Journal of Banking & Finance

Transcription

1 Journal of Banking & Finance 35 (2011) Contents lists available at ScienceDirect Journal of Banking & Finance journal homepage: Stock and option market divergence in the presence of noisy information Carl R. Chen a, J. David Diltz b, Ying Huang c, Peter P. Lung b, a Department of Economics and Finance, University of Dayton, Dayton, OH 45469, USA b Department of Finance and Real Estate, University of Texas at Arlington, TX 76019, USA c Academy of Financial Research and College of Economics, Zhejiang University, China article info abstract Article history: Received 18 May 2010 Accepted 14 January 2011 Available online 22 January 2011 JEL classification: G1 G13 G14 Keywords: Information and market efficiency Price pressure Options Informed trading We examine market behavior of the stock and option markets upon the arrival of noisy information in the form of CNBC s Mad Money recommendations. If stock and option markets are not equally efficient, they should respond differently to noisy information, with the less efficient market more susceptible to noise. We find that the stock market is less efficient than the option market. The abnormal difference between option-implied and actual stock returns is negative and significant upon exposure to noisy information. This difference may yield an economically significant monthly trading profit of up to 5%. We conclude that the stock market is more susceptible to noisy information than the option market and is therefore less efficient. Published by Elsevier B.V. 1. Introduction Are investors in different trading venues equally rational when it comes to reacting to noisy information? We attempt to answer this question in the case of stock and option markets by studying the differential response of stock and option-implied stock prices to the arrival of noisy information. It has been argued that the stock market should react to information quicker and more accurately than the option market because option prices rely on their underlying stock prices. Proponents of this hypothesis have investigated whether the stock market leads in information discovery using Granger causality and similar techniques. Many researchers consistently find no significant lead in the option market. For example, Stephan and Whaley (1990) find that option-implied stock prices cannot predict future stock returns. Chan, Chung, and Johnson (1993) analyze the lead lag relation between stock and option markets, and find no evidence that option price changes lead stock price changes. Similar findings have been reported by Diltz and Kim (1996), Finucane (1999), O Connor (1999), Chan, Chung, and Fong (2002), and McIntyre and Jackson (2009). Previous literature on implied volatilities in the option market also Corresponding author. Tel.: ; fax: addresses: [email protected] (C.R. Chen), [email protected] (J.D. Diltz), [email protected] (Y. Huang), [email protected] (P.P. Lung). documents short-horizon underreaction and long-horizon overreaction to information arrival (i.e., Stein (1989), Poteshman (2001)). 1 Opponents of the stock-leads-option hypothesis suggest that option market may be the preferred habitat for informed trading due to the absence of short-sale constraints, built-in downside protection, and opportunities to exploit leverage. 2 If informed traders prefer the option market, the option prices should be less responsive to noisy information. Manaster and Rendleman (1982) are among the first to use option prices to predict prices in the underlying stock market. They suggest that option-implied stock prices represent the option market s assessment of the underlying asset s value, and that the option-implied stock prices contain information that is not fully reflected in actual stock prices. Kumar et al. (1992) document abnormal option returns prior to block trading in the underlying stock. Easley, O Hara, and Srinivas (1998) use signed option trading volume to show that the option markets contain information about stock 1 Other studies focus on the price discovery in the futures markets. For example, Chen and Gau (2010) find that spot foreign exchange rates provide more price discovery than the CME futures trades. 2 See, for example, Black (1975), Cox and Rubinstein (1985), Easley, O Hara, and Srinivas (1998), and Chakravarty, Gulen, and Mayhew (2004). After the SEC removed short-sale constraints in July 2007, the informational role of the option market could have changed. This is beyond the scope of this study /$ - see front matter Published by Elsevier B.V. doi: /j.jbankfin

2 2002 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) price changes. Research using sequential-trade models also suggests that informed traders trade in the option market. 3 More recently, Chakravarty et al. (2004) show that about 17% of price discovery occurs in the option market. Of particular interest to this paper, Lamont and Thaler (2003) find that in the stock carve-out case of Palm/3Com, the actual price of Palm shares is considerably higher than the option-implied stock price for Palm in the option market. Ofek et al. (2004) also show that option-implied stock prices forecast future stock returns. Taylor, Yadav, and Zhang (2010) show that option forecasts are nearly always more informative for firms that have more actively traded options. Yu, Lui, and Wang (2010) find that option-implied volatility is superior to either historical volatility or GARCH type volatility forecasts. Most recently, Xing, Zhang, and Zhao (2010) find that the volatility smirk has a capacity for predicting future equity returns. Cremers and Weinbaum (2010) document that the deviations from put-call parity contain information about future stock price movements. They also find that option prices are more likely to deviate from put-call parity when underlying stocks face more information risk. All of these recent findings are consistent with the notion that equity prices may reflect one set of beliefs while option prices reflect another. However, existing literature has not directly examined the comparative efficiency of trading behavior in the stock and option markets when noisy information is present. We aim to understand which market behaves more efficiently when stock and option markets diverge. Efficient trading behavior reveals market quality, and thus where informed traders are more likely to operate. Divergence between stock and option markets should signal the difference in beliefs between informed versus noise traders. We test market quality (i.e., efficiency) based on the price pressure hypothesis first proposed by Scholes (1972). The price pressure hypothesis asserts that security prices may diverge temporarily from efficient information values. Uninformed shifts in excess demand compensate liquidity providers as prices return to equilibrium values. Past research has documented abnormal returns and trading volumes around the arrival of noisy information. Driven by noise trading from naïve investors, abnormal returns are reversed shortly thereafter. 4 We gauge market quality by the strength of price pressure effects in response to noisy information in the form of recommendations made by CNBC s Mad Money flamboyant host Jim Cramer. 5 We provide evidence that his buy recommendations are a good example of noisy information, and stock and option prices diverge as a result. However, his sell recommendations are more credible and no divergence between the stock and option market is observed. Although Mad Money is popular among individual investors, the fact that it disseminates noisy information is known by investment professionals and some academics. Thus, Mad Money stock picks constitute a reasonable test for market rationality. If the option market responds to Mr. Cramer s recommendations similar to the stock market, we conclude that the option market possesses no quality advantage. If informed traders prefer the option market 3 In sequential-trade models, informed traders can trade in either the stock or the option market. These models suggest that the amount of informed trading in option markets should be related to the depth or liquidity of both the stock and option markets, and the amount of leverage achievable with options. See, for example, Biais and Hillion (1994), Easley et al. (1998), Mayhew et al. (1995), and Pan and Poteshman (2006). 4 See, for example, Harris and Gurel (1986), Lynch and Mendenhall (1997), Wurgler and Zhuravskaya (2002), Mitchell, Pulvino, and Stafford (2002), Coval and Stafford (2004), Corwin (2003), Liang (1999), Carhart, Kaniel, Musto, and Reed (2002), Cohen, Gompers, and Vuolteenaho (2002), Hotchkiss and Strickland (2003), and Engelberg, Sassville, and Williams (2009). 5 The show s animated host, Jim Cramer, draws more than 398,000 viewers daily according to the Philadelphia Enquirer, January 8, Recent estimates provided by Nielsen range from 400,000 to 600,000. The show airs three times a day during weekdays at 6:00 pm, 9:00 pm, and 12:00 midnight. and uninformed investors prefer stock trading, the stock price implied by the corresponding option (hereafter option-implied stock price) should exhibit significantly less price pressure than the observed stock price. In extreme cases, the option-implied stock price may even move in the opposite direction in response to noisy information. Using the standard event time methodology, 6 we examine trading behavior in the stock and option markets in response to Mr. Cramer s recommendations. Our final sample consists of 317 Cramer buy recommendations and 337 sell recommendations from July 2005 through April 2007 following deletion of inconsistent and news confounded observations. 7 We document three major findings. First, for buy recommendations, in the presence of a price pressure effect, the option market behaves more efficiently than the stock market. A 2.92% abnormal stock return on the day after Mr. Cramer s buy recommendations suggests the existence of a price pressure effect in the stock market for small-cap stocks. This is consistent with other studies examining Mad Money effects on the stock market. These results are reversed during the ensuing two weeks. In contrast, the option-implied stock returns are significantly lower than the actual stock returns following the buy recommendations. This indicates that either the option market is less responsive to the noisy information than the stock market, or the option traders may actually trade against the naïve stock investors upon the arrival of Cramer s buy recommendations. 8 On the other hand, Mr. Cramer s sell recommendations are more credible. The stock prices of sell recommendations drop for the following trading days without any price reversal. The option-implied stock returns move along with the actual stock returns without any significant deviation after the sell recommendations. Thus, Mr. Cramer s sell recommendations appear to be more trustworthy than his buy recommendations. Second, we find that following the buy recommendations, bidask spreads in the option market decrease significantly and option trading volumes are abnormally high. The abnormal trading activity lasts for more than 5 days following the recommendations. This finding strengthens the abnormal price results, and it suggests that option market makers may narrow the bid-ask spreads in anticipation of reduced adverse selection bias resulting from the buy recommendations. Furthermore, abnormal trading activity is particularly strong for put options. This result is consistent with Chakravarty et al. (2004) and Xing et al. (2010) in that informed traders prefer to trade put options when the stock prices are expected to drop. For the sell recommendations, the bid-ask spreads in the option market also decrease on day +1. This is consistent with the notion that the sell recommendations disseminate information that reduces adverse selection bias. The option trading volume following sell recommendations increases, particularly for put options. Third, we show that the abnormal divergence between the actual stock prices and option-implied stock prices due to buy recommendations could yield a maximum risk-adjusted return of 5.06% during the 30-day event period. This result is from the option-implied stock prices associated with out-of-the-money put options. It suggests that out-of-the-money put options are the best predictors of the price reversal pattern following the price pressure effect on day +1. Again, this finding supports the results in Xing et al. (2010). Out-of-the-money put options tend to convey 6 We follow the event study design documented in Mikkelson and Partch (1988). 7 Indeed, the Mad Money show makes more buy recommendations than sell recommendations. However, more stocks are repeatedly recommended to buy and fewer stocks are repeatedly recommended to sell. Therefore, we have more first-time sell recommendations than first-time buy recommendations. 8 We also conduct the test based on the full sample (without deleting news confounded observations), and the results are qualitatively similar to those of the more restricted sample.

3 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) more information when stock prices are expected to drop. Nevertheless, we need to interpret this finding with caution. This result sustains only if investors can freely short-sell stocks. In a market with short-sale constraints or high short-sale costs, the return may diminish or disappear altogether. The paper is organized as follows. Section 2 presents the event study methodology and hypotheses to be tested. Data sources and sample characteristics are shown in Section 3, and empirical results are discussed in Section 4. Section 5 reports the results of robustness tests. Trading profit potential is reported in Section 6, and Section 7 concludes the paper. 2. Methodology and hypothesis We conduct event studies on actual stock prices and optionimplied stock prices to examine the quality of trading in the stock and option markets in response to the recommendations made by CNBC s Mad Money host Jim Cramer. Although the show s noisy background, dramatic camera effect, and the boisterous host may seem silly to some, evidence shows that Mr. Cramer s buy recommendations actually affect stock prices in the short run. We explore the possibility of a quality gap between stock and option markets according to the relation between the innovations in the observed stock prices in the stock market and option-implied stock prices in the option market. The existing literature and this study both find a price pressure effect in the stock market following Cramer s stock buy picks. As a result, we assert that if noise traders are equally likely to trade in the stock or option market, both option-implied and observed stock prices should rise, being consistent with Mr. Cramer s buy recommendations. If noise traders dominate the stock market, then price pressure would cause stock prices to be more responsive than option-implied stock prices. On the other hand, if naïve traders are more active in the option market, price pressure would cause option-implied stock prices to be more responsive than actual ones. Thus, the divergence between option-implied and observed stock prices reveals a difference in quality between the two markets, with the more efficient market exhibiting a weaker reaction to the buy recommendations. Prior studies find that analysts sell recommendations are more credible, with large traders more likely to respond. 9 Based upon this strand of earlier works, we assert that an initial negative reaction to Cramer s sell recommendations should occur, and no price reversal should be observed in the stock market. If option market traders are informed, one should observe no abnormal behavior in option-implied prices following the recommendations. Section 2.1 discusses the methodology for estimating option-implied stock prices from the option market. Section 2.2 describes the event study procedure. Section 2.3 develops the hypotheses The put-call parity approach In this section, we outline the procedures for estimating optionimplied stock prices from the option market using the put-call parity. Assuming an absence of arbitrage, it is well known that for European options, the put-call parity follows: the dividends during the life of the option. For American options, we need to consider the early exercise premium. Following the analytical valuation formulas appearing in previous literature (e.g., Ofek, Richardson, and Whitelaw (2004), and Battalio and Schultz (2006)), we can rewrite Eq. (1) as: ^S0 ¼ C P þ Ke rt þ D EEP call þ EEP put ; where ^S 0 is the option-implied stock price based on the information in the option market. EEP call and EEP put are the early exercise premium for American call and put options, respectively. 10 We estimate the early exercise premium based on the method specified in Barone-Adesi and Whaley (1986). 11 The early exercise premium is the difference between the American and European option prices. All the put-call parity conditions are adjusted for this estimate as in Eq. (2). It should be noted that if dividends are zero during the life of the options, both D and EEP call disappear Event study procedure We use a standard event study procedure (see, for example, Mikkelson and Partch (1988) and Liang (1999)) to test the abnormal stock returns around the event date. We now discuss the procedures to obtain abnormal returns in both stock and option markets Stock market returns To analyze stock price behavior, we first specify a benchmark return and define the daily abnormal return in the event window as the difference between the actual return and the benchmark return. We use the following three-factor model to generate benchmark returns: R i;t ¼ a i þ b m R m;t þ b SMB SMB t þ b HML HML t þ e i;t ; where R i,t is the log return for common stock i on day t, R m is the log return for the CRSP value-weighted market index, SMB is the difference between the daily returns on portfolios of small and large stocks, and HML is the difference between the daily returns on stock portfolios of high and low book-to-market values. e i,t is the random error term of stock i on day t. We define the announcement day as day zero, and we estimate the model parameters based on a 150- day window from day 154 through day 5. Event day zero is defined as the day on which the announcement is made, with the announcement broadcast after trading hours. 12 Thus, we calculate the market response from day +1. The abnormal returns for common stock i during our event window from day +1 to day +30 are estimated from: AR i;t ¼ R i;t ð^a i þ ^b m R mt þ ^b SMB SMB t þ ^b HML HML t Þ¼e 0 i;t ; t ¼ 1; 2;...; 30: As in Boehmer, Poulsen, and Musumeci (1991), we employ the following test statistic for the abnormal stock returns: P N i¼1 test statistic t ¼ q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi SAR i;t=n 1=NðN 1Þ P N i¼1 ½SAR i;t 1=N P ; ð5þ N i¼1 SAR i;tš 2 ð2þ ð3þ ð4þ S 0 ¼ C P þ Ke rt þ D; ð1þ where S 0 is the stock price, C is the call option price, P is the put option price, K is the strike price, r is the continuously compounded risk-free rate, T is time to maturity, and D is the present value of 9 See, for example, Womack (1996), Michaely and Womack (1999), Carleton, Chen, and Steiner (2002), Green (2006), Clarke, Ferris, Jayaraman, and Lee (2006), Mikhail, Walther, and Willis (2007), Cliff (2007), and Mokoaleli-Mokoteli, Taffler, and Agarwal (2009). 10 Although this approach is well accepted in the literature, it does not consider the market frictions (such as bid-ask spreads), and the estimates for the early exercise premium may not be perfect. In the robustness test section in this paper, we employ another proxy for the implied option price based on option boundary conditions only. We find consistent results in both approaches. 11 We also use the method in Ho, Stapleton, and Subrahmanyam (1994). The test results are similar. 12 The Mad Money show is scheduled at 6:00 pm EST.

4 2004 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) where SAR i is the standardized abnormal returns of stock i, defined as: SAR i;t ¼ AR i;t r i ; ð6þ where r i is the standard deviation of the estimation-period abnormal returns for stock i. Boehmer et al. (1991) show that this test statistic is not affected by event-induced variance changes. The cumulative abnormal returns for the stock i from day t 1 to day t 2 are: CAR i ¼ Xt 2 t¼t 1 AR i;t : The test statistic for the cumulative abnormal stock returns takes the following form: P N i¼1 test statistic ¼ q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi SCAR i;t=n 1=NðN 1Þ P N i¼1 ½SCAR i 1=N P ; ð8þ N i¼1 SCAR iš 2 where SCAR i is the standardized cumulative abnormal return of the ith stock, estimated as: CAR i SCAR i ¼ q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : ð9þ ðt 2 t 1 þ 1Þr 2 i Option-implied stock returns The abnormal behavior of option-implied stock returns is analyzed similarly to that in the stock market. Because the value of an option depends on its underlying asset, we first specify a benchmark relation between the option and stock market returns. We then estimate the abnormal returns in the event window. The benchmark relation between option and stock for stock i is specified as: IS it ¼ a i þ b i R it þ s it ; ð7þ ð10þ where IS it is the option-implied log stock return based on optionimplied price for common stock i on day t calculated based on Eq. (2). That is, IS i;t ¼ logð^s i;t =^S i;t 1 Þ: R it is the actual stock return for common stock i on day t. Similar to the construction of the benchmark in the stock market, the benchmark for the option market is estimated over a 150-day period from day 154 to day 5. The abnormal option-implied stock returns for common stock i from day +1 to day +30 are estimated from: AR IS i;t ¼ IS it ð^a i þ ^b i R it Þ¼s 0 it ; t ¼ 1; 2;...; 30: ð11þ The cumulative abnormal option-implied stock returns for stock i from day t 1 to day t 2 are: CAR IS i ¼ Xt 2 t¼t 1 AR IS i;t : ð12þ The test statistics for AR_IS i,t and CAR_IS i,t are computed in the same manner as AR i,t and CAR i,t in Eqs. (5) and (8), respectively Abnormal bid-ask spreads and turnover Similar to Liang (1999), we compute the abnormal trading turnover on day t and its standard deviation as follows. The stock turnover ratio is specified as: TO it ¼ VOL it ; SHROUT it ð13þ where VOL it and SHROUT it are the daily trading volume and shares outstanding for stock i on day t, respectively. The average daily turnover for stock i is calculated using the daily turnover in days 154 to 5: TO i ¼ X 5 t¼ 154 TO i;t =150: ð14þ The daily turnover for day t is the simple average turnover for all stocks in the sample: TO t ¼ 1 N X N i¼1 TO it TO i ; ð15þ where N is the number of stocks for day t. The abnormal turnover, AT t, is then computed as: AT t ¼ TO t 1: The variance of the abnormal turnover is: r 2 AT ¼ X 5 t¼ 154 ðat t ATÞ 2 ; where AT ¼ 1=150 P 5 t¼ 154 AT t. The test statistic for the abnormal turnover is: ð16þ ð17þ test statistic ¼ AT t q ffiffiffiffiffiffiffi : ð18þ r 2 AT The abnormal stock bid-ask spread, AS t, on day t and its standard deviation are estimated as follows. We define the spread as: Spread it ¼ Ask it Bid it : ð19þ Ask it þbid it 2 Then, we specify the daily stock spread as: Spread t ¼ XN i¼1 Spread i;t =N; ð20þ where N is number of stocks on day t. The average stock spread during the estimation window is computed as: Spread Est: ¼ X 5 t¼ 154 Spread t =150: The abnormal spread is estimated as: AS t ¼ Spread t Spread Est: : The test statistic for the stock abnormal spread is: ð21þ ð22þ t stat ¼ AS t q ffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; ð23þ r 2 AS Est: where r 2 AS Est: is the variance of the abnormal spread during the estimation window. Using the same procedure from Eqs. (19) (23), we compute and test the abnormal option trading activity and abnormal option spread. The option trading activity, estimated according to the option trading volume ratio, VR, is specified as: VR i;t ¼ OptionVolume i;t StockVolume i;t ; ð24þ where OptionVolume i,t and StockVolume i,t are the option trading volume and stock trading volume for stock i on day t, respectively. The option spread is calculated in the same manner as the stock spread in Eq. (19) Hypotheses We examine, separately, Cramer s buy and sell recommendations to test the relative market efficiency for the stock and options markets. This separation is required because the literature has

5 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) documented that analysts buy recommendations are generally less credible than their sell recommendations. (e.g., Carleton, Chen, and Steiner, 2002; Mikhail, Walther, and Willis, 2007; Cliff, 2007, and Mokoaleli-Mokoteli, Taffler, and Agarwal, 2009). 13 Based on the presence of a price pressure effect in prior studies, we expect the stock market to react positively to Cramer s buy recommendations, followed by a reversal. Therefore, if AR_IS (or CAR_IS) is negative (positive) and significant, the option market is less (more) responsive to the noisy information than the stock market. We can conclude that the option market behaves more (less) efficiently than the stock market. Alternatively, if AR_IS (or CAR_IS) is not significantly different from zero, the option market has the same response to noisy information as the stock market. We can conclude the option and stock markets behave with equal efficiency in response to noisy information. Based upon findings in the literature that sell recommendations are more credible than buy recommendations, we expect the stock market to react negatively to the sell recommendations without a price reversal effect. Therefore, if AR_IS (or CAR_IS) is significantly different from zero, the option market does not respond to the credible information (i.e., Cramer s sell recommendations) as the stock market does. Alternatively, if AR_IS (or CAR_IS) is not significantly different from zero, the option market responds to the credible information (i.e., Cramer s sell recommendations) as the stock market does. We can conclude the option market behaves as efficiently as the stock market in response to credible information Data Our data sample is described in Table 1. We initially obtain 923 first-time buy recommendations and 969 first-time sell recommendations from MyMoneyWatch.com from July 2005 through April Indeed, the Mad Money show makes more buy recommendations than sell recommendations. However, more stocks are repeatedly recommended as buys than sells. Therefore, we have more first-time sell recommendations than first-time buy recommendations. We then use the recommendations from a second source, TheStreet.com, to verify the validity of our original sample. As a result, we remove 107 buy recommendations and 88 sell recommendations that appear in MyMoneyWatch.com but not in The- Street.com. After that, we follow Asquith, Mikhail, and Au (2005), Hansen and Zhou (2005), and Altinkilic and Hansen (2009) to remove any confounding news that can contaminate our tests. We remove those observations with news in the 3-day window surrounding the recommendations. Based on I/B/E/S, Security Data Company (SDC), and Factiva.com in search for news, we classify the news into four categories: (1) earnings reports, (2) earnings guidance (not associated with earnings reports), (3) stock repurchase and issuance, and (4) others (including debt financing, debt credit rating, merger and acquisition, asset sale, workforce cut, divestiture, accounting issues and lawsuits). After matching our sample with the news, we discard 418 buy recommendations and 465 sell recommendations that are associated with news. We note that more than half of the recommendations made by Cramer are relevant to firm news. We then match the remaining 13 Similar findings are also found in Womack (1996), Michaely and Womack (1999), Green (2006), and Clarke, Ferris, Jayaraman, and Lee (2006). 14 It is also worth noting that the expected differential responses to Cramer s buy and sell recommendations depend not only on the perceived credibility of Cramer s recommendations, but also on the trader types. Mikhail et al. (2007) find that small investors trade more often on buy recommendations than sell recommendations, while large traders tend to be net sellers following downgrade and sell recommendations. Since large traders are more likely to be informed, the price effect of their trade tends to be permanent rather than transitory. Table 1 The description for sample size. This table reports the number of observations for both buy and sell recommendations in this study. The observations included in this study must satisfy four criteria. First, the recommendations are first-time recommendations, since the viewer always remembers the first recommendation and is more likely to act on the first recommendation than others. Second, the recommendations must appear in both MyMoneyWatch.com and TheStreet.com. This is to eliminate possible disagreements between the two data sources. Third, the recommendations are not associated with any news in the 3-day window surrounding the recommendation. This is to eliminate any confounding effects. Fourth, the stocks must have options listed during our sample period. First-time appearance in Mad Money TV show Disagreement between MyMoneyWatch.com and TheStreet.com Recommendations appearing in both resources News: Earnings reports Earnings guidance (not associated with earnings reports) Stock repurchase and issuance Others Remaining recommendations after removing news related recommendations Stocks without options listed during our sample period Observations included in this study MyMoneyWatch.com TheStreet.Com Buy Sell Buy Sell recommendations with the OptionMetrics data. After the matching, 317 buy recommendations and 337 sell recommendations are available for analysis. We use CRSP daily stock returns, the CRSP value-weighted index returns, and the Fama French SMB and HML data to estimate benchmark stock returns and calculate CARs. The daily trading volume and shares outstanding are used to calculate the abnormal trading volume. We use daily closing bid and ask prices to calculate the abnormal spread. We also explore a possible size effect by partitioning sample firms into small-, medium-, and large-capitalization categories based on the firm s market capitalization. We obtain option data from OptionMetrics which provides daily bid-ask quotes, open interest, trading volume, implied volatility, corresponding stock prices and dividends. To mitigate the volatility term-structure effect, options with maturity of 5 90 days are used to derive the option-implied stock returns. We discard options with absolute value of delta ( delta ) smaller than 0.02 or greater than 0.98 to account for thin trading. To analyze the potential interaction between the option moneyness and the test results, we classify the options into at-the-money (ATM), in-the-money (ITM), and out-of-the-money (OTM). Similar to Bollen and Whaley (2004), we use the delta to classify the options into different moneyness. Essentially, moneyness in option markets is constructed to reflect an option s likelihood of being inthe-money at expiration. Typically, it is measured as the relative difference between the price of the underlying asset and the option s exercise price. The greater (lower) the level of moneyness, the more (less) likely a call (put) will be exercised at expiration. However, this traditional measure for moneyness fails to account for the fact that the likelihood that the option will be in-the-money at expiration also depends heavily on the volatility rate of the underlying asset and the time remaining to expiration of the option. To account for these effects, we measure moneyness using the option s delta, which ranges in value from zero to one, and may be loosely interpreted as the risk-neutral probability that

6 2006 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) Table 2 Descriptive statistics for stocks and options. This table shows the basic descriptive statistics for stocks and options. Panel A reports the basic statistics for the stocks. The stocks are equally divided into three groups by market capital. The market capital equals the average market capital value (price times shares outstanding) of the stock from day 1 to 5. The return is daily log return. Spread is the bid-ask spread divided by the closing price, and turnover equals trading volume divided by the shares outstanding. Panel B reports the basic statistics for options. Volume Ratio is scaled by 100. We use the options with time to maturity between 5 and 90 days and with an absolute value of delta between 0.02 and These options are classified into at-the-money (ATM), in-the-money (ITM), and out-of-the-money (OTM) based on delta. ATM options have delta values between 0.4 and 0.6; ITM, between 0.6 and 0.98; and OTM, between 0.4 and The option spread is calculated in the same manner as the stock spread. The option volume ratio is calculated as the option trading volume divided by the stock trading volume. Panel A: Descriptive statistics for stocks Buy recommendations Sell recommendations All Small-cap Mid-cap Large-cap All Small-cap Mid-cap Large-cap Obs Size ($ mil) 14, ,557 11, ,753 Spread 0.10% 0.15% 0.09% 0.06% 0.12% 0.19% 0.11% 0.07% Volume (,000) Turnover 1.20% 1.72% 1.19% 0.70% 1.33% 1.77% 1.36% 0.87% Beta Panel B: Option descriptive statistics for the buy recommendations Call options Put options Obs. Spread (%) Volume Volume ratio 10 2 (%) Spread (%) Volume Volume ratio 10 2 (%) ALL Small-cap ALL ATM ITM OTM Mid-cap ALL ATM ITM OTM Large-cap ALL ATM ITM OTM Panel C: Option descriptive statistics for the sell recommendations Call options Put options Obs. Spread (%) Volume Volume Ratio 10 2 (%) Spread (%) Volume Volume Ratio 10 2 (%) ALL Small-cap ALL ATM ITM OTM Mid-cap ALL ATM ITM OTM Large-cap ALL ATM ITM OTM the option will be in-the-money at expiration. In this study, ATM refers to the options with delta value between 0.4 and 0.6, ITM, between 0.6 and 0.98, and OTM, between 0.4 and We also eliminate options with no open interest and stocks with price less than $5. To avoid possible recording errors, we discard option pairs if either put or call options have a bid-ask spread that is greater than 50% of the option price at the midpoint. We calculate daily continuously compounded interest rates using a zero 15 For a robustness check, we also run our tests based on the traditional approach to measure the moneyness. In the robustness check, we classify call options as ATM if S/K is between 0.9 and 1.1, where S is the stock price and K is the strike price; ITM if S/K is greater than 1.1; OTM if S/K is lower than 0.9. We classify put options in a similar manner. The test results are qualitatively similar. curve provided by OptionMetrics. We interpolate to match the corresponding time to maturity. Because each option-implied stock price is constructed from a put-call pair with the same time to maturity, strike price, and stock price, it can be derived from a pair of ATM call and ATM put, ITM call and OTM put, or OTM call and ITM put. On any given date and for any given stock there may be multiple pairs that satisfy the maturity and moneyness criteria. We use the average of the option-implied stock prices from those multiple pairs as the proxy for the option-implied stock price for the stock on that date. Table 2 shows the basic descriptive statistics for stocks and options during the estimation period. Panel A reports the basic statistics for the stocks. The stocks are divided equally into three groups by market capitalization. The market capital is calculated as the

7 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) Table 3 Results for stock abnormal returns and option-implied abnormal returns. AR i;t ¼ R i;t ð^a i þ ^b mr mt þ ^b SMBSMB t þ ^b HMLHML tþ¼e 0 i;t ; t ¼ 1; 2;...; 30: ð4þ AR IS i;t ¼ R e it ð^a i þ ^b i R it Þ¼e 0 it ; t ¼ 1; 2;...; 30: ð11þ This table reports the test results for the abnormal returns and cumulative abnormal returns based on the actual stock prices and the option-implied stock prices. AR denotes the stock abnormal returns and CAR is the cumulative stock abnormal returns estimated based on the three-factor model. AR_IS and CAR_IS refer to the abnormal returns and cumulative abnormal returns based upon the option-implied stock prices, respectively. The t-stat is calculated based on Boehmer, Poulsen, and Musumeci (1991). Buy recommendations Sell recommendations Day AR (%) t-stat CAR (%) t-stat AR (%) t-stat CAR (%) t-stat Panel A: Abnormal returns and cumulative abnormal stock returns * * * * * * * * * * * * 2.04 Panel B: Abnormal returns and cumulative abnormal option-implied returns Buy recommendations Sell recommendations AR_IS (%) t-stat CAR_IS (%) t-stat AR_IS (%) t-stat CAR_IS (%) t-stat * * * * * * * * * Indicates the test statistic is significant at less than the 5% level. average market capital value (price times shares outstanding) of the stock from day 1 to 5. Spread is the bid-ask spread divided by the average bid-ask, and Turnover equals the trading volume divided by the shares outstanding. For both buy and sell recommendations, small-cap stocks have higher bid-ask spreads, higher share turnover, and higher betas. For example, for buy recommendations, the bid-ask spread for small-cap stocks (0.15%) is more than twice that of the large-cap (0.06%). This is consistent with the notion that small-cap market makers face greater adverse selection and (or) higher inventory/order processing costs (see Madhavan et al. (1997), and Liang (1999)). Panel B reports the basic statistics for options of buy recommendations. For all of the options, the average option spreads are 16.60% and 16.38% for calls and puts, respectively. The option trading volume ratio for call options is 5.79% 10 2, which is higher than 3.56% 10 2 for put options. This indicates that call options have higher liquidity than put options. As we classify the options by size and moneyness, we find that large-cap stocks have a lower option bid-ask spread and a greater option volume ratio than small-cap stocks. For example, the volume ratio of call options for large-cap stocks is 7.04% 10 2, whereas it is 4.99% 10 2 for the small-cap stocks. This result suggests that the options on large-cap stocks have higher liquidity than those on small-cap stocks. We find that OTM options have the highest volume ratio and liquidity among all the options. For large-cap stocks, for instance, the volume ratio for OTM put options is 2.40% 10 2, while it is only 1.22% 10 2 for ATM put options. Panel C exhibits the basic statistics for options associated with the sell recommendations. Similar to Panels B and C shows that the call options are more heavily traded than put options, and large-cap stock options are more heavily traded than small-cap stock options. 4. Empirical results In this section, we report abnormal returns and abnormal trading activities in the stock and option markets following Cramer s recommendations. Section 4.1 shows results for the entire sample; Section 4.2 presents results based upon firm size; and Section 4.3 reports results based on option moneyness Abnormal returns and trading activities Table 3 reports the abnormal returns and cumulative abnormal returns for the stocks and options. AR denotes the abnormal stock returns and CAR is the cumulative abnormal stock returns estimated based on the three-factor model discussed earlier. AR_IS and CAR_IS refer to the abnormal returns and cumulative abnormal returns for the option-implied stock prices, respectively. The t-statistic is calculated based on Boehmer et al. (1991). 16 Abnormal returns for the stocks from day +1 to day +30 along with cumulative abnormal returns are reported in Panel A. For the buy recommendations, results in Panel A indicate that the abnormal stock return, AR, on day +1 (the day the market first trades on the information) is positive 1.42%, which is significant at less than the 5% level. However, this price effect lasts for one day only. AR turns negative starting at day +2. The cumulative abnormal stock return, CAR, shows that the price pressure effect on 16 This test statistic is not affected by event-induced variance changes. We also calculate the t-stat based on Patell (1976) and perform sign tests. All results are qualitatively the same. To save space, we only report the t-stat based on Boehmer, Poulsen, and Musumeci (1991).

8 2008 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) day +1 is carried through until day +5. It eventually becomes a 1.05% and statistically significant at less than 5% level on day +30. This result indicates a significant price reversal following the price jump on day +1. For Cramer s sell recommendations, we find that AR is negative 0.41% on day +1. CAR is negative for all 30 trading days following the sell recommendations, decreasing from 0.41% on day +1 to 1.91% on day +30. This result suggests that, unlike the buy A AR of buy recommendations AR of sell recommendations B AR_IS of buy recommendations AR_IS of sell recommendations C CAR of buy recommendations CAR of sell recommendations D CAR_IS of buy recommendations CAR_IS of sell recommendations Fig. 1. Abnormal stock returns and abnormal option-implied stock returns (day 4 to +30). This figure plots the abnormal stock returns, and abnormal option-implied stock returns from day 4 to 30. The abnormal stock returns and cumulative abnormal stock returns are calculated based on Eqs. (4) and (7), respectively. The abnormal optionimplied stock returns and cumulative abnormal option-implied stock returns are specified in Eqs. (11) and (12), respectively. (A) Abnormal stock returns; (B) abnormal option-implied stock returns; (C) cumulative abnormal stock returns and (D) cumulative abnormal option-implied returns.

9 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) recommendations, the sell recommendation stocks do not experience a price reversal. This is consistent with extant literature, where sell recommendations are more credible than buy recommendations. (e.g., Stickel, 1995; Michaely and Womack, 1999; Barber, Lehavy, McNichols, and Trueman, 2001; Carleton, Chen, and Steiner, 2002; Mikhail, Walther, and Willis, 2007; Cliff, 2007, and Mokoaleli-Mokoteli, Taffler, and Agarwal, 2009). Panel B in Table 3 reports the test results in the option market. For the buy recommendations, the abnormal option-implied stock return, AR_IS, is a negative 0.23% on day +1, which is significant at less than the 5% level. Based on the pattern of the cumulative abnormal option-implied stock return, CAR_IS, we also find a negative and significant abnormal option-implied return on day +1, which remains significant up to day +15. This finding suggests that there exists a price pressure effect in the stock market that is not duplicated in the option market in response to Cramer s buy recommendations. This is consistent with Wang s (2010) argument that informed investors take large opposite positions against the naive speculators when these speculators trade on noise as if it were information. Also consistent with Ofek et al. (2004), we find that traders in stock and option markets react differently in the wake of noisy information arrival. Based upon the evidence presented here, we assert that the option market is of better quality than the stock market. In contrast, for the sell recommendations, neither AR_IS nor CAR_IS is significant. This finding suggests that the option market moves in step with the stock market upon the arrival of the sell recommendations. Both the stock market and the options market respond to credible information similarly, and no price reversal is observed. Fig. 1 plots the abnormal returns and cumulative abnormal returns for the stock and option markets. Abnormal stock returns (AR) and abnormal option-implied stock returns (AR_IS) are shown in Fig. 1A and B, respectively. Fig. 1A shows that AR for the buy recommendations is significantly positive on day +1, while Fig. 1B shows that AR_IS is significantly negative on day +1. This pattern is consistent with the argument that option market does not respond to Cramer s buy recommendations the same manner as the stock market. Cumulative abnormal returns are presented in Fig. 1C and D, and the same pattern emerges. On the other hand, for the sell recommendations, both AR and AR_IS are negative on day +1 as shown in Fig. 1A and B, respectively. AR_IS moves in step with AR, indicating that both the option and stock markets react to the credible information similarly. Moreover, both CAR and CAR_IS for the sell recommendations in Fig. 1C and D, respectively, do not show any sign of price reversal, suggesting that the sell recommendations are more credible than the buy recommendations. In addition to the price effect, we also examine the bid-ask spread and trading volumes, and relevant results are reported in Table 4. InTable 4, AS is the abnormal stock spread; AT refers to the abnormal turnover; AS_Call and AS_Put are abnormal option spreads for calls and puts, respectively; AVR_Call and AVR_Put are the abnormal option trading volume ratios for call and put options, respectively. In Panel A, the stock spread (AS) for the buy recommendations decreases significantly on day +1 and lasts more than 5 days. Market microstructure theory may suggest that market makers narrow the spread because they face less information asymmetry and adverse selection in the wake of Cramer s recommendations. Nevertheless, the abnormal stock spread goes back to normal after 5 days. The abnormal stock turnover (AT) for the Table 4 Results on the bid-ask spreads and trading activities. This table reports the test results on the bid-ask spread and trading activities for both stocks and options for the entire sample. AS refers to the abnormal stock spread, specified in Eq. (22). AT is stock turnover computed based on Eq. (16). AS_Call is the abnormal spread for call options and AS_Put for put options. AVR_Call and AVR_Put are abnormal option trading volume ratio for calls and puts scaled by 100, respectively. The abnormal option spreads and option trading volume ratios are calculated in the same manner as the stock spread based on Eqs. (19) (23). Buy recommendations Sell recommendations Day AS (%) t-stat AT (%) t-stat AS (%) t-stat AT (%) t-stat Panel A: Abnormal stock spreads and abnormal turnover * * * * * * * * * * * * * * * * * Panel B: Abnormal option spreads and abnormal volume ratio for call options AS_Call (%) t-stat AVR_Call 10 2 (%) t-stat AS_Call (%) t-stat AVR_Call 10 2 (%) t-stat * * * * * * * * * * * * * * * Panel C: Abnormal option spreads and abnormal volume ratio for put options AS_put (%) t-stat AVR_put 10 2 (%) t-stat AS_put (%) t-stat AVR_put 10 2 t-stat * * * * * * * * * * * * * * * * * * * * Indicates the test statistic is significant at less than the 5% level.

10 2010 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) A AS of buy recommendations AS of sell recommendations B AS_Call of buy recommendations AS_Call of sell recommendations C AS_PUT of buy recommendations AS_PUT of sell recommendations D AT of buy recommendations AT of sell recommendations E AVR_Call of buy recommendations AVR_Call of sell recommendations F AVR_Put of buy recommendations AVR_Put of sell recommendations Fig. 2. The abnormal bid-ask spread and trading activity for stocks and options (day 4 to +30). This figure plots the abnormal bid-ask spreads and trading activity for stocks and options from day 4 to 30. The abnormal stock spread, is specified in Eq. (22). The abnormal stock turnover is computed based on Eq. (16). The abnormal option spreads and option trading volume ratios are calculated in the same manner as the stock spread based on Eqs. (19) (23). (A) Abnormal stock spreads; (B) abnormal call option spreads; (C) abnormal put option spreads; (D) abnormal stock turnover; (E) abnormal trading volume ratios for call options and (F) abnormal trading volume ratios for put options.

11 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) buy recommendations increases dramatically on day +1, and the abnormal trading activities last 4 days. These results show that Cramer s buy recommendations not only bring price pressure to stock prices, but they also affect the trading activities for the initial four trading days after the event. Panel A also exhibits AS and AT for the sell recommendations. AS drops significantly on day +1, suggesting that sell recommendations also reduce information asymmetry and adverse selection in the wake of Cramer s recommendations. At the same time, AT increases for 3 days following the sell recommendations. This suggests that Cramer s sell recommendations affect trading activity in the stock market as well. Panel B shows that the bid-ask spread for call options (AS_Call) decreases significantly for the buy recommendations, which is similar to the reaction in the stock market. This implies that option market makers are also aware of the price pressure effect and encounter less information asymmetry. Similar to the turnover in the stock market, the option trading volume ratio for calls (AVR_Call) increases significantly as well. The significant abnormal call trading volume lasts for 5 days. For the sell recommendations, AS_Call decreases for 2 days. It shows that the sell recommendations may also reduce the information asymmetry faced by option market makers. AVR_Call increases significantly for the 2 days after the sell recommendations, suggesting that trading activities on call options are also affected by sell recommendations. Panel C reports the impact of the recommendations on put options. Similar to the findings in call options for the buy recommendations, the bid-ask spread for put options decreases significantly. The significant and negative put spread persists for 5 days. The trading activity for put options also increases significantly on day +1 and lasts more than 5 days after the buy recommendations. Bid-ask spreads and trading activities respond to the sell recommendations similarly. AS_Put drops on day +1 significantly, suggesting less information asymmetry and adverse selection following the sell recommendations. Moreover, AVR_Put increases from days +1 to +5. It is worth noting that the jump in AVR_Put on day +1 (1.44% 10 2 ) after sell recommendations is higher than that (0.88% 10 2 ) after buy recommendations, and it is also greater than the jump in AVR_Call (0.74% 10 2 ) after sell recommendations (shown in Panel B). This result suggests that put option trading is heavier after sell recommendations than after buy recommendations, and more puts are traded than calls after Cramer makes sell recommendations. The innovations in the abnormal bid-ask spreads are plotted in Fig. 2A C. These figures depict the abnormal stock spread, the abnormal call spread, and the abnormal put spread, respectively. For both buy and sell recommendations, all figures show a significant drop in the bid-ask spread, and the decline in the spread continues for days following day +1. The innovations in abnormal trading activities are plotted in Fig. 2D F. These figures depict the abnormal stock turnover, the abnormal trading volume ratio for call options, and the abnormal trading volume ratio for put options, respectively. All figures show that the trading activity jumps substantially in response to both the buy and sell recommendations on the event day, and the effect then gradually subsides Abnormal returns and trading activities by firm size Since investors are less informed about small stocks, we now investigate whether investors react to Cramer s recommendations Table 5 Event test results for stock abnormal returns and option-implied abnormal returns by size. AR i;t ¼ R i;t ð^a i þ ^b m R mt þ ^b SMB SMB t þ ^b HML HML t Þ¼e 0 i;t ; t ¼ 1; 2;...; 30: ð4þ AR IS i;t ¼ R e it ð^a i þ ^b i R it Þ¼e 0 it ; t ¼ 1; 2;...; 30: ð11þ This table reports the test results for the abnormal returns according to three different firm sizes small-cap, mid-cap, and large-cap. Three firm sizes are based on the average market capital value of the stock from day 1 to 5. AR denotes the abnormal returns estimated based on the three-factor model. AR_IS refers to the abnormal returns calculated from the option-implied stock prices. The t-stat is calculated based on Boehmer, Poulsen, and Musumeci (1991). Day Buy recommendations Sell recommendations AR (%) t-stat AR_IS (%) t-stat AR (%) t-stat AR_IS (%) t-stat Small-cap * * * Mid-cap * * * Large-cap * * * * Indicates the test statistic is significant at less than the 5% level.

12 2012 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) Table 6 Results for the bid-ask spreads and trading activities by size. This table reports the test results for the bid-ask spread and trading activities for stocks and options in three firm size groups small-cap, mid-cap, and large-cap. We group the entire sample equally into three firm sizes based on the average market capital value of the stock from day 1 to 5. AS refers to the abnormal stock spread, specified in Eq. (22). AT is stock turnover computed based on Eq. (16). AS_Call is the abnormal spread for call options and AS_Put for put options. AVR_Call and AVR_Put are abnormal option trading volume ratio for calls and puts scaled by 100, respectively. The abnormal option spreads and option trading volume ratios are calculated in the same manner as the stock spread based on Eqs. (19) (23). Day Buy recommendations Sell recommendations AS (%) t-stat AT (%) t-stat AS (%) t-stat AT t-stat Panel A: The bid-ask spreads and trading activities for stocks Small-cap * * * * * * * * * * * * * * * * * * * Mid-cap * * * * * * * * * * * Large-cap * * * * * * Day AS_Call (%) t-stat AVR_Call 10 2 (%) t-stat AS_Call (%) t-stat AVR_Cal 10 2 (%) t-stat Panel B: The bid-ask spreads and trading activities for call options Small-cap * * * * * * * * * * * * * * Mid-cap * * * * * * * * * * Large-cap * Day AS_Put (%) t-stat AVR_Put 10 2 (%) t-stat AS_Put (%) t-stat AVR_Put 10 2 (%) t-stat Panel C: The bid-ask spreads and trading activities for put options Small-cap * * * * * * * * * * * * * * * * Mid-cap * * * * * * * 3.45

13 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) Table 6 (continued) Day Buy recommendations Sell recommendations AS (%) t-stat AT (%) t-stat AS (%) t-stat AT t-stat * * Large-cap * * * Indicates the test statistic is significant at less than the 5% level. differently based on firm size. In Table 5 we report the abnormal returns partitioned into three firm sizes small-cap, mid-cap, and large-cap. For the sake of brevity, we report the abnormal returns only because the cumulative abnormal returns exhibit a similar pattern as shown in Table 3. Similar to the results in Table 3 based on the entire sample, the price pressure effect in the stock market in response to the buy recommendations occurs only on day +1 regardless of the size of the firm. The abnormal returns across all sized groups turn negative quickly afterward. Moreover, we find that the smaller the size, the greater and more significant the price pressure effect tends to be. For example, AR for the small-cap stocks on day +1 is a positive 2.92% and significant, while it is a much smaller 0.44% for large-cap stocks. It indicates the noisy information has a relatively weaker impact on large-cap stocks, hence, the market for large-cap stocks is more efficient. For the abnormal option-implied stock returns, AR_IS is negative and significant only on day +1 across all the size groups, which resembles the pattern observed in Table 3. For the sell recommendations, AR for small-cap stocks is negative 0.76% (t-stat = 3.07) on day +1, but it drops to 0.27% (t-stat = 2.13) for mid-cap stocks and 0.22% (t-stat = 1.99) for large-cap stocks. This result indicates that Cramer s sell recommendations have a greater impact on small-cap stocks. AR_IS for Table 7 Event test results for option-implied abnormal returns by moneyness. AR IS i;t ¼ R e it ð^a i þ ^b i R it Þ¼e 0 it ; t ¼ 1; 2;...; 30: ð11þ This table reports the test results for implied abnormal return by moneyness. The options are classified into at-the-money (ATM), in-the-money (ITM), and out-of-the-money (OTM) based on delta. ATM options have delta values between 0.4 and 0.6; ITM, between 0.6 and 0.98; and OTM, between 0.4 and AR_IS and CAR_IS denote the abnormal returns and cumulative abnormal returns for the option-implied stock return, respectively. Because each AR_IS is constructed from a pair of call and put with the same time to maturity, strike price, and stock price, it can be estimated based on a pair of ATM Call and ATM put, ITM Call and OTM put, or OTM Call and ITM put. The t-stat is calculated based on Boehmer, Poulsen, and Musumeci (1991). Day Buy recommendations Sell recommendations AR_IS (%) t-stat CAR_IS (%) t-stat AR_IS (%) t-stat CAR_IS (%) t-stat AR_IS based on pairs of ATM call and ATM put * * * * * * * * AR_IS based on pairs of ITM call and OTM put * * * * * * * * AR_IS based on pairs of OTM call and ITM put * * * * * * * Indicates the test statistic is significant at less than the 5% level.

14 2014 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) the sell recommendations turns out to be insignificant across all firm size categories, showing that the stock and option markets do not diverge on the sell recommendations. Table 6 reports the results for the bid-ask spreads and trading activities for stocks and options by firm size. Panel A reports the bid-ask spreads and trading activities for stocks. Similar to the finding in Table 4 for the entire sample, the stock bid-ask spreads narrow significantly on day +1 for both sell and buy recommendations. Furthermore, the smaller the firm size, the deeper the plunge in spreads and the longer the effect lasts. For example, for the buy recommendations, the abnormal stock spread (AS) for small-cap stocks is negative 6.13% on day +1. The significant abnormal stock spreads last more than 10 days. On the other hand, AS for large-cap stocks is negative 1.32%, and the significant spreads last only 3 days. This result suggests the existence of a larger reduction in information asymmetry and adverse selection for smaller stocks. For the abnormal trading activity, regardless of buy or sell recommendations, small-cap stocks have higher abnormal turnover (AT) than other sized stocks. The impact of the recommendations for small-cap stocks also lasts longer. For instance, based on buy recommendations, AT for small-cap stocks is % and statistically significant. This finding indicates that the abnormal trading activity for small-cap stocks almost triples the next day after Cramer makes buy recommendations. This abnormal stock turnover effect lasts more than 10 days. On the other hand, AT for largecap stocks is a much smaller 8.11%. The same findings also apply to the sell recommendations. These results reinforce the price effect reported earlier and suggest that Cramer s recommendations exert a stronger impact on small-cap stocks. The bid-ask spreads and trading activities for call and put options are reported in Panels B and C, respectively. Again, similar to the pattern found in Panel A, abnormal bid-ask spreads decrease in firm size and abnormal trading activities decrease in firm size as Table 8 Results on the bid-ask spreads and trading activities for options by moneyness. This table reports the test results on the bid-ask spread and trading activities for options by three moneyness groups. The options are classified into at-the-money (ATM), in-the-money (ITM), and out-of-the-money (OTM) based on delta. ATM options have delta values between 0.4 and 0.6; ITM, between 0.6 and 0.98; and OTM, between 0.4 and AS_Call is the abnormal spread for call options and AS_Put for put options. AVR_Call and AVR_Put are abnormal option trading volume ratio for calls and puts scaled by 100, respectively. The abnormal option spreads and option trading volume ratios are calculated in the same manner as the stock spread based on Eqs. (19) (23). Day Buy recommendations AS_Call AS_Put ATM (%) t-stat ITM (%) t-stat OTM (%) t-stat ATM (%) t-stat ITM (%) t-stat OTM (%) t-stat Panel A: Abnormal option bid-ask spread * * * * * * * * * * * * * * * * * * * * * ** * * * * * Sell recommendations AS_Call AS_Put * * * * * * * * * * * Buy recommendations AVR_Call 10 2 AVR_Put 10 2 Panel B: Abnormal option volume ratio * * * * * * * * * ** * * * * * * * * * * * * * * * Sell recommendations AVR_Call 10 2 AVR_Put * * * * * * * * * * * * * * Indicates the test statistic is significant at less than the 5% level.

15 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) well for both the buy and sell recommendations. For example, the abnormal call option spread, AS_Call, for small-cap stocks is a negative 2.72% (t-stat = 3.40) on day +1, while it is only 0.43% (t-stat = 0.72) for large-cap stocks. The results in Table 6 suggest that trading activities are firm-size dependent, and this finding holds for both stock and option markets. It is also worth noting that put options for small-caps experience the highest abnormal trading volume ratio in Panel C. For instance, based on sell recommendations, AVR_put on day +1 for small-cap stocks is 2.09% 10 2 (t-stat = 9.64), while it is only 0.69% 10 2 (t-stat = 4.31) for largecap stocks. This finding indicates that put options for small-cap stocks are the most heavily traded among all the options on day Abnormal returns and trading activity by option moneyness As option traders may exhibit a preference for certain option deltas, it is interesting to examine the differential price effect, if any, across option moneyness. Table 7 reports such results for implied abnormal returns by option moneyness. As we divide the entire option sample into three groups based on moneyness, we find that AR_IS for the buy recommendations are negative and significant across all the moneyness groups, which is consistent with the findings in Tables 3 and 5. These negative and significant AR_IS values reaffirm that the option market is less responsive to the noisy information than the stock market. Although AR_IS for the buy recommendations is negative and significant only on day +1, the cumulative effect of the disagreement between the option and stock markets actually persists for up to 15 days as shown by CAR_IS. Among the three moneyness categories, it is also worth noting that AR_IS based on pairs of ITM call and OTM put shows the strongest deviation between the option-implied stock returns and actual stock returns. For instance, AR_IS based on pairs of ITM call and OTM put on day +1 is 0.27%, which is twice as large as 0.13% for AR_IS based on pairs of OTM call and ITM put. This test result is consistent with Chakravarty et al. (2004) and Xing et al. (2010) in that OTM put options are more heavily employed in informed trading. As option investors predict the price pressure effect and trade more heavily on OTM puts, AR_IS based on pairs of ITM call and OTM put may display greater deviation between the optionimplied stock returns and actual stock returns. On the other hand, all AR_IS and CAR_IS estimates associated with the sell recommendations are insignificant. This finding suggests that options across different monenyess categories move in step with the underlying stocks because the sell recommendations, unlike the buy recommendations, are not perceived as noisy to investors. Thus, sell recommendations do not cause the stock and option markets to diverge. Table 8 presents the results on the abnormal bid-ask spreads and trading activities for options by moneyness. Similar to the findings in Tables 4 and 6, Panel A shows the abnormal option bid-ask spread drops significantly on day +1 across most of the moneyness groups for both call and put options. Panel B shows that the abnormal option volume ratio is positive and significant on day +1 for all moneyness groups. These findings indicate that the response of the option market to Cramer s recommendations is consistent regardless of option moneyness. In addition, among all the moneyness categories for calls and puts, we can see that OTM put options experience the greatest abnormal volume ratio for both the buy and sell recommendations. For example, in Panel B for buy recommendations, AVR_Put in the OTM category is 0.72% 10 2 on day +1, while it is only 0.54% 10 2 for AVR_Call in the OTM category, 0.31% 10 2 for AVR_Call in the ITM category, and 0.26% 10 2 for AVR_Put in the ATM category. Consistent with the result in Table 7 for buy recommendations, this finding shows that investors predicting the price reversal tend to trade more heavily on OTM puts, which makes AR_IS based on a pair of ITM call and OTM put experience greater divergence between the option-implied stock returns and actual stock returns. 5. Robustness checks 5.1. Sub-period analysis In this section, we conduct a sub-period analysis to examine whether market participants adjust their trading behavior over the sample period, and to test whether the price effect is robust over time. Table 9 reports the results for the abnormal returns and cumulative abnormal returns based on the actual and option-implied stock prices in sub-periods. The first sub-period is from July 2005 through June 2006; the second sub-period is from July 2006 through April Similar to the findings in Table 3, test results show that the price pressure effect occurs in both sub-periods under buy recommendations. The abnormal stock return is 1.48% for the first sub-period and 1.31% for the second sub-period. However, the cumulative effect in the stock market lasts for more than 5 days for the first sub-period, while it lasts for 3 days in the second sub-period. Option market test results are again similar to the findings in Table 3. The abnormal option-implied stock return is negative and significant on day +1 for both sub-periods. It shows that the option market consistently exhibits a strong and negative deviation from the stock market as the stock market experiences a price pressure effect. As shown in Panel B, the test results on sell recommendations are also similar to the findings in Table 3. AR is 0.42% and significant in the first sub-sample period; 0.40% and also significant in the second sub-period. AR_IS and CAR_IS do not show any significant deviation between the stock and option markets. Again, this indicates that sell recommendations are more credible than buy recommendations, and both the stock and options markets respond to the credible information similarly. Therefore, our test results appear robust across different sub-periods Model specification, liquidity, and time to maturity This section provides several additional robustness tests to deal with any model-misspecification, liquidity, and options time-tomaturity issues. First, because the option-implied stock return is calculated based on the put-call parity with adjustment of early exercise premium for American options, we use the midpoint value between bid and ask option prices in the calculation. The early exercise premium is estimated based on Barone-Adesi and Whaley s (1986) American option pricing model. As a result, the test results could be affected by market frictions and modelmisspecification. To mitigate this potential problem, we propose a new measure, Divergence, to gauge the deviation between stock and option markets based on the option boundary conditions for American options. This measure incorporates option bid and ask quotes and does not depend on any option pricing models. Thus, the test results from Divergence should not be affected by market frictions and/or model-misspecification. The Divergence variable is constructed in the following way. With market frictions, the upper boundaries for American call and put options are specified in inequalities (25) and (26), respectively: ðp a þ S a Ke rxt ÞþðT X þ T S þ T P ÞC b T C ; ðc a S b þ D þ Ke rxt ÞþðT X þ T S þ TÞCÞ P b T P ; ð25þ ð26þ where S, P, C, K, r, s and D refer to the observed stock price, put premium, call premium, strike price, the risk-free rate, time to

16 2016 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) Table 9 The event test results for stock abnormal returns and option-implied abnormal returns in sub-periods. AR i;t ¼ R i;t ð^a i þ ^b m R mt þ ^b SMB SMB t þ ^b HML HML t Þ¼e 0 i;t ; t ¼ 1; 2;...; 30: ð4þ AR IS i;t ¼ R ~ it ð^a i þ ^b i R it Þ¼e 0 it ; t ¼ 1; 2;...; 30: ð11þ This table reports the event study test results for the abnormal returns and cumulative abnormal returns based on the actual stock prices and the option-implied stock prices in sub-periods. We divide the entire sample period into two sub-periods. The first half of the sample period is from July 2005 through May 2006, the second half, from June 2006 through April AR denotes the abnormal returns and CAR is the cumulative abnormal returns estimated based on the three-factor model. AR_IS and CAR_IS refer to the abnormal returns and cumulative abnormal returns for the option-implied stock return, respectively. The t-stat is calculated based on Boehmer, Poulsen, and Musumeci (1991). Day Sub-period 1 (July 2005 May 2006) Sub-period 2 (June 2006 April 2007) Obs. AR (%) t-stat AR_IS (%) t-stat Obs AR (%) t-stat AR_IS (%) t-stat Panel A: Buy recommendations * * * * Obs CAR t-stat CAR_IS t-stat Obs CAR t-stat CAR_IS t-stat * * * * * * * * * * * * * * * * * * * * * Sub-period 1 (July 2005 May 2006) Sub-period 2 (June 2006 April 2007) Obs. AR (%) t-stat AR_IS (%) t-stat Obs AR (%) t-stat AR_IS (%) t-stat Panel B: Sell recommendations * * Obs CAR (%) t-stat CAR_IS (%) t-stat Obs CAR (%) t-stat CAR_IS (%) t-stat * * * * * * * Indicates the test statistic is significant at less than the 5% level. maturity, and the present value of cash dividends during the life of the option, respectively. The superscripts, a and b, denote ask and bid of the quotes, respectively. T X, T S, T P, and T C are the transaction costs for exercising options, trading stocks, trading puts, and trading calls, respectively. Given inequality (25), the lower boundary of a stock price in option markets can be expressed as: S a C b P a þ Ke rxt ðt X þ T S þ T P þ T C Þ¼L ow : ð27þ According to inequality (26), the upper boundary of a stock price in option markets is: S a P b þ D þ K þðt X þ T S þ T P þ T C Þ¼H igh : ð28þ These two inequalities yield a range for the option-implied stock price. L ow S a measures the distance between the lower bound and the observed stock price. The greater the distance, the higher the call premium is relative to the put premium. By the same token, S b H igh determines the distance between the observed stock price and the upper bound. The greater the distance, the higher is the put premium relative to the call premium. Hence, the difference between (L ow S a ) and (S b H igh ) gauges how the optionimplied stock price relates to the actual stock price. It can be specified as: Divergence ¼ðL ow S a Þ ðs b H igh Þ ¼ðC a þ C b P a P b Þ ðs a þ S b DÞþKð1 þ e rs Þ: ð29þ Eq. (29) suggests that the larger the Divergence, the higher the option-implied stock price relative to the observed stock price, and vice versa. 17 The abnormal Divergence change, AR_D, is calculated in the same manner as the abnormal option-implied stock returns in Eq. (11). 17 This argument is consistent with Bodurtha and Courtadon (1986).

17 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) Table 10 Robustness tests for the response of the option market. This table reports the robustness tests for the response of the option market to Cramer s recommendations. Panel A reports a model-free measure of the divergence between stock and option markets based on the option boundary conditions. AR_D denotes the abnormal change of the divergence between stock and option markets, which is calculated according to Eq. (29). CAR_D is the cumulative abnormal change of the divergence. Panel B shows the test results based on the options with minimal trading volume of 5 contracts. Panel C exhibits the test results based on options with time to maturity from 91 to 182 days. AR_IS and CAR_IS refer to the abnormal returns and cumulative abnormal returns for the option-implied stock return, respectively. The t-stat is calculated based on Boehmer, Poulsen, and Musumeci (1991). Day Buy recommendations Sell recommendations AR_D (%) t-stat CAR_D (%) t-stat AR_D (%) t-stat CAR_D (%) t-stat Panel A: Abnormal divergence change * * * * * * * * Day AR_IS (%) t-stat CAR_IS (%) t-stat AR_IS (%) t-stat CAR_IS (%) t-stat Panel B: Abnormal option-implied returns for options that are actually traded * * * * * * * * Panel C: Abnormal option-implied returns for options with longer time to maturity * * * * * * * Indicates the test statistic is significant at less than the 5% level. Second, because some stock options recorded in OptionMetrics may not be actually traded, a liquidity problem could also affect the test results. We alleviate liquidity concerns by excluding options without any trading volume. Third, to this point, we have used the options with time to maturity of 5 90 days in our empirical tests. We now employ options with time to maturity between 91 and 182 days to verify whether our results are robust with respect to time to maturity. Table 10 reports these additional robustness tests results. Panel A reports results for abnormal Divergence changes. AR_D denotes the abnormal change of the Divergence, which is calculated using Eq. (29). CAR_D is the cumulative abnormal change of the Divergence. For the buy recommendations, we find AR_D is 8.65% and significant on day +1. This indicates that the Divergence is 8.65% lower than usual. Thus, investors in the option market disagree with the price effect in the stock market, and their disagreement is reflected in the divergence measure. Therefore, our test results appear robust and unaffected by market frictions or potential model-misspecification problems. In contrast, both AR_D and CAR_D for sell recommendations are insignificant. This suggests that the options and stock markets are in synch in response to the sell recommendations. Panel B shows test results based on the options with trading volume greater than zero in OptionMetrics. The abnormal optionimplied stock return is negative and larger than that reported in Table 3. For example, based on buy recommendations, AR_IS is 0.29%, while it is 0.23% in Table 3. This implies that our results are not likely affected by the liquidity problem. Panel C shows the test results based on the options with time to maturity between 91 and 182 days. Again, the abnormal option-implied stock return for the buy recommendations is negative ( 0.26%) and significant. Thus, our results are robust with respect to time to maturity. 6. Trading profit on the divergence between the stock and option markets So far, we have found that there exists significant divergence between stock and option markets in response to Cramer s buy recommendations, and significant information content appears to be embedded in the option-implied stock return. 18 A logical question thus arises as to whether the significant divergence between the option and stock markets can predict future stock returns and yield an economically significant trading profit. 19 Following Ofek et al. (2004), Xing et al. (2010), and Cremers and Weinbaum (2010), we use a portfolio-forming approach to evaluate stock return predictability. If option-implied stock prices on the event day contain economically significant information about the price reversal following the price pressure effect on day +1, the smaller the abnormal optionimplied stock return (the larger negative value), the greater is the price reversal it predicts. This implies that a low AR_IS should be associated with a low future stock return, and vice versa. To study this possibility, we first form five quintile portfolios based on the AR_IS on day +1 and then calculate the cumulative 18 Because we do not find any significant divergence between stock and option markets for Cramer s sell recommendations, we restrict our trading simulation on the buy recommendations only. 19 Cullen, Gasbarro, and Monroe (2010) report that mutual funds that systematically trade counter to the public information form a subset of the privately informed funds and this group of funds exhibits a superior average performance.

18 2018 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) Table 11 The risk-adjusted return for portfolios formed according to the abnormal optionimplied return on day +1 for the whole sample. This table shows the risk-adjusted return for portfolios formed according to the abnormal option-implied stock return (AR_IS) on day +1 for the whole sample. We form five quintile stock portfolios. The bottom 20% stocks are classified in Portfolio Low, 20 40% in Portfolio 2, 40 60% in Portfolio 3, 60 80% in Portfolio 4, and the top 20% in Portfolio High. All denotes the AR_IS calculated based on all the options. ATM Call/ATM Put refers to the AR_IS calculated based on pairs of ATM call and ATM put options. ITM Call/OTM put refers to the AR_IS computed based on pairs of ITM call and OTM put options. OTM Call/ITM Put denotes the AR_IS calculated from pairs of OTM call and ITM put options. The returns are the cumulative daily risk-adjusted returns (based on Fama French three-factor model) over the event window from day +2 to day +30. Group All ATM Call/ATM put ITM Call/OTM put OTM Call/ITM put Panel A: Trading profits based on the entire sample Low 3.04% 1.99% 3.76% 1.18% % 1.57% 1.09% 1.18% % 1.54% 0.99% 2.25% % 0.98% 0.53% 0.93% High 0.03% 0.55% 0.19% 0.82% High low 3.01% * 1.44% * 3.57% * 0.36% t-stat Panel B: Trading profits by firm size Small-cap Low 4.40% 3.72% 5.23% 1.40% % 1.93% 1.87% 2.24% % 2.11% 1.43% 3.82% % 1.14% 0.80% 1.66% High 0.13% 0.63% 0.17% 0.94% High low 4.27% * 3.09% * 5.06% * 0.46% t-stat Mid-cap Low 2.99% 1.34% 3.75% 0.72% % 1.78% 0.75% 0.21% % 1.76% 0.78% 1.17% % 1.08% 0.44% 0.22% High 0.18% 0.62% 0.16% 0.78% High low 2.80% * 0.72% 3.59% * 0.06% t-stat Large-cap Low 1.72% 0.91% 2.30% 1.41% % 0.99% 0.66% 1.09% % 0.75% 0.75% 1.75% % 0.72% 0.36% 0.91% High 0.21% 0.40% 0.24% 0.74% High low 1.94% * 0.50% 2.06% * 0.67% t-stat * Indicates the test statistic is significant at less than the 5% level. daily risk-adjusted return (based on Fama French three-factor model) over the event window from day +2 to day +30. The bottom 20% stocks are grouped in Portfolio Low, 20 40% in Portfolio 2, 40 60% in Portfolio 3, 60 80% in Portfolio 4, and the top 20% in Portfolio High. If information contained in the AR_IS is predictive, the return of Portfolio High should be significantly greater than the return of Portfolio Low. Table 11 reports the test results. Panel A shows trading profits based on the entire sample period, and Panel B exhibits the trading profits by firm size. In Panel A, the cumulative daily risk-adjusted return increases monotonically from the Low to the High Portfolio. Most of the returns are negative because of price reversal following the price pressure effect on day +1. This indicates that AR_IS is able to predict future stock performance. Moreover, long portfolio High ( 0.03% return) and short portfolio Low ( 3.04% return) produces a positive and significant risk-adjusted return of 3.01%. As we control option moneyness to form portfolios, AR_IS computed from pairs of ITM call and OTM put options shows the highest and most significant trading profit. (High low) return is 3.57%, which suggests that AR_IS based on pairs of ITM calls and OTM puts has the greatest predictability for the price reversal pattern following the price pressure effect on day +1. This result is consistent with the finding in Table 8 where OTM put options are most likely to be traded by investors exploiting the price reversal following buy recommendations. It also complements the finding of Xing et al. (2010) which shows evidence that informed traders with negative news prefer to trade out-of-the-money put options. The trading profit by firm size is reported in Panel B. Two findings are worth mentioning. First, the trading profit decreases with firm size. Based on all of the options, we find that long portfolio high and short portfolio low produces a positive 4.27% riskadjusted return for the small-cap group, while the same statistic is 1.94% for the large-cap group. Second, when we control the option moneyness to form portfolios, the AR_IS computed based on pairs of ITM call and OTM put options shows the highest and most significant trading profit. (High low) return is 5.06% for the small-caps, while the number is reduced to 2.06% for the largecaps. Similar to the results reported in Panel A, the primary source of the trading profit comes from the pairs of ITM call and OTM put. Again, this result is consistent with the finding in Table 8, showing that OTM put options are the most likely to be traded. Although Table 11 suggests profitable opportunities may exist, we caution the interpretation of these profitability results. Compared to significant trading profits in other studies, a riskadjusted return of 5.06% for holding the portfolio for 29 trading days seems overly optimistic. For example, using the information content in volatility smirk, Xing et al. (2010) document an annual risk-adjusted return of 10.9%. Cremers and Weinbaum (2010) employ the information from put-call disparity and find a weekly return of 50 basis points. Nevertheless, the trading return of 5.06% found in this study holds only if investors can freely short the stock portfolio. In a market with short-sale constraints and/or high short-sale costs, the return will clearly be reduced. 7. Conclusions This paper explores the relative market efficiency of the stock and option markets in response to the arrival of information in the form of buy and sell recommendations made by a popular television show host, Jim Cramer. Extant literature has dealt extensively with lead lag relations and the price discovery process between these two markets. An unexplored issue is whether stock and option markets behave with equal efficiency upon the arrival of noisy information. Little attention has been given to the unusual situation in which the two markets diverge. If the stock and option markets are occasionally out of sync with each other, it is important to discover the reason, and to determine which market behaves more efficiently. Our empirical results have implications for the preferred habitat of informed trading hypothesis. We find that a short-lived price pressure effect associated with the buy recommendations of Mad Money show host Jim Cramer provides an ideal experimental environment. We conduct event studies to examine the effect of Mr. Cramer s buy and sell recommendations on stock and option markets. For the buy recommendations, small-cap stocks show a strong shortlived price run-up followed by a price reversal, consistent with the findings in Balcarcel and Chen (2007) and Engelberg et al. (2009). We assert this behavior may represent an irrational reaction of naïve investors to Mr. Cramer s buy recommendations. However, if as past research suggests (see, e.g., Black, 1975; Cox and Rubinstein, 1985; Easley et al., 1998, and Chakravarty et al., 2004), the option market is the preferred habitat for informed traders, option-implied stock prices should be less responsive or completely unresponsive to the noisy information. Thus a divergence between the actual stock price and option-implied stock price occurs, especially in the presence of short-sale constraints and in the

19 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) absence of arbitrage. 20 If informed investors anticipate the price pressure effect from Cramer s buy recommendations, they may even trade against naïve stock investors, causing the option-implied stock price to be abnormally lower than the actual stock price. Our empirical results show that the option-implied stock returns are significantly and abnormally smaller than actual stock returns around the event window of Mr. Cramer s buy recommendations. Hence the price pressure effect observed in the stock market is missing in the option market. We conclude that the option market behaves more efficiently in response to noisy information than the stock market. In addition to the lower option-implied stock returns, we also find higher trading volumes and narrower bid-ask spreads. The increased trading volumes and narrower option spreads imply some option trades are bets against stock trades on the buy recommendations. Trading strategies taking the advantage of this market divergence, i.e., long portfolio with high divergence and short portfolio with low divergence, yield a remarkable profit as much as 5.06% over our event window from day +2 to day +30. For Cramer s sell recommendations, however, we find that the stock and option markets move in step with each other without any significant divergence. This finding should not be surprising since current literature has documented that sell recommendations are more credible than buy recommendations. Thus, the stock and option markets react to credible information similarly. Acknowledgement Ying Huang acknowledges the financial support provided by the Program for New Century Excellent Talents in University (NCET) of China. References Altinkilic, O., Hansen, R.S., On the information role of stock recommendation revisions. Journal of Accounting and Economics 48, Asquith, P., Mikhail, M.B., Au, A.S., Information content of equity analyst reports. Journal of Financial Economics 75, Balcarcel, A.L., Chen, C.R., Mad Money: Jim Cramer vs. Efficient Market. Working Paper, University of Dayton. Barber, B., Lehavy, R., McNichols, M., Trueman, B., Can investors profit from the prophets? Security analyst recommendations and stock returns. The Journal of Finance 56, Barone-Adesi, G., Whaley, R.E., The valuation of American call options and the expected ex-dividend stock price decline. Journal of Financial Economics 17, Battalio, R., Schultz, P., Options and the bubble. Journal of Finance 61, Biais, B., Hillion, P., Insider and liquidity trading in stock and options markets. Review of Financial Studies 7, Black, F., Fact and fantasy in the use of options. Financial Analysts Journal 31, Bodurtha, J., Courtadon, G., Efficiency tests of foreign currency options market. Journal of Finance 41, Boehmer, E., Poulsen, A., Musumeci, J., Event study methodology under conditions of event-induced variance. Journal of Financial Economics 30, Bollen, N., Whaley, R., Does net buying pressure affect the shape of implied volatility functions? The Journal of Finance 59, Carhart, M.M., Kaniel, R., Musto, D.K., Reed, A.V., Leaning for the tape: evidence of gaming behavior in equity mutual funds. Journal of Finance 57, Carleton, W., Chen, C.R., Steiner, T., Optimism biases among brokerage and non-brokerage firms equity recommendations: agency costs in the investment industry. Financial Management 31, Chakravarty, S., Gulen, H., Mayhew, S., Informed trading in stock and option markets. Journal of Finance 58, Chan, K., Chung, Y.P., Fong, W.M., The informational role of stock and option volume. Review of Financial Studies 15, Chan, K., Chung, Y.P., Johnson, H., Why option prices lag stock prices: a trading based explanation. Journal of Finance 48, The large bid-ask spread in the option market may reduce the arbitrage opportunity. See also arguments about the short sale constraint on put-call parity in Ofek et al. (2004). Chen, Y.L., Gau, Y.F., News announcements and price discovery in foreign exchange spot and futures markets. Journal of Banking and Finance 34, Clarke, J., Ferris, S.P., Jayaraman, N., Lee, J., Are analyst recommendations biased? Evidence from corporate bankruptcies. Journal of Financial and Quantitative Analysis 41, Cliff, M.T., Do affiliated analysts mean what they say? Financial Management 36, Cohen, R.B., Gompers, P.A., Vuolteenaho, T., Who underreacts to cash-flow news? Evidence from trading between individuals and institutions. Journal of Financial Economics 66, Corwin, S.A., The determinants of underpricing for seasoned equity offers. Journal of Finance 58, Coval, J., Stafford, E., Asset Fire Sales in Equity Markets. Working Paper, Harvard Business School. Cox, J.C., Rubinstein, M., Option Markets. Prentice-Hall, Englewood Cliffs, New Jersey. Cremers, M., Weinbaum, D., Deviations from put-call parity and stock return predictability. Journal of Financial and Quantitative Analysis 45, Cullen, G., Gasbarro, D., Monroe, G.S., Mutual fund trades and the value of contradictory private information. Journal of Banking and Finance 34, Diltz, J.D., Kim, S., The relationship between stock and option price changes. Financial Review 31, Easley, D., O Hara, M., Srinivas, P.S., Option volume and stock prices: evidence on where informed traders trade. Journal of Finance 53, Engelberg, J., Sassville, C., Williams, J., Market Madness? The Case of Mad Money. Working Paper, University of North Carolina, Chapel Hill. Finucane, T.J., A new measure of the direction and timing of information flow between markets. Journal of Financial Markets 2, Green, T.C., The value of client access to analyst recommendations. Journal of Financial and Quantitative Analysis 41, Hansen, R.S., Zhou, X., Are Analysts Recommendations Always Informative? Tulane University, unpublished manuscript. Harris, L., Gurel, E., Price and volume effects associated with changes in the S&P500 list: new evidence for the existence of price pressures. Journal of Finance 41, Ho, T., Stapleton, R., Subrahmanyam, M., A simple technique for the valuation and hedging of American options. Journal of Derivatives 1, Hotchkiss, E., Strickland, D., Does shareholder composition matter? Evidence from the market reaction to corporate earnings announcements. Journal of Finance 58, Kumar, R., Sarun, A., Shastri, K., The behavior of option price around large block transactions in the underlying securities. Journal of Finance 47, Lamont, O.A., Thaler, R.H., Can the market add and subtract? Mispricing in tech stock carve-outs. Journal of Political Economy 111, Liang, B., Price pressure: evidence from the Dartboard column. Journal of Business 72, Lynch, A.W., Mendenhall, R.R., New evidence on stock price effects associated with changes in the S&P 500 index. Journal of Business 70, Madhavan, A., Richardson, M., Roomans, M., Why do security price change? A transaction-level analysis of NYSE stocks. Review of Financial Studies 10, Manaster, S., Rendleman, R., Option prices as predictors of equilibrium stock prices. Journal of Finance 4, Mayhew, S., Sarin, A., Shastri, K., The allocation of informed trading across related markets: an analysis of the impact of changes in equity-option margin requirements. Journal of Finance 55, McIntyre, M.L., Jackson, D., Market moves and the information content of option prices. International Review of Economics and Finance 18, Michaely, R., Womack, K.L., Conflict of interest and the credibility of underwriter analyst recommendations. The Review of Financial Studies 12, Mikhail, M.B., Walther, B.R., Willis, R.H., When security analysts talk, who listens? Accounting Review 82, Mikkelson, W.H., Partch, M.M., Withdrawn security offerings. Journal of Financial and Quantitative Analysis 23, Mitchell, M., Pulvino, T., Stafford, E., Limited arbitrage in equity markets. Journal of Finance 57, Mokoaleli-Mokoteli, T., Taffler, R.J., Agarwal, V., Behavioural bias and conflicts of interest in analyst stock recommendation. Journal of Business Finance and Accounting 36, O Connor, M.L., The cross-sectional relationship between trading costs and lead/lag effects in stock & option markets. Financial Review 34, Ofek, E., Richardson, M., Whitelaw, R., Limited arbitrage and short sales restrictions: evidence from the options markets. Journal of Financial Economics 74, Pan, J., Poteshman, A.M., The information in option volume for stock prices. Review of Financial Studies 19, Patell, J., Corporate forecasts of earnings per share and stock price behavior: empirical tests. Journal of Accounting Research 14, Poteshman, A., Underreaction, overreaction, and increasing misreaction to information in the options market. Journal of Finance 56, Scholes, M., The market for corporate securities: substitution versus price pressure and the effects of information on share price. Journal of Business 45,

20 2020 C.R. Chen et al. / Journal of Banking & Finance 35 (2011) Stein, J., Overreaction in the options market. Journal of Finance 44, Stephan, J., Whaley, R., Intraday price change and trading volume relations in the stock and stock options markets. Journal of Finance 45, Stickel, S.E., The anatomy of the performance of buy and sell recommendations. Financial Analysts Journal 51, Taylor, S.J., Yadav, P.K., Zhang, Y., The information content of implied volatilities and model-free volatility expectations: evidence from options written on individual stocks. Journal of Banking and Finance 34, Wang, A., Informed arbitrage with speculative noise trading. Journal of Banking and Finance 34, Womack, K., Do brokerage analysts recommendations have investment value? Journal of Finance 51, Wurgler, J., Zhuravskaya, E., Does arbitrage flatten demand curves for stocks? Journal of Business 75, Xing, Y., Zhang, X., Zhao, R., What does the individual option volatility smirk tell us about future equity returns? Journal of Financial and Quantitative Analysis 45, Yu, W.W., Lui, E.C.K., Wang, J.W., The predictive power of the implied volatility of options traded OTC and on exchanges. Journal of Banking and Finance 34, 1 11.