Seasonality of Optimism in Options Markets Kelley Bergsma, Andy Fodor, and Danling Jiang June 2016 Abstract We study how seasonality in option implied volatilities and returns is related to predictable investor mood fluctuations and retail versus sophisticated investor trading. Using Gallup survey data, we show that mood is predictably more favorable before holidays and on Fridays, while mood is predictably less favorable on Mondays. We find that this seasonality in optimism and pessimism is associated with significant changes in implied volatility spreads and subsequent returns between the out-of-the-money (OTM) and at-the-money (ATM) options. This seasonality in options is primarily driven by higher retail vs. sophisticated investor demand for OTM options. Specifically, optimistic, retail investors heightened demand for OTM calls increases OTM-ATM implied volatility spreads, while pessimistic, retail investors reduced demand for OTM calls decreases spreads. Collectively, our findings suggest retail investor mood plays a dominant role in generating seasonality in call options, while for put options, both retail demand and hedging are important factors for seasonality. Kelley Bergsma is an Assistant Professor in Finance at the College of Business, Ohio University, Athens, OH. Andy Fodor is the Leona Hughes Associate Professor in Finance at the College of Business, Ohio University, Athens, OH. Danling Jiang is the SunTrust Professor and Associate Professor of Finance at the College of Business, Florida State University, Tallahassee, FL.
Seasonality of Optimism in Options Markets June 2016 Abstract We study how seasonality in option implied volatilities and returns is related to predictable investor mood fluctuations and retail versus sophisticated investor trading. Using Gallup survey data, we show that mood is predictably more favorable before holidays and on Fridays, while mood is predictably less favorable on Mondays. We find that this seasonality in optimism and pessimism is associated with significant changes in implied volatility spreads and subsequent returns between the out-of-the-money (OTM) and at-the-money (ATM) options. This seasonality in options is primarily driven by higher retail vs. sophisticated investor demand for OTM options. Specifically, optimistic, retail investors heightened demand for OTM calls increases OTM-ATM implied volatility spreads, while pessimistic, retail investors reduced demand for OTM calls decreases spreads. Collectively, our findings suggest retail investor mood plays a dominant role in generating seasonality in call options, while for put options, both retail demand and hedging are important factors for seasonality. 2
1. Introduction Relative to rational expectations, options traders particularly retail traders are not immune to suboptimal behavior. These investors misreact to changes in implied volatility and instantaneous variance (Stein 1989; Poteshman 2001), exercise options too early (Poteshman and Serbin 2003), buy options on growth stocks that subsequently underperform (Lakonishok et al 2007; Mahani and Poteshman 2008), and are unduly influenced by sentiment without learning from their past mistakes (Han 2008; Chang, Hsieh, and Wang 2015). In this paper, we provide the first evidence that retail investor trading associated with seasonality in mood affects option pricing. We find retail demand for out-of-the-money (OTM) calls connected to mood seasonality is significantly related to seasonality in call OTM-ATM implied volatility spreads and subsequent return corrections of these options. Prior research has shown investors are more optimistic before holidays (Autore, Bergsma, and Jiang 2016) and are in a predictably better (worse) mood on Friday (Monday) (Rystorm and Benson 1989). Although it is well documented that these expected mood swings impact stockholders, our study is the first to explore how unsophisticated vs. sophisticated options traders behave during predictable changes in mood throughout the year. Our study is related to earlier work by Fabozzi, Ma, and Briley (1994) who find significantly higher returns in futures prior to holidays and attribute their findings to a positive preholiday investor mood. Yet, they do not examine options trading or the role of retail investors. Dickinson and Peterson (1989) and Jones and Shemesh (2012) document lower call and put option returns from Friday close to Monday close but have not linked the seasonality in options return to retail investor trading. Doran, Jiang, and Peterson (2012) show that retail investors demand significantly more OTM call options for lottery-type stocks during January; however, they have not examined seasonality across days of the week or holidays. 3
Our paper fills this gap by examining how retail vs. sophisticated investor demand impacts option pricing (e.g., implied volatilities) and returns prior to holidays and on Mondays and Fridays. Using Gallup survey data, we first confirm the predictable variation in mood preceding holidays and by days of the week. We provide evidence that these anticipated mood changes impact OTM-ATM implied volatility spreads, which are driven in part by higher retail open buy demand relative to sophisticated trader open buy demand for OTM options. In our tests, we focus on OTM options on individual stocks, based on the notion that long positions in OTM calls represents strong optimism whereas long positions in OTM puts are indicative of strong pessimism. Consistent with increased optimism prior to holidays and on Fridays, we find that call OTM-ATM implied volatility spreads widen on these days. The increased spread is related to higher unsophisticated relative to sophisticated open buy demand for OTM calls on those days. In contrast, on Mondays when investors are pessimistic, call OTM-ATM implied volatility spreads shrink and put OTM-ATM implied volatility spreads widen. We find that higher retail vs. sophisticated investor open buy demand for OTM puts significantly increases put OTM-ATM spreads on Mondays. Our additional tests verify the robustness of our results. The predictable patterns in option trading following mood changes are unaffected by restricting the sample to only options within 30 trading days of an anticipated earnings announcements, suggesting that the mood effects are not subsumed by changes in implied volatility prior to earnings announcements. We observe some seasonality in S&P 500 Index options, although the results for the index options are somewhat weaker than those for individual stock options. By demonstrating that mood plays an important role in option pricing (Kliger and Levy 2008), our findings provide further empirical evidence that behavioral theory is applicable to option markets (Gurevich, Kliger, and Levy 2009; Delisle, Diavatopoulos, Fodor, and Krieger 2015). 4
The remainder of the paper is organized as follows. Section 2 provides the motivation and hypotheses. Section 3 describes the data and methodology. Section 4 discusses the empirical results. Section 5 concludes. 2. Motivation and Hypotheses Compared with stock trading data, option trading data provides a richer set of information on investors level of bullish or bearish sentiment. Investors purchases of stock would be construed as a bullish signal, while sales of equity would be a bearish signal. Assuming no leverage, however, it is difficult to clearly identify trader sentiment from equity trading data. In contrast, investors increased purchases of OTM calls (vs. ATM calls) demonstrates a stronger level of optimism, whereas traders buying more OTM puts (vs. ATM puts) indicates a greater degree of pessimism. OTM and ATM options trading provides a new setting to gauge investors levels of optimism and pessimism. In the context of our study, trading in OTM and ATM options provides a novel test as to whether seasonality in mood is reflected in heightened bullish and/or bearish sentiment. We study predictable changes in mood throughout the year based on growing literature on the day of the week effect (French 1980; Keim and Stambaugh 1984; Jones and Shemesh 2012) and preholiday effect (Lakonishok and Smidt 1988; Pettengill 1988; Ariel 1990; Fabozzi, Ma, and Briley 1994; Frieder and Subrahmanyam 2004). Prior work suggests that mood follows predictable patterns based on the day of the week and proximity to holidays (Rystrom and Benson 1989; Autore, Bergsma, and Jiang 2016). We use the Gallup Daily U.S. Mood survey data to verify these mood patterns. The Gallup Daily U.S. Mood survey data tracks daily measures of Americans mood using telephone interviews of approximately 500 randomly sampled adults, and is available from January 2008 to November 2012. Following Autore, Bergsma, and Jiang (2016), we use the mood variable, 5
NetHappy, which captures the net percentage of individuals who are happy as opposed to worried on a given day. 1 Figure 1 presents consistent patterns in the average NetHappy values throughout time. Specifically, Figure 1A plots the average NetHappy values for Preholiday and Ordinary day. Preholiday is a preholiday trading day, defined as within two trading days prior to a major U.S. holiday and includes the holiday itself if the market is open that day (Autore, Bergsma, and Jiang 2016). 2 Ordinary day is comprised of all other days that are not defined as Preholiday. Average mood is higher for Preholiday as compared with Ordinary day, corroborating the findings of Autore, Bergsma, and Jiang. 3 Figure 1B plots the average NetHappy values by day of the week. Consistent with Rystorm and Benson s (1989) hypothesis of a Blue Monday effect, NetHappy is the lowest on Monday and the highest on Friday. In untabulated results, we confirm these graphical observations regarding mood changes are also statistically significant mood is significantly higher on Fridays and on preholidays, while mood is significantly lower on Mondays. [Insert Figure 1 here] Having established that seasonality in optimism varies by day, our first goal is to investigate whether these mood fluctuations impact option pricing. Since less sophisticated investors use OTM call options to bet on positive underlying asset returns (Shefrin and Statman 2000; Statman 2004), we would expect more optimistic investors would exhibit a higher demand for OTM relative to ATM call options. Buying OTM options represents a stronger bullish view than buying ATM options as OTM options are more lottery-like (Kumar 2009). The same should be true for put options when investors are pessimistic, predicting greater demand for long positions in OTM puts. 1 For further information on this dataset, refer to Autore, Bergsma, and Jiang (2015). 2 We study 13 major U.S. holidays that have been celebrated in the United States for at least 100 years: New Year s Day, Valentine s Day, Presidents Day, St. Patrick s Day, Easter, Mother s Day, Memorial Day, Father s Day, Independence Day (Fourth of July), Labor Day, Halloween, Thanksgiving, and Christmas. 3 Autore, Bergsma, and Jiang (2016) use change in NetHappy relative to the same weekday of the prior week in order to control for the day of the week effect, while in this paper raw measures are used as our aim is to study the day of the week effect. 6
These demand pressures will influence OTM-ATM implied volatility spreads. Thus, our first hypothesis is as follows. Hypothesis 1: The call OTM-ATM implied volatility spread will be greater when individuals are in a predictably favorable mood and will be reduced when mood is unfavorable. The put OTM-ATM implied volatility spread will widen when individuals are in a predictably pessimistic mood and will shrink when mood is optimistic. Moreover, since retail investors are more subject to the influence of mood (Bergsma and Jiang 2016), we expect retail investor demand to drive predicted changes in implied volatility spreads associated with mood fluctuations. We calculate Net Long, a measure described in Section 3, which captures the difference in net long positions between retail trades and sophisticated trades in OTM options. We hypothesize Net Long interacted with our mood-related seasonality variables will significantly explain the increases and decreases in implied volatility spreads related to predictable mood changes. This motivates our second hypothesis. Hypothesis 2: When mood is optimistic, retail traders will demand more OTM call options and fewer OTM put options relative to sophisticated traders, contributing to increased call OTM-ATM implied volatility spreads and decreased put OTM-ATM implied volatility spreads. When mood is unfavorable, retail investors will demand fewer OTM call options and more OTM put options relative to sophisticated investors, contributing to decreased call OTM-ATM implied volatility spreads and increased put OTM-ATM implied volatility spreads. Last, we expect implied volatility changes induced by retail investor mood swings will be subsequently corrected in future option returns. Options that are overpriced due to excess optimism (or pessimism) will earn lower subsequent returns as the market corrects mispricing. Conversely, options that are underpriced due to excess pessimism (or optimism) will earn higher future returns as mispricing is corrected. These ideas are described in our third hypothesis. Hypothesis 3: When investors are predictably optimistic, OTM calls will become overpriced (vs. ATM calls), leading to subsequent abnormally low call OTM-ATM returns, whereas OTM puts will become underpriced (vs. ATM puts), leading to subsequent abnormally high put OTM-ATM returns. When investors are predictably pessimistic, OTM 7
calls will become underpriced (vs. ATM calls), resulting in subsequent abnormally high call OTM-ATM returns, while OTM puts will become overpriced (vs. ATM puts), leading to subsequent abnormally low put OTM-ATM returns. 3. Data and Methodology To test our three main hypotheses, we use individual stock options from the OptionsMetrics dataset from January 1996 to August 2014. We define OTM calls as those with the ratio of the strike price to the stock price greater than 1.05 and OTM puts as those with the ratio of the strike price to the stock price less than 0.95. ATM options have the ratio of the strike price to the stock price between 0.95 and 1.05 inclusive. Implied volatility is measured at the close of the day for options with time to expiration between 11 to 92 days. 4 Stocks without trading volume for any ATM and OTM options are excluded. We compute average open interest weighted implied volatility separately for OTM and ATM calls for each stock and then calculate OTM-ATM spread. Table 1 presents summary statistics for the full sample and then separately for preholidays, Mondays, and Fridays. Ret is the underlying stock s return of the prior day. MktCap, is the market capitalization at the end of the prior month. Book-to-market (B/M) is defined as in Fama and French (1992). Momentum is the cumulative monthly stock return from month t 12 to t 2. Turnover is average monthly turnover from month t 13 to t 1. Idiosyncratic volatility (Ivol) is the standard deviation of the residuals of the previous month s daily returns (at least 17 days) regressed on the Fama-French three factors (Fama and French 1993; Ang, Hodrick, Xing, and Zhang 2006). [Insert Table 1 here] We use International Securities Exchange (ISE) Open/Close Trade profile dataset, available from May 2005 to August 2014, to classify each option s trades as retail investor trades or sophisticated trades. We identify retail trades as those of at most 200 contracts, while trades above 4 Options without 1-10 days to expiration are omitted due to expiration day concerns. 8
200 contracts are classified as sophisticated trades. In the spirit of Doran, Fodor, and Jiang (2013), we calculate a Net Long measure to capture the difference in net long positions between retail traders and sophisticated traders in OTM options: Net Long = (Retail Open Buy Retail Close Sell) (Sophisticated Open Buy Sophisticated Close Sell) (1) Net Long is in units of number of contracts. We define CNL as Call Net Long for OTM calls and PNL is Put Net Long for OTM puts. Option returns are calculated as Option Return t = (Price t Price t-1)/price t-1 where Price t is the price at the close of the current day and Price t-1 is the price at the close of prior trading day. All prices are the average of the closing bid and ask prices. In unreported tests, we choose one OTM call, one ATM call, one OTM put, and one ATM put from our sample for each firm day that is the nearest to expiration. If more than one option in each category has the nearest expiration, we select ATM options with K/S closest to 1, the OTM call option closest to 1.05, and the OTM put option closest to 0.95. Results are similar using this alternative methodology. 4. Empirical Results 4.1. The Impact of Seasonality in Optimism on Implied Volatility Spreads Our first hypothesis predicts that call OTM-ATM implied volatility spreads will widen when investors are optimistic (before holidays and on Fridays) and shrink when investors are pessimistic (on Mondays). In addition, put OTM-ATM implied volatility spreads will widen when traders are in an unfavorable mood and shrink when traders are in a favorable mood. We test our first hypothesis in Table 2. Using OLS regressions with firm and year fixed effects, we regress each stock s average daily OTM-ATM implied volatility spread on the following seasonality 9
dummy variables: Preholiday Monday, and Friday. 5 In each regression, we include the control variables of the prior day s stock return, market capitalization, B/M, momentum, stock turnover, stock idiosyncratic volatility, and Expiration Friday. 6 Expiration Friday is a dummy variable for the third Friday of each month (or Thursday in the case of a Friday holiday), a day when options typically expire and unusually high volume occurs (Stoll and Whaley 1986, 1987). The dependent variable for each regression is either Call OTM-ATM IV or Put OTM-ATM IV, where IV refers to implied volatility spread. The specification is: Call or Put OTM-ATM IV = β 1 Seasonality Dummy + β 2 Ret + β 3 Log(MktCap) + β 4 B/M + β 5 Momentum + β 6 Turnover + β 7 Ivol + ε (2) [Insert Table 2 here] In Table 2, we examine OTM-ATM implied volatility spreads on preholidays, Mondays, and Fridays. On preholidays, regressions (1) and (4) show that the OTM-ATM implied volatility spread for both calls and puts significantly widens. The positive coefficient of 0.27% (t = 29.98) on Preholiday for calls is consistent with Figure 1A showing that investors exhibit a more positive preholiday mood, leading to stronger bullish sentiment (Hypothesis 1). However, the coefficient of 0.14% (t = 11.82) on Preholiday for puts is consistent with investors attempts to hedge downside risk prior to holiday market closures. 7 For Mondays, regressions (2) and (5) show that the OTM-ATM implied volatility spread for calls significantly shrinks, by 0.03% (t = 4.85), and spread for puts widens, by 0.12% (t = 13.74). This evidence supports the assertion that pessimistic mood leads to a more bearish outlook. In contrast, we find positive and significant estimates for Friday coefficients 5 Results are similar using Fama-MacBeth quarterly regressions. 6 Results are similar using contemporaneous daily stock return. 7 Rather than shorting the stock over a trading break, investors may choose to buy an OTM put, as it is cheaper than an ATM put and both gain value as stock prices fall. 10
for both calls and puts in regressions (3) and (6), indicating OTM-ATM implied volatility spreads for calls and puts widen by 0.24% (t = 30.72) and 0.14% (t = 14.02) respectively on Fridays. These results are similar to our findings for Fridays, as in both cases investors are in a more positive mood and market closures occur (Kaplanski and Levy 2015). An important caveat is that if underlying stock volatility is greater during a trading break (weekend or holiday market closure), our findings for Fridays and some preholidays could be explained by market closures as opposed to investor mood changes. Yet, Jones and Shemesh (2012) point out that weekend volatility is no different than weekday volatility, confirming an earlier observation of French and Roll (1986). Thus, the increases in call OTM-ATM implied volatility spreads on Fridays and on preholidays cannot be explained by underlying volatility differences. Rather, evidence that call OTM-ATM implied volatility spreads widen can be explained by investors exhibiting seasonality in optimism which influences option demand and pricing. Nevertheless, since we include holidays associated and not associated with holiday market closures, we examine the two types of holidays separately in the Appendix and find that higher put OTM-ATM implied volatility spreads occur only in the case of holidays with non-weekend trading breaks, consistent with a hedging story. However, the wider call OTM-ATM implied volatility spread in both cases (with or without market closures) support a preholiday investor optimism explanation. In addition, an examination of the impact of the turn of the year on OTM-ATM implied volatility spreads is provided in the Appendix, as New Year s Day represents a distinct holiday that has been investigated separately (Dickinson and Peterson 1989; Doran, Jiang, and Peterson 2012). We define PreNY and PostNY as five trading days before and after New Year s Day, following Dickinson and Peterson. Consistent with an optimistic New Year holiday mood, we find that call OTM-ATM implied volatility spreads widen around the New Year. Moreover, put OTM-ATM implied volatility spreads also widen surrounding the turn of the year, consistent with Dickinson and 11
Peterson s finding that OTM puts may be overvalued in January. However, as the turn of the year effect in the U.S. market is intertwined with tax-loss selling, these results should be interpreted with caution. Prior literature has shown that implied volatilities increase before earnings announcements (Patell and Wolfson 1979, 1981). Given this fact, we separate the sample into two groups options of firms within and not within 30 trading days of a Compustat quarterly earnings announcement in order to investigate whether options on stocks with upcoming earnings announcements exhibit similar seasonality in implied volatility spreads. Results in Table 3 demonstrate seasonality similar to full sample findings. The one exception to our main findings is for options not within 30 trading days of an earnings announcement, put OTM-ATM implied volatility spreads do not significantly widen on preholidays, suggesting hedging pressure is less for this category of options. Overall, our findings on options seasonality cannot be attributed to implied volatility changes associated with earnings announcements. [Insert Table 3 here] 4.2. The Role of Retail vs. Sophisticated Open Buy Demand in Seasonality in Optimism and Implied Volatility Spreads Next we examine the role of retail vs. sophisticated open buy demand on changes in implied volatility spreads associated with the seasonality in mood. As described in our second hypothesis, since retail investors are more subject to the influence of mood swings than are sophisticated investors, we expect that when optimistic, retail traders will open more long positions in OTM calls and fewer in OTM puts relative to sophisticated traders. Conversely, we hypothesize that pessimistic retail traders will open fewer long positions in OTM calls and more positions in OTM puts relative 12
to sophisticated traders. We are most interested in open buy demand for OTM calls as it represents the strongest bullish outlook and OTM puts as it signals the strongest bearish view. In Table 4, we explore whether retail vs. sophisticated open buy demand drives seasonality in implied volatilities reported in Table 2. As described in Section 3, we calculate two measures, CNL (Call Net Long) and PNL (Put Net Long), which express the difference in net long positions between retail and sophisticated investors for call and put options. We interact CNL and PNL with our seasonality dummy variables (Preholiday, Monday, and Friday) and repeat the regressions in Table 2. [Insert Table 4 here] In Table 4, our results demonstrate greater retail investor buying activity on preholidays than other days. Specifically, in model (1), when retail traders open more long positions in OTM calls relative to sophisticated investors, the call OTM-ATM implied volatility spread widens by an additional 0.54% (t = 2.89) after controlling for Net Long demand for OTM calls (CNL) and the preholiday effect. While the Preholiday CNL coefficient is both statistically and economically significant, Preholiday PNL is insignificant in model (4), suggesting less sophisticated investors increased closing (relative to sophisticated investors) of long OTM put positions has no significant impact on the put OTM-ATM implied volatility spread. Thus, the evidence suggests the increased call OTM-ATM implied volatility spread on preholidays is related to greater retail vs. sophisticated buy demand on preholidays when investors are in favorable preholiday mood. In contrast, we find that the Monday CNL coefficient is insignificant, but the Monday PNL coefficient is positive and significant (0.51%, t = 3.54). Therefore, after controlling for Net Long demand in OTM puts (PNL) and the Monday effect, our results indicate retail traders increased opening of long positions in OTM put options relative to sophisticated investors 13
significantly increases the put OTM-ATM implied volatility spread by an additional 0.51%. 8 This is consistent with the more bearish outlook of retail investors on Mondays influencing option pricing. Furthermore, on Fridays, less sophisticated investors increased opening of long positions (relative to sophisticated investors) in OTM put options widens the call OTM-ATM implied volatility spread by an additional 0.31% (t = 2.27) on average. The coefficient of Friday PNL is insignificant. Our findings support the idea of greater optimism on Fridays influencing call OTM- ATM implied volatility spreads through the channel of retail vs. sophisticated investor demand. In sum, we find seasonality in OTM-ATM implied volatility spreads is significantly related to differences in retail vs. sophisticated traders open buy demand on preholidays, Mondays, and Fridays. Our results provide strong support for Hypothesis 2, particularly regarding call options. 4.4. S&P 500 Options It is a natural extension to study whether the seasonality evident among individual stock options are also manifest in S&P 500 index options (SPX). To analyze these options, we use ordinary least squares regressions and cluster standard errors by quarter (Table 5). We regress each day s average OTM-ATM implied volatility spread on the following predictable mood pattern dummy variables: Monday, Friday, and Preholiday. Control variables include Expiration Friday, as defined in Table 2, and S&PRet, the S&P 500 return of the prior day. As with prior tests, the dependent variable is either Call OTM-ATM IV or Put OTM-ATM IV, where IV refers to implied volatility spread. Consistent with investor optimism, the Friday coefficient is positive for call options, whereas the Monday coefficient is positive for put options (indicative of greater pessimism) but is not significant for call options. Consistent with prior results, the Friday coefficient is also positive for puts, supporting 8 The mean CNL is 15 contracts and the mean PNL is 18 contracts. Thus, relative to the unconditional mean of 18 OTM put contracts, a one contract increase of retail investor buy demand for OTM puts increases the Monday put OTM-ATM implied volatility spread by an additional 0.51% on average. 14
a hedging explanation. The Preholiday coefficient is insignificant for both puts and calls. It is not surprising that seasonality is stronger in individual stock options than in S&P 500 options. For example, Jones and Shemesh (2012) find the effect of weekends and mid-week holidays on option returns is much weaker among S&P 500 options than among portfolios of individual stock options. Nevertheless, S&P 500 OTM-ATM implied volatility spreads exhibit significant seasonality consistent with increased optimism on Fridays and greater pessimism on Mondays. [Insert Table 5 here] 4.5. Option Returns We last examine whether seasonality in OTM-ATM implied volatility spreads associated with periodic mood changes leads to predictable subsequent option returns, as mood fluctuations recede and options mispricing is corrected. We calculate the difference between each stock s OTM and ATM option returns for calls and puts on the trading days immediately following preholidays, Mondays, and Fridays, labeled as Call OTM-ATM Return and Put OTM-ATM Return respectively. Based on our third hypothesis, when investors are predictably optimistic, we predict greater demand for OTM calls and lesser demand for OTM puts will lead to overpricing of OTM calls and underpricing of OTM puts. We expect this initial, mood-induced mispricing to be subsequently corrected, implying lower call OTM-ATM return spreads and higher put OTM-ATM return spreads in the near future. In contrast, when traders are in an unfavorable mood, we predict that reduced demand for OTM calls and increased demand for OTM puts will lead to underpricing of OTM calls and overpricing of OTM puts. The outcome of this scenario is higher future call OTM-ATM return spreads and lower put OTM-ATM return spreads than under normal circumstances. Prior literature suggests the impact of mood on security prices is transitory and will be corrected in the following days. For example, mispricing induced by negative sentiment associated with aviation disasters is 15
corrected within two days (Kaplanski and Levy 2010). In Table 6, we present results from regressing subsequent one- and two-day option returns on and Preholiday, Monday, and Friday [Insert Table 6] We find evidence of reversal in call returns following preholidays that continues for the next two days. The call OTM-ATM return spread is 0.73% on the subsequent day (t = 9.57) and 1.28% (t = 11.15) for the following two days. Our results for Mondays suggest reversal takes two days to be detectable. Although the one-day return spread is negative and significant, two-day call OTM-ATM returns are 0.49% (t = 5.48) after Mondays, suggesting a reversal to unusually low call OTM-ATM implied volatility spreads at the Monday close. The reversal in call OTM-ATM return spreads following Fridays happens more swiftly one- and two-day call OTM-ATM return spreads are 1.72% and 2.05% lower following Fridays and are both highly significant. These results imply market correction of abnormally high call OTM-ATM implied volatility spreads on Fridays occurs within two days (by Tuesday close). Overall, we find unusually high (low) call OTM-ATM implied volatility spreads are followed by significantly lower (higher) future OTM-ATM return spreads. For put options, the OTM-ATM return spread following Mondays is 0.13% (t = 2.39) for the one-day period, but insignificant for the two-day period. We document strong reversal of elevated OTM-ATM implied volatility spreads for preholidays and Fridays. As previously stated, the increased put OTM-ATM implied volatility spread on preholidays and Fridays is likely due to hedging pressure. Our main results in Table 2 show hedging pressure subsumes the mood effect for Fridays and preholidays. Pronounced reversal of widened put OTM-ATM implied volatility spreads results in the significantly lower subsequent put OTM-ATM returns. 9 9 Our findings of significant negative return spreads for call OTM-ATM and put return spreads from Friday close to Monday close are in agreement with Jones and Shemesh (2012) with two important distinctions. Jones and Shemesh (2012) neither examine the differences between OTM and ATM return spreads nor explore the role of investor mood in option pricing. 16
In short, we find support for Hypothesis 3 among call OTM-ATM return spreads, but weaker evidence for put return spreads. Our results demonstrate that widened call OTM-ATM implied volatility spreads on Fridays and preholidays are followed by negative OTM-ATM return spreads. This evidence supports the assertion that optimism-induced demand for OTM call options leads to overpricing relative to ATM calls and lower future OTM-ATM return spreads. In untabulated results, evidence suggest that higher OTM-ATM implied volatility spreads are directly related to options lower future returns. Results available upon request. 5. Conclusion While prior research provides evidence of retail investors suboptimal behavior in option markets, we are the first to document the impact of mood seasonality in retail option investor demand as well as associated implied volatilities and returns. Using Gallup survey data, we establish individuals seasonality in optimism and pessimism prior to holidays and on Mondays and Fridays. We find that call OTM-ATM implied volatility spreads widen when individuals are optimistic and shrink in times of pessimism. These changes in implied volatility spreads are driven by retail investor trading as less sophisticated investors demand more OTM calls when mood is favorable relative to when mood is unfavorable. Our results also provide similar evidence, albeit weaker, for puts as retail demand for OTM puts reflects pessimism, yet put trading is also influenced by hedging demand. Moreover, we find widened call OTM-ATM implied volatility spreads are immediately followed by subsequent lower abnormal call OTM-ATM returns, suggesting a correction of optimism-induced overpricing of OTM calls. As these mood patterns are predictable across the calendar year, our findings imply trading opportunities in options markets are generated by retail investors seasonality in mood. 17
References Ariel, R.A. 1990. High stock returns before holidays: Existence and evidence on possible causes. Journal of Finance 45, 1611-1626. Ang, A., R. Hodrick, Y. Xing, and X. Zhang. 2006. The cross-section of volatility and expected returns. Journal of Finance 61, 259-299. Autore, D., K. Bergsma, and D. Jiang. 2016. The preholiday corporate announcement effect. Working paper. Bergsma, K. and D. Jiang. 2016. Cultural New Year holidays and stock returns around the world. Financial Management 45, 3-35. Chang, C-C., Hsieh, P-F., and Y-H Wang. 2015. Sophistication, sentiment, and misreaction. Journal of Financial and Quantitative Analysis 50, 903-928. Delisle, J., D. Diavatopoulos, A. Fodor, and K. Krieger. 2015. Prospect theory, mental accounting, and option prices. Working paper. Dickinson, A. and D. Peterson. 1989. Seasonality in the option market. The Financial Review 24, 529-540. Doran, J., D. Jiang, and D. Peterson. 2012. Gambling preference and the New Year effect of assets with lottery features. Review of Finance 16, 685-731. Doran, J.S., A. Fodor, and D. Jiang. 2013. Call-put implied volatility spreads and option returns. Review of Asset Pricing Studies 3, 258-290. Fabozzi, F., C. Ma, and J. Briley. 1994. Holiday trading in futures markets. Journal of Finance 49, 307-324. Fama, E.F. and K.R. French. 1992. The cross-section of expected stock returns. Journal of Finance 47, 427-465. French, K. 1980. Stock returns and the weekend effect. Journal of Financial Economics 8, 55 69. French, K., and R. Roll. 1986. Stock return variances: The arrival of information and the reaction of traders. Journal of Financial Economics 17, 5-26. Frieder, L. and A. Subrahmanyam. 2004. Nonsecular regularities in returns and volume. Financial Analysts Journal 60, 29-34. Gurevich, G., D. Kliger, and O. Levy. Decision-making under uncertainty A field study of cumulative prospect theory. Journal of Banking and Finance 2009 33, 1221-1229. Han, B. 2008. Investor sentiment and option prices. Review of Financial Studies 21, 387-414. Jones, C.S. and J. Shemesh. 2012. The weekend effect in equity option returns. Working paper. Kaplanski, G., and H. Levy. 2010. Sentiment and stock prices: The case of aviation disasters. Journal of Financial Economics 95, 174-201. 18
Kaplanski, G. and H. Levy. 2015. Trading breaks and asymmetric information: The option markets. Journal of Banking and Finance 58, 390 404. Keim, D.B. and R. Stambaugh. 1984. A further investigation of the weekend effect in stock returns. Journal of Finance 39, 819-835. Kliger, D. and O. Levy. 2008. Mood impacts on probability weighting functions: Large gamble evidence. Journal of Socio-Economics 37, 1397-1411. Lakonishok, J., and S. Smidt. 1988. Are seasonal anomalies real? A ninety-year perspective. Review of Financial Studies 1, 403-425. Lakonishok, J., I. Lee, N.D. Pearson, and A.M. Poteshman. 2007. Option market activity. Review of Financial Studies 20, 813-857 Mahani, R.S. and A.M. Poteshman. 2008. Overreaction to stock market news and misevaluation of stock prices by unsophisticated investors: Evidence from the option market. Journal of Empirical Finance 15, 635-655. Patell, J., and M. Wolfson. 1979. Anticipated information releases reflected in call option prices. Journal of Accounting and Economics 1, 117 140. Patell, J., and M. Wolfson. 1981. The ex-ante and ex-post price effects of quarterly earnings announcements reflected in option and stock prices. Journal of Accounting Research 19, 434 458. Pettengill, G.N. 1989. Holiday closings and security returns. Journal of Financial Research 12, 57-67. Poteshman, A.M. 2001. Underreaction, overreaction, and increasing misreaction to information in the options market. Journal of Finance 56, 851-876. Poteshman, A.M. and V. Serbin. 2003. Clearly irrational financial market behavior: Evidence from the early exercise of exchange traded stock options. Journal of Finance 58, 37-70. Rystrom, D.S. and E.D. Benson. 1989. Investor psychology and the day-of-the-week effect. Financial Analysts Journal 45, 75-78. Shefrin, H., and M. Statman. 2000. Behavioral portfolio theory. Journal of Financial and Quantitative Analysis 35, 127-151. Statman, M. 2004. The diversification puzzle. Financial Analysts Journal 60, 44-53. Stein, Jeremy. 1989. Overreactions in the options market. Journal of Finance 44, 1011-1023. Stoll, H., and R. Whaley. 1986. Expiration day effects of index futures and options: Empirical tests. Review of Research in Futures Markets 5, 292-304. Stoll, H., and R. Whaley. 1987. Program trading and expiration day effects. Financial Analysts Journal 43, 16-28. 19
Figure 1: Mood Patterns This figure presents the average values of individuals reported levels of happiness minus levels of worry (NetHappy) according to the Gallup Daily U.S. Mood survey responses. Figure 1A presents the average values for Preholiday vs. Ordinary. Preholiday is a preholiday trading day, defined as within two trading days prior to a major U.S. holiday and include the holiday itself if the market is open that day (Autore, Bergsma, and Jiang 2016). Ordinary day consists of all non-preholiday trading days. Figure 1B presents the averages by day of the week. The sample period is from January 2008 to November 2012. 44 Figure 1A: Preholiday vs. Ordinary day 42 40 38 36 34 Preholiday Ordinary day Figure 1B: Day of the Week 48 46 44 42 40 38 36 34 32 30 Monday Tuesday Wednesday Thursday Friday 20
Table 1: Summary Statistics This table reports the summary statistics for all options in our sample and separate summary statistics on preholidays, Mondays, and Fridays. Preholiday is a preholiday trading day, defined as within two trading days prior to a major U.S. holiday and include the holiday itself if the market is open that day (Autore, Bergsma, and Jiang 2016). Call OTM-ATM IV refers to the out-of-the-money minus at-the-money implied volatility spread for call options. Put OTM-ATM IV refers to the out-of-the-money minus at-the-money implied volatility spread for put options. Implied volatility is reported in percent (annualized). Ret is the underlying stock s daily return of the prior day. Mktcap is market capitalization in billions. Book-to-market (B/M) is defined as in Fama and French (1992). Momentum is the cumulative monthly stock return from month t 12 to t 2 in percent. Turnover is average monthly stock turnover from month t 13 to t 1 in percent. The sample period is from January 1996 to August 2014. N Call OTM-ATM IV Put OTM-ATM IV Ret Log(MktCap) B/M Momentum Turnover Ivol All 5,802,273-0.77% 5.91% 1.49% 2.20 0.44 24.65% 263.69 1.83% Preholiday 620,007-0.56% 6.06% 1.91% 2.23 0.44 25.04% 264.57 1.86% Monday 1,120,888-0.77% 6.03% 1.39% 2.14 0.44 24.31% 261.44 1.83% Friday 1,155,679-0.69% 5.90% 1.46% 2.21 0.44 24.50% 263.94 1.83%
Table 2: The Impact of Seasonality in Optimism on Implied Volatility Spreads This table tests whether individual stocks OTM-ATM implied volatility spreads differ on preholidays, Mondays, and Fridays. Using OLS regressions with quarter and firm fixed effects, we regress each stock s average OTM-ATM implied volatility spread on the following predictable mood pattern dummy variables: Preholiday, Monday, and Friday. Preholiday is dummy variable equaling one if the date is within two trading days prior to a major U.S. holiday and includes the holiday itself if the market is open that day (Autore, Bergsma, and Jiang 2016). In each regression, we include Ret, Log(MktCap), B/M, Momentum, Turnover, Ivol, and Expiration Friday as controls. Expiration Friday is a dummy variable for the third Friday of each month (or Thursday in the case of a Friday holiday), a day when options typically expire and unusually high option volume occurs. The dependent variable for each regression is labeled as either Call OTM-ATM IV or Put OTM-ATM IV, where IV refers to implied volatility spread in percent (annualized). *, **, and *** indicate 10%, 5%, and 1% level of significance respectively using a two-tailed test. The t-statistics are reported in parentheses. The sample period is from January 1996 to August 2014. Call OTM-ATM IV Put OTM-ATM IV (1) (2) (3) (4) (5) (6) Preholiday 0.27*** 0.14*** (29.98) (11.82) Monday -0.03*** 0.12*** (-4.85) (13.74) Friday 0.24*** 0.14*** (30.72) (14.02) Ret -0.24*** -0.23*** -0.22*** 1.32*** 1.33*** 1.33*** (-7.96) (-7.66) (-7.58) (32.50) (32.70) (32.71) Log(MktCap) -0.15*** -0.14*** -0.14*** -0.06-0.05-0.05 (-3.86) (-3.55) (-3.49) (-1.17) (-1.01) (-1.00) B/M -0.06* -0.06-0.06 0.21*** 0.21*** 0.21*** (-1.69) (-1.64) (-1.61) (4.54) (4.53) (4.57) Momentum 0.09*** 0.09*** 0.09*** -0.13*** -0.13*** -0.13*** (10.53) (10.52) (10.60) (-12.06) (-12.03) (-12.02) Turnover 0.32*** 0.32*** 0.32*** -0.01 0.00 0.00 (21.87) (22.07) (21.98) (-0.26) (-0.18) (-0.23) Ivol 0.18*** 0.19*** 0.19*** 0.28*** 0.28*** 0.28*** (18.22) (18.63) (18.55) (22.08) (22.27) (22.22) Expiration Friday -0.39*** -0.37*** -0.56*** -0.60*** -0.56*** -0.69*** (-29.53) (-27.74) (-37.98) (-35.06) (-32.60) (-36.96) N 2,083,998 2,083,998 2,083,998 1,686,444 1,686,444 1,686,444
Table 3: Implied Volatility Spreads Prior to Earnings Announcements This table tests whether within 30 trading days prior to a quarterly earnings announcement, individual stocks OTM-ATM implied volatility spreads differ on preholidays, Mondays, and Fridays. Panel A displays the results for options within 30 days of an earnings announcement, while Panel B presents the findings for options not within 30 days of an earnings announcement for comparison. Using OLS regressions with quarter and firm fixed effects, we regress each stock s average OTM-ATM implied volatility spread on the following predictable mood pattern dummy variables: Preholiday, Monday, and Friday. Preholiday is dummy variable equaling one if the date is within two trading days prior to a major U.S. holiday and includes the holiday itself if the market is open that day (Autore, Bergsma, and Jiang 2016). Control variables are as in Table 2. The dependent variable for each regression is labeled as either Call OTM-ATM IV or Put OTM-ATM IV, where IV refers to implied volatility spread in percent (annualized). *, **, and *** indicate 10%, 5%, and 1% level of significance respectively using a two-tailed test. The t-statistics are reported in parentheses. The sample period is from January 1996 to August 2014. Panel A: Options within 30 days prior to an earnings announcements Call OTM-ATM IV Put OTM-ATM IV (1) (2) (3) (4) (5) (6) Preholiday 0.35*** 0.28*** (25.03) (15.55) Monday -0.03*** 0.14*** (-3.41) (10.94) Friday 0.22*** 0.11*** (19.54) (7.61) Control Variables as in Table 2 N 1,021,340 1,021,340 1,021,340 824,097 824,097 824,097 Panel B: Options not within 30 days prior to an earnings announcements Call OTM-ATM IV Put OTM-ATM IV (1) (2) (3) (4) (5) (6) Preholiday 0.17*** 0.02 (15.62) (1.62) Monday -0.03*** 0.11*** (-3.12) (9.65) Friday 0.21*** 0.14*** (22.08) (11.58) Control Variables as in Table 2 N 1,062,657 1,062,657 1,062,657 862,347 862,347 862,347 23
Table 4: The Role of Retail vs. Sophisticated Buy Demand in Seasonality in Optimism and Implied Volatility Spreads This table tests whether net long demand among retail vs. sophisticated investors drives the difference in individual stocks OTM-ATM implied volatility spreads associated with predictable mood patterns. Similar to Doran, Fodor, and Jiang (2013), we calculate a Net Long measure which captures the difference in net long positions between less sophisticated trades and sophisticated trades. Net Long is defined as (Retail Open Buy Retail Close Sell) (Sophisticated Open Buy Sophisticated Close Sell). We define CNL as Call Net Long for OTM calls and PNL is Put Net Long for OTM puts. Using OLS regressions with quarter and firm fixed effects, we regress each stock s average OTM-ATM implied volatility spread on the following predictable mood pattern dummy variables: Preholiday, Monday, and Friday. Preholiday is dummy variable equaling one if the date is within two trading days prior to a major U.S. holiday and includes the holiday itself if the market is open that day (Autore, Bergsma, and Jiang 2016). Control variables are as in Table 2. The dependent variable for each regression is labeled as either Call OTM-ATM IV or Put OTM-ATM IV, where IV refers to implied volatility spread in percent (annualized). *, **, and *** indicate 10%, 5%, and 1% level of significance respectively using a two-tailed test. The t-statistics are reported in parentheses. The sample period is from May 2005 to August 2014. Preholiday CNL 0.54*** (2.89) Monday CNL -0.10 (-0.81) Call OTM-ATM IV Put OTM-ATM IV (1) (2) (3) (4) (5) (6) Friday CNL 0.31** (2.27) Preholiday PNL -0.23 (-0.94) Monday PNL 0.51*** (3.54) Friday PNL -0.20 (-1.24) Preholiday 0.30*** (24.28) Monday -0.09*** 0.14*** (-9.36) (9.01) Friday 0.25*** 0.15*** (24.32) (12.05) CNL -0.01 0.05-0.02 0.12*** (-0.26) (0.93) (-0.39) (9.07) PNL -0.19*** -0.37*** -0.16** (-2.77) (-4.60) (-2.15) Control Variables as in Table 2 N 830,894 830,894 830,894 735,940 735,940 735,940 24
Table 5: S&P 500 Index Options Implied Volatility Spreads This table tests whether S&P 500 Index options OTM-ATM implied volatility spreads differ on Mondays, Fridays, and prior to and on holidays. Using OLS regressions with quarter fixed effects, we regress each stock s average OTM-ATM implied volatility spread on the following predictable mood pattern dummy variables: Preholiday, Monday, and Friday. Preholiday is dummy variable equaling one if the date is within two trading days prior to a major U.S. holiday and includes the holiday itself if the market is open that day (Autore, Bergsma, and Jiang 2016). Control variables include Expiration Friday, as defined in Table 2, and S&PRet, the S&P 500 return of the prior day. The dependent variable for each regression is labeled as either Call OTM-ATM IV or Put OTM-ATM IV, where IV refers to implied volatility spread in percent (annualized). We cluster standard errors by year-month. *, **, and *** indicate 10%, 5%, and 1% level of significance respectively using a two-tailed test. The t-statistics are reported in parentheses. The sample period is from January 1996 to August 2014. Call OTM-ATM IV Put OTM-ATM IV (1) (2) (3) (4) (5) (6) Preholiday 0.11-0.16 (0.83) (-1.20) Monday 0.07 0.30*** (0.68) (2.82) Friday 0.33*** 0.32*** (2.83) (2.74) S&PRet 4.56 4.64 4.82 1.61 1.44 1.65 (1.36) (1.39) (1.44) (0.48) (0.43) (0.49) Expiration Friday -0.84*** -0.81*** -1.10*** -0.81*** -0.77** -1.10*** (-4.30) (-4.15) (-5.06) (-4.11) (-3.89) (-5.00) N 4,767 4,767 4,767 4,767 4,767 4,767 25
Table 6: Option Returns This table tests whether individual stocks OTM-ATM option returns after Mondays or Fridays and prior to and on holidays. Options returns are calculated as Return t = (Price t - Price 0)/Price 0 where Price 0 is the price at the close of the current day and Price t is the price at the close of the last trading day. Using OLS regressions with quarter and firm fixed effects, we regress each stock s average OTM-ATM implied volatility spread on the following predictable mood pattern dummy variables: Preholiday, Monday, and Friday. Preholiday is dummy variable equaling one if the date is within two trading days prior to a major U.S. holiday and includes the holiday itself if the market is open that day (Autore, Bergsma, and Jiang 2016). Control variables are as in Table 2. We calculate the difference between OTM and ATM option returns for calls and puts, labeled as Call OTM-ATM Return and Put OTM-ATM Return, respectively. *, **, and *** indicate 10%, 5%, and 1% level of significance respectively using a two-tailed test. The t-statistics are reported in parentheses. The sample period is from January 1996 to August 2014. Panel A: Dependent variable is subsequent one-day return Call OTM-ATM Return Put OTM-ATM Return (1) (2) (3) (4) (5) (6) Preholiday -0.73*** -0.80*** (-9.57) (-11.62) Monday -0.42*** -0.13** (-7.26) (-2.39) Friday -1.72*** -1.06*** (-27.17) (-18.59) Ret 11.15*** 11.12*** 11.11*** -6.62*** -6.66*** -6.66*** (46.16) (46.03) (46.00) (-28.11) (-28.26) (-28.29) Log(MktCap) 0.21 0.17 0.16 1.13*** 1.09*** 1.08*** (0.69) (0.58) (0.55) (3.87) (3.73) (3.70) B/M 0.48*** 0.48*** 0.48*** -0.26* -0.26** -0.26** (3.60) (3.59) (3.59) (-1.95) (-1.96) (-1.97) Momentum -0.02*** -0.02*** -0.02*** -0.07-0.07-0.08 (-0.33) (-0.34) (-0.39) (-1.18) (-1.21) (-1.24) Turnover -0.64*** -0.65*** -0.64*** -1.51*** -1.51*** -1.51*** (-5.80) (-5.86) (-5.79) (-15.27) (-15.34) (-15.28) Ivol -2.69*** -2.71*** -2.70*** -1.79*** -1.80*** -1.79*** (-32.00) (-32.13) (-32.05) (-23.60) (-23.78) (-23.70) Expiration Friday -3.58*** -3.75*** -2.23*** -2.07*** -2.19*** -1.28*** (-33.17) (-34.62) (-18.63) (-21.23) (-22.32) (-11.81) N 1,594,367 1,594,367 1,594,367 1,212,982 1,212,982 1,212,982 26