days prior to the option expiration dates. Options with between 2 and 11 months to expiration on the initial measurement day are included in measurement of open interest variables. Within ME quintiles, firms are placed into portfolios based on these measures and remain until the second trading day after the next measurement. Each day, mean portfolio returns are calculated for high and low open interest quintiles. The dependent variable in regressions is the difference between the mean daily high and low quintile returns. The sample period is from January 1996 through September Alphas are presented with standard errors in parentheses. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. ME Quintile Low High Open Interest Measure Call Put C/P Ratio C/P Ratio *** *** (0.018) (0.018) (0.016) (0.017) ** *** (0.013) (0.014) (0.012) (0.012) *** *** *** (0.012) (0.012) (0.010) (0.010) (0.010) (0.011) (0.009) (0.009) *** (0.010) (0.010) (0.008) (0.010)

24 Table 6: Calendar-Time Regressions by Book-to-Market Equity Quintile This table presents alphas from regressing daily high-low open interest quintile returns on the three factors of Fama and French (1993) and the momentum factor provided by Ken French s website after first dividing the sample into quintiles each period based on book-to-market equity (BM). BM is calculated using month t-1 market equity and the last reported book equity (month t-1 or prior). Call and Put are calculated by measuring open interest the first trading day after an option expiration date and again two trading days prior to the next option expiration date. C/P Ratio is the change in the ratio of call open interest to put open interest and is measured in the same manner. Changes are then calculated on a percentage basis. C/P Ratio is the ratio of call open interest to put open interest, measured two trading days prior to the option expiration dates. Options with between 2 and 11 months to expiration on the initial measurement day are included in measurement of open interest variables. Within BM quintiles, firms are placed into portfolios based on these measures and remain until the second trading day after the next measurement. Each day, mean portfolio returns are calculated for high and low open interest quintiles. The dependent variable in regressions is the difference between the mean daily high and low quintile returns. The sample period is from January 1996 through September Alphas are presented with standard errors in parentheses. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. BM Quintile Low High Open Interest Measure Call Put C/P Ratio C/P Ratio ** *** *** (0.015) (0.015) (0.013) (0.014) ** ** (0.012) (0.012) (0.011) (0.011) ** * (0.011) (0.011) (0.011) (0.011) * *** ** (0.011) (0.011) (0.010) (0.011) *** ** (0.015) (0.015) (0.013) (0.015)

25 Table 7: Fama-Macbeth Regression Results This table presents Fama-Macbeth (1973) regression results. Regressions are performed each period where the dependent variable is firm buy-and-hold returns over the period. Market equity (ME) is measured at the end of month t-1 where the return period begins in month t. Book-to-market equity (BM) is calculated using month t-1 market equity and the last reported book equity (month t-1 or prior). Momentum (MOM) is the firm buy-and-hold return from month t-12 through month t-1. IV is moneyness-weighted implied volatility measured two trading days prior to the option expiration dates before the return measurement period begins. Call and Put are calculated by measuring open interest the first trading day after an option expiration date and again two trading days prior to the next option expiration date. Changes are calculated on a percentage basis. C/P Ratio is the change in the ratio of call open interest to put open interest and is measured in the same manner. C/P Ratio is the ratio of call open interest to put open interest, measured two trading days prior to the option expiration dates. Options with between 2 and 11 months to expiration on the initial measurement day are included. The sample period is from January 1996 through September ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. α Call OI Put OI *** C/P Ratio *** C/P Ratio ME BM *** *** *** *** MOM *** *** *** *** IV