+ All Categories
Home > Documents > Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

Date post: 02-Jun-2018
Category:
Upload: javier-murphys
View: 218 times
Download: 0 times
Share this document with a friend

of 33

Transcript
  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    1/33Electronic copy available at: http://ssrn.com/abstract=1571786

    Stock Returns during Option Expiration Weeks

    and the Option-Stock Volume Ratio1

    Chris Stivers

    Terry College of Business

    University of Georgia

    Athens, GA 30602

    Licheng Sun

    Department of Finance

    College of Business

    Old Dominion University

    Norfolk, VA 23529

    This version: March 15, 2010

    1Comments welcome. Please address comments to Chris Stivers (e-mail: [email protected]; phone:

    (706) 542-3648) or to Licheng Sun (e-mail: [email protected]; phone: (757) 683-6552).

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    2/33Electronic copy available at: http://ssrn.com/abstract=1571786

    Stock Returns during Option Expiration Weeks and

    the Option-Stock Volume Ratio

    Abstract

    Over 1983 to 2008, we find that large-cap stocks with actively traded options tend to have

    substantially higher average weekly returns during option-expiration weeks (ending on a months

    third-Friday), with consistent subperiod results. This empirical regularity is also evident in large-

    cap-dominated stock portfolios. Further, over the 1996 to 2008 period with available Option Metrics

    data, we find that the average weekly stock returns for option-expiration weeks tend to be apprecia-

    bly larger for option-expiration weeks that also experience a relatively high option trading volume,

    relative to the underlying stock trading volume. In contrast, the average weekly returns for the

    option-expiration week are not much different than other weeks for smaller-cap stock portfolios,

    which contain stocks with relatively less option trading activity. Further, for our sample of large-cap

    stocks, the average weekly stock returns for the third-Friday of a calendar month are not different

    than other weeks for a pre-option-market sample over 1948 to 1972. We provide additional evidence

    that, along with the option market evidence in Lakonishok, Lee, Pearson, and Poteshman (2007),

    suggests that the sizable written call activity from non-market makers may contribute towards

    understanding our empirical findings.

    JEL Classification: G12, G13, G14

    Keywords: Option Expiration, Stock Returns, Option and Stock Trading Volume

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    3/33Electronic copy available at: http://ssrn.com/abstract=1571786

    1. Introduction

    The past three decades have experienced a dramatic growth in the size of the options market.

    According to the Chicago Board Options Exchange (CBOE), in 2007 CBOE options contract

    volume reached an all-time record of 944,471,924 contracts. The growth rates in CBOE options

    contract volume were 33%, 44%, and 40% in 2005, 2006, and 2007 respectively.1 Further, recent

    research indicates that option-related activity may have an influence on the underlying stock price

    behavior (Ni, Pearson, and Poteshman (2005)), or may contain information about the subsequent

    behavior of stock prices (Roll, Schwartz, and Subrahmanyam (2010)). Ni, Pearson, and Poteshman

    (2005) document that the closing prices of optionable stocks tend to cluster at option strike prices

    on option-expiration days. They attribute the pricing patterns to hedge rebalancing by option

    market makers and stock price manipulation by firm proprietary traders.

    Given the tremendous growth in option trading activities and this recent research, this paper

    re-examines the behavior of stock returns during option-expiration weeks. A few older studies

    examine stock returns around the expiration day with return horizons beyond the actual expiration

    day. Klemkosky (1978) analyzes weekly stock returns before and after option expiration for a

    limited sample of 14 expiration periods in 1975 and 1976. He finds that 12 of the 14 expiration

    weeks have negative returns (10 of the 14 weeks are significantly negative) with an average return

    of about 1% in the expiration week. The returns in the week after option expiration are largely

    positive with an average return of 0.4%. Using a longer sample from January 1979 to June 1985,

    Cinar and Vu (1987) examine the effect of option expiration on six large-cap stocks (IBM, GM,

    Eastman Kodak, 3M, Sears and Exxon). They find that the mean returns of these stocks during

    the expiration weeks are mostly statistically indistinguishable from zero.2 Mayhew (2000) surveys

    the empirical evidence on expiration-day effects and concludes that although there appear to have

    been specific instances of price movements on expiration days that appear to have been related to

    1See the website http://cboe.com/AboutCBOE/History.aspx.2Other studies examine option-expiration effects on the expiration day or the end of the expiration day. For

    example, Stoll and Whaley (1987, 1991) look at expiration-day effects by focusing exclusively on the trading activity

    during the last hour of trading on triple-witching days (days on which index futures, index options and options on

    index futures expire simultaneously) and first half-hour of the following day.

    1

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    4/33

    traders unwinding their positions, there is little evidence of a strong, systematic price effect around

    expiration.

    With the huge growth in option trading activities in recent years and the much longer time-series

    now available, these earlier studies seem both outdated and limited in their scope. In this article, we

    re-examine stock returns over option-expiration weeks for both large individual stocks and various

    stock indices. We focus on the weekly returns of large individual firms that have actively traded

    options over the 1996 to 2008 period, corresponding to option-volume data availability from Option

    Metrics. However, our study also examines a longer 1983 to 2008 period (with 1983 coinciding with

    the introduction of stock index options) and a pre-option-market period over 1948 to 1972.

    This article is also the first to examine whether the relative trading activity in options has a

    role in understanding the return patterns during option expiration weeks. We follow from Roll,

    Schwartz, and Subrahmanyam (2010), hereafter RSS, and examine the ratio of option trading vol-

    ume to stock trading volume, or O/S. RSS study the time-series properties and cross-sectional

    determinants of the O/S ratio. Their findings include that the O/S tends to be higher around earn-

    ings announcements and that the O/S ratio is positively related to the absolute post-announcement

    returns.

    The intuition behind examining the O/S ratio is simple. If weekly returns appear to behave

    differently during the option-expiration week (in contrast to the other weeks) and the behavior is

    tied to option-related activity, then it seems likely that the associated patterns would be stronger

    during option-expiration weeks that experience a relative large O/S. Accordingly, we also separately

    examine the mean and volatility of returns for option-expiration weeks that have a relatively high

    O/S value.3

    In stark contrast to earlier studies, we present striking new empirical evidence that average

    stock returns tend to be higher during option-expiration weeks as compared to average returns in

    other weeks, at least for large-cap stocks with very active option trading. First, for our primary

    sample of 28 large-cap individual stocks with active option activity over 1996 to 2008, the average

    3We use a weekly O/S, defined as the average of the option trading volume for the five trading days concluding in

    the last day of the calendar week (normally Friday) divided by the comparable average of stock trading volume over

    the same five trading days.

    2

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    5/33

    weekly return during option-expiration weeks is 45.0 basis points. In contrast, for the other weeks,

    the average weekly return is only 11.8 basis points. This spread in means tends to be larger for

    the stocks with more actively traded options. We note that the average return volatility for the

    option-expiration weeks is only modestly higher than the return volatility of the other weeks, at

    4.97% versus 4.57%. One-half subperiod results are consistent.

    Second, for the same sample of 28 stocks over 1996 to 2008, we examine the option-expiration

    weeks where the O/S volume ratio is relatively high (above the 75th percentile value of the O/S

    ratio for each respective stock). For our sample of large individual stocks, about 7.5% of the weeks

    meet the double conditioning of being an option-expiration week with a high O/S value. We find

    that the spread in means is even more dramatic. For the option-expiration weeks with a high O/S,

    the average weekly return is 82.5 basis points. For the remaining 92.5% of the weeks, the average

    weekly stock return is only 14.0 basis points. We note that the average return volatility for the

    option-expiration weeks with a high O/S is only modestly higher than the return volatility of the

    other weeks, at 4.87% versus 4.64%. One-half subperiod results are again consistent.

    Third, given the above patterns in the mean and volatility of the weekly stock returns, the

    Sharpe ratio during the option-expiration weeks is much higher than the Sharpe ratio during the

    other weeks. When examining all the option-expiration weeks, the average Sharpe ratio for the

    option-expiration weeks is 0.57 versus an average Sharpe ratio of 0.07 for the other weeks.4 When

    examining the option-expiration weeks that also experienced a relatively high O/S ratio, the average

    Sharpe ratio for the high O/S option-expiration weeks is 1.25 versus an average Sharpe ratio of

    0.11 for the other weeks.

    Fourth, we examine the same set of 28 firms in the earlier 1983 to 1995 period and find quali-

    tatively consistent results for the difference in means, as related to the option expiration week. We

    do not examine the O/S results in this earlier period, since the Option Metrics data begins in 1996.

    Fifth, for stock indices, we examine the S&P 500 and size-based decile stock portfolios. Over

    1983 to 2008, the option-expiration spreads in means for the S&P 500 and the large-cap stock

    portfolio are positive and statistically significant. However, the option-expiration spread in means

    4For these Sharpe ratios, we annualize the weekly return statistics by multiplying the mean excess return by 52

    and multiply the sample standard deviation by the square root of 52.

    3

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    6/33

    is much smaller and statistically insignificant for the smaller-cap portfolios, presumably portfolios

    composed of stocks with relatively much lower option activity.

    Sixth, we examine our sample of large individual stocks over an earlier pre-option-market period

    of comparable length, 1948 to 1972. For the 15 firms with available return history over the entire

    earlier period, we find essentially no spread in means between the third-Friday weeks and other

    weeks. The average spread in means between the third-Friday weeks and other weeks is only 0.4

    basis points and none of the spreads are statistically significant. The same result holds for the S&P

    500 index in the pre-1983 sample (preceding index option trading).

    We also discuss possible explanations for these return patterns and present related evidence in

    Section 6. The dramatically higher Sharpe ratios for the option-expiration weeks suggest that the

    high average returns during the option-expiration weeks cannot be rationally attributed to an ex

    ante risk-return tradeoff.

    Lakonishok, Lee, Pearson, and Poteshman (2007) document that nonmarket maker investors

    (in aggregate) have more written than purchased open interest, with written calls making up the

    majority of this written open interest. This suggests three non-fundamental-value avenues that

    could promote the observed option-expiration spread in weekly mean returns. First, retail investors

    could be purchasing the stock to cover their uncovered written calls or be purchasing stock so that

    they do not have to lower their position in the underlying stock in the event their covered calls

    expires in the money. Second, especially during option-expiration weeks with stock price increases

    for other reasons, retail investors may be offsetting their written call positions. Presumably, such

    offsetting would serve to lower the net long call position of market makers. This, in turn, could

    result in market makers reducing their short stock positions that they may have in place to delta

    hedge their long call positions, which might exert some upward pressure on stock prices. Third,

    since the market makers tend to have a net long call position, then they would benefit from stock

    price increases during the option-expiration week for their long call positions that are not delta

    hedged. Thus, market-makers might have an incentive to promote stock price appreciation during

    the option-expiration week.

    To examine the notion that the stock price might be driven up beyond the fair economic value

    4

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    7/33

    during by the option expiration, we also examine the average weekly returns for the week following

    the option expiration (under the assumption that non-fundamentals price movements would tend

    to be reversed in the subsequent week). For our sample of 28 large individual stocks over 1996 to

    2008, the average return for fourth-Friday weeks is -2.9 basis points versus an average return of

    18.1 basis points for the other weeks that are not option-expiration weeks (referring to a months

    first-Friday, second-Friday, and fifth-Friday weekly return). Recall that option-expiration weeks

    have a mean of 45.0 basis points, so this result suggests a return reversal in the week following the

    option expiration, which is consistent with the explanation suggested in the prior paragraph.

    Finally, in regard to our O/S volume ratio results for option-expiration weeks, we acknowledge

    that it is possible that option activity responds to an increasing stock price (rather than the option

    activity contributing to the observed high average weekly returns for the option-expiration weeks).

    If so, it seems likely that other non-option-expiration weeks that have a high O/S value would also

    be associated with higher average weekly stock returns. We also investigate this possibility for

    our sample of 28 large stocks over 1996 to 2008. For non-option-expiration weeks that also have a

    relatively high O/S volume ratio (an O/S above the 75th percentile value of the firms respective

    O/S distribution), the average weekly stock return is 38.7 basis points across our 28 stocks. While

    sizably positive, this average weekly return is only about 47% as large as the average weekly returns

    for the option-expiration weeks that also have a high O/S ratio (which have an average weekly

    return of 82.5 basis points). We interpret this as additional circumstantial evidence that suggest

    option-related trading activity tends to promote stock price increases in the option-expiration week.

    This article proceeds as follows. Section 2 reviews related theory and other related empirical

    evidence. Sections 3 and 4 present our results for large individual stocks with active option trading

    and size-based stock portfolios, respectively. Section 5 extends the results to the S&P 500 index.

    Finally, Section 6 concludes and also provides some exploratory analysis on the likely causes of the

    documented patterns for option-expiration weeks.

    5

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    8/33

    2. Additional Background and Literature Review

    Why would trading activities in options have any impact on the underlying stock prices? After all,

    in the Black-Scholes world, options are viewed as redundant assets whose payoffs may be replicated

    using a dynamic trading strategy in stocks and cash, and therefore should have no impact on

    underlying stock prices. A number of theoretical articles have offered some interesting perspectives

    on this issue. For example, Detemple and Selden (1991) demonstrate that in incomplete markets,

    the valuation of option prices and their underlying stock prices is a simultaneous pricing problem.

    Although in general stock prices depend on the contractual characteristics of options available for

    trading, they find that, in a mean-variance model, option trading activity can lead to an increase

    in the equilibrium stock price as well as a reduction in the stocks volatility. Cao (1999) identifies a

    similar effect under an alternative theoretical framework. His model predicts that the availability of

    options contracts can incentivize investors efforts to collect information about asset payoffs, which

    in turn leads to an increase in the expected prices of the underlying as well as positively correlated

    assets, and a reduction in price volatility. Mayhew (2000) provides a comprehensive review on the

    topic.

    Other studies evaluates the pricing effects related to the introduction of option contracts; see

    Conrad (1989), Detemple and Jorion (1990), and Sorescu (2000). These studies find that option

    listing is associated with positive abnormal stock returns in early years (prior to 1981) but insignif-

    icant or negative returns after 1981, which is likely due to some regulatory changes in 1981.

    Given our primary empirical results that indicate option-expiration weeks have higher average

    stock returns and much higher Sharpe ratios than other weeks for large-cap stock with activity

    traded options, the results in Ni, Pearson, and Poteshman (2005) seem particularly relevant. They

    consider pricing influences related to option activity that can be described as price movements not

    strictly related to fundamental value information. Specifically, they attribute the pricing patterns

    they uncover to hedge rebalancing by option market makers and to stock price manipulation by

    firm proprietary traders.

    6

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    9/33

    3. Weekly Returns for Large Stocks with Active Option Trading

    3.1. Sample Selection

    We desire to examine a set of individual stocks that have actively traded options. Further, we

    wish to examine large-cap stocks, because of their economic importance and because large-cap

    stocks account for most option activity, see Lakonishok, Lee, Pearson, and Poteshman (2007).

    Additionally, we elect to limit the number of firms in our initial investigation so that we may report

    results on each of the important stocks, firm by firm. Finally, our initial firm-level investigation is

    limited to 1996 through 2008, based on availability of option trading volume from Option Metrics.

    Given our goals and data limitations, we initially investigate the following set of 28 large-capstocks with actively traded options: International Business Machines (IBM), Texas Instruments

    (TXN), 3M (MMM), Eastman Kodak (EK), Hewlett-Packard (HPQ), Coca-Cola (KO), Boeing

    (BA), General Electric (GE), Wal-Mart(WMT), Merck (MRK), Halliburton (HAL), Schlumberger

    (SLB), Alcoa (AA), Bristol-Myers Squibb (BMY), Johnson & Johnson (JNJ), McDonalds (MCD),

    International Paper (IP), Occidental Petroleum (OXY), Honeywell (HON), Xerox (XRX), Ford (F),

    Pepsico (PEP), Black & Decker (BDK), Dow Chemical (DOW), Gap (GPS), Baxter International

    (BAX), Avon Products (AVP), and HJ Heinz (HNZ) (ticker in parentheses, with the firms ordered

    based on the firms 75th percentile value of their respective weekly O/S volume ratio). These 28

    firms are a subset of the 50 individual stocks from Dennis, Mayhew, and Stivers (2006) (DMS);

    where the 50 firms from DMS are the 50 individual firms that have the largest option trading

    volume over 1988 to 1995. These 28 firms are chosen (out of the 50) because they have both a

    complete return history over 1996 through 2008 and available option trading volume from Option

    Metrics. Using a sample of firms that were chosen for their high option trading activity over the

    prior 1988 to 1995 period provides a set of firms that are expected to have high option activity

    and provides ob jectivity to our firm selection. We note that these firms continue to have actively

    traded options over our later sample period, but that there is a sizable cross-sectional dispersion

    in the firms average O/S volume ratio across the 28 firms.

    Table 1 reports summary statistics on these firms weekly stock returns, average daily stock

    7

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    10/33

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    11/33

    This spread in means tends to be larger for the stocks with a high O/S volume ratio. For the

    top 10 O/S stocks (ranked by the 75th percentile O/S volume ratio for each respective stock), the

    average weekly return during option-expiration weeks is 58.3 basis points, versus 6.6 basis points

    for the other weeks.

    For the 28 stocks, we also find that the average return volatility for the option-expiration weeks

    is only modestly higher than the return volatility of the other weeks, at 4.97% versus 4.57%. This

    modest different in return volatility casts doubt on a risk-return tradeoff explanation for the high

    average returns during the option-expiration weeks. Given these patterns in the mean and volatility

    of the weekly stock returns, the Sharpe ratio during the option-expiration weeks is much higher

    than the Sharpe ratio during the other weeks. The average Sharpe ratio for the option-expiration

    weeks across the 28 stocks is 0.57 versus an average Sharpe ratio of 0.07 for the other weeks.

    We also examine the option-expiration weeks in a GARCH system, since the volatility of the

    weekly returns during the option-expiration weeks tends to be a little higher than for the other

    weeks and to control for heteroskedasticity in general. We evaluate the mean and weekly volatility

    of the third-Friday weekly returns in a GARCH framework that allows both the weekly mean and

    volatility to be different for the third-Friday week (as compared to the other weeks). We estimate

    the following model:

    Ri,t= 0+ 1DumFr3t + t (2)

    vi,t= 0+ 12

    t1+ 2vi,t1+ 3DumFr3t (3)

    where vi,t is the conditional variance oft from equation (2) for week t; 2t1 is the lagged squared

    return shock; the s are coefficients to be estimated for the conditional variance equation; and the

    other terms are as defined for equation (1). We estimate the GARCH system, given by equations

    (2) and (3), simulataneously, using maximum likelihood with a conditional normal distribution. We

    compute T-statistics for each estimated coefficients, based on heteroskedastic- and autocorrelation-

    consistent standard errors.

    We find the following. The conclusions for the conditional mean equation are the same as

    depicted in Table 2. The estimated 1 coefficients indicates that the conditional mean for the

    option-expiration weeks is 38.4 basis points higher than the conditional mean for the other weeks,

    9

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    12/33

    with seven of the 1 estimates being positive and statistically significant across the 28 firms. Next,

    for the variance equation, we find that the estimated 3 coefficients are positive for 21 of the 28

    firms and positive and statistically significant for 2 of the 28 firms (versus negative and statistically

    significant for 0 of the 28 firms). This indicates that the third-Friday volatility also tends to be

    modestly higher than that for the other weeks, as indicated by the summary volatility results

    reported earlier.

    3.3. Average Weekly Returns for Option-Expiration Weeks with a Relatively

    High O/S Volume Ratio

    We next investigate whether the spread in mean returns between the option-expiration weeks and

    other weeks is stronger when the option-expiration week also has a relatively high O/S option

    volume ratio.

    Here, we first estimate the following model to calculate the spread in weekly mean returns:

    Ri,t= 0+ 1(DumFr3t Dum

    HiOSt ) + t (4)

    WhereDumHiOSt is a dummy variable that equals one if a weeks O/S volume ratio is above the 75th

    percentile value of the weekly O/S distribution for each respective stock. The other terms are as

    defined for equation (1). Thus, the product of the two dummy variables indicates option-expiration

    weeks that also have a relatively high O/S volume ratio. We calculate and report T-statistics in

    parentheses, based on heteroskedastic- and autocorrelation-consistent standard errors, to indicate

    whether the estimated1 is reliably different than zero.

    Table 3 reports the results. For our primary sample of 28 large-cap individual stocks over

    1996 to 2008, the average weekly return is 82.5 basis points for option-expiration weeks that also

    experience a relatively high O/S volume ratio. In contrast, for the other weeks, the average weekly

    return is only 14.0 basis points. This spread in means is pervasive and sizable, with 11 of the 28

    stocks having a spread in the weekly mean returns of 100 basis points or larger. For nine of the

    individual stocks, the return spread is statistically significant with a 10% p-value or greater. Also,

    as shown in Table 4, Panel B, one-half subperiod results are consistent with the return patterns in

    Table 3.

    10

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    13/33

    This spread in means tends to be larger for the stocks with a high O/S volume ratio. For the

    top 10 O/S stocks (ranked by the 75th percentile weekly O/S for each respective stock), the average

    weekly return is 129.1 basis points, versus 8.6 basis points for the other weeks and the return spread

    is positive for all 10 stocks.

    For the 28 stocks, we also find that the average return volatility for the option-expiration weeks

    with a high O/S volume ratio is only modestly higher than the return volatility of the other weeks,

    at 4.87% versus 4.64%. Again, in our view, this very modest difference in return volatility casts

    doubt on a risk-return tradeoff explanation for the high average returns during the option-expiration

    weeks. Given these patterns in the mean and volatility of the weekly stock returns, the Sharpe

    ratio during the option-expiration weeks is much higher than the Sharpe ratio during the other

    weeks. The average Sharpe ratio for the option-expiration weeks across the 28 stocks is 1.25 versus

    an average Sharpe ratio of 0.11 for the other weeks.

    Some readers might be concerned that an increasing time trend in the O/S volume ratio might

    result in essentially all of the high O/S realizations occurring later in our sample. If so, then the

    O/S might not be informative, but rather our O/S results could just indicate a a higher average

    weekly return for option-expiration weeks later in our sample. Our investigation indicates this is

    not a material concern for the following reasons. First, as shown in Table 4, our O/S results are

    reliably evident for both one-half subperiods separately (the January 1996 to June 2002 and July

    2002 to December 2008 periods). Second, when estimating the O/S model over our entire 1996

    to 2008 period, we tabulate what proportion of the high O/S weeks occur in the first one-half

    subperiond versus the second one-half subperiod. Across our 28 stocks, 40% of the high O/S weeks

    occur in the first one-half subperiod and 60% in the second one-half subperiod, and 8 of the 28

    stocks have more high O/S weeks in the first half of our sample. We conclude that our O/S results

    are not due to an increasing time-trend in the O/S.

    We also examine the option-expiration weeks in a GARCH system, since the volatility of the

    weekly returns during the option-expiration weeks tends to be a little higher than for the other

    weeks and to control for heteroskedasticity in general. We evaluate the mean and weekly volatility

    of weekly returns in a GARCH framework that allows both the weekly mean and volatility to be

    11

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    14/33

    different for third-Friday weeks that also have a high O/S realization (as compared to the other

    weeks). We estimate the following model:

    Ri,t= 0+ 1(DumFr3t Dum

    HiOSt ) + t (5)

    vi,t= 0+ 12

    t1+ 2vi,t1+ 3(DumFr3t Dum

    HiOSt ) (6)

    where vi,t is the conditional variance oft from equation (4) for week t; 2t1 is the lagged squared

    return shock; the s are coefficients to be estimated for the conditional variance equation; and the

    other terms are as defined for equation (4). We estimate the GARCH system, given by equations

    (5) and (6), simulataneously, using maximum likelihood with a conditional normal distribution. We

    compute T-statistics for each estimated coefficients, based on heteroskedastic- and autocorrelation-

    consistent standard errors.

    We find the following. The conclusions for the conditional mean equation are the same as

    depicted in Table 3. The estimated 1 coefficients indicates that the conditional mean for the

    option-expiration weeks is 73.4 basis points higher than the conditional mean for the other weeks,

    with 12 of the 1 estimates being positive and statistically significant across the 28 firms. Next,

    for the variance equation, we find that the estimated 3 coefficients are positive for 13 of the 28

    firms and positive and statistically significant for only one of the 28 firms (versus negative and

    statistically significant for 0 of the 25 firms). This indicates that the volatility for option-expiration

    weeks with a high O/S tends to be little different than the volatility for the other weeks, as indicated

    by the summary volatility results reported earlier.

    3.4. Individual Stock Results for other Sample Periods

    Next, we examine our sample of 28 firms over two alternate earlier periods. First, we examine the

    1983 to 1995 period. We choose this period because it is the same length as our primary 1996 to

    2008 period (13 years). Also, 1983 was the year that trading in stock-index options commenced,

    which we examine later in this study. We do not examine the O/S volume ratio results in this

    earlier period, since the Option Metrics data begins in 1996.

    Column two of Table 5 reports the results for the 1983 to 1995 period for our sample of 28

    individual stocks. We find that the spread in means between the option-expiration weeks and the

    12

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    15/33

    other weeks also tends to be positive, with an average spread in means of 26.4 basis points and

    with a positive spread observed for 22 of the 28 firms. This finding supports our earlier evidence

    that indicates average weekly returns tend to be higher for option-expiration weeks for individual

    stocks with activity traded options.

    For comparison, we also evaluate the spread in means for our sample of 28 firms over the entire

    26 year period from 1983 to 2008. Column one of Table 5 reports the results. For this case, we find

    that the average spread in means is 29.8 basis points and the option-expiration spread is positive

    for 26 of the 28 stocks.

    Over our primary sample of 1996 to 2008, 2008 is noteworthy as a dramatic bear market with

    the S&P 500 down over 38% for the year and with extreme day to day return volatility. Readers

    might wonder whether this year has a strong influence on our results. Accordingly, we also re-

    estimate our primary models over the 1996 to 2007 period (omitting the 2008 data). The results

    are quite similar to those reported in Tables 2 and 3.

    Finally, we also evaluate 1948 to 1972 as a period that predates the opening of option trading on

    the Chicago Board Option Exchange. This 25 year period is essentially the same length as our 1983

    to 2008 period. Column four of Table 5 reports the results for the 1948 to 1972 period. For this

    earlier period, our evaluation only includes 15 of the 28 individual stocks in our primary sample,

    because we only include the stocks that have the full stock return history back to January 1948.

    Over 1948 to 1972, the average weekly stock return for the third-Friday weeks is essentially the same

    as for the other weeks. The average spread in means is only 0.4 basis points and only about half

    the stocks have positive spreads. None of the 15 stocks have a spread in means that is statistically

    significant. This finding provides additional circumstantial evidence that option-related activity

    may be important for understanding our primary results.

    To conclude, over 1983 to 2008 and for inclusive subperiods, the weekly stock return for option-

    expiration weeks tends to be higher for large, individual firms with actively traded options; with the

    difference being sizable and statistically reliable in many cases. Further, the average weekly return

    is even higher for option-expiration weeks that also experience a relatively high option trading. In

    contrast, over the pre-option period, the weekly stock returns for the 3rd Friday-week of a calendar

    13

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    16/33

    month is essentially the same as the other weekly returns.

    4. Option-Expiration Weekly Returns for Size-based Portfolios

    If option-related activity is a factor behind the patterns documented in Tables 2 through 5, then

    it seems that the return patterns would be stronger for firms with actively traded stock options.

    The comparisons in Tables 2 and 3 for stocks with a higher O/S volume ratio versus stocks with

    a lower O/S ratio suggest the degree of option activity may be important. If so, then the return

    patterns seem likely to be stronger in large-cap stock portfolios, as compared to smaller-cap stock

    portfolio, since large firms have more actively traded options and many small firms have very little

    option trading or may not even have publicly-traded options.

    Accordingly, we investigate the weekly returns of size-based stock portfolio and report the results

    in Table 6. We evaluate the weekly returns of the value-weighted, size-based, decile stock portfolios

    from the French data library. The results are reported in rows one through ten of the table.

    We find that the largest size-based decile exhibits a statistically reliable spread between the

    average weekly returns for the option-expiration weeks and the average weekly return of the other

    weeks. The spread in weekly returns for the largest size-based decile is 27.5 basis points, with an

    average option-expiration weekly return of 41.7 basis points versus an average weekly return for

    the other weeks of only 14.2 basis points. The spread in means is statistically significant at a 10%

    p-value.

    For the other nine size-based decile portfolios, the spread between the average weekly return

    for the third Friday versus the other weeks is also positive for all cases. However, the magnitude

    is appreciably lower, as compared to the largest size-based decile, and none of the spreads are

    statistically reliable for the smaller nine size-based deciles.

    To further investigate the differences between the largest size-based decile and the smaller

    firms, we calculate a weekly return spread defined as the difference between the weekly return for

    the largest size-based decile and the average return for that week for the smallest five size-based

    decile portfolios. Then, we evaluate the average of this weekly return spread to see if it is reliably

    different for the third-Friday of a calendar month. The results are reported in row 11 of Table 6.

    14

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    17/33

    We find that the average difference between the large-firm and small-firm portfolio is 0.177% for

    the third-Friday week, as compared to a -0.042% for the other weeks. The difference of 0.220% is

    statistically reliable, indicating that there a reliable difference between the behavior in the large-firm

    and small-firm stock portfolios.

    To conclude, for the large-firm portfolio in Table 6, we find that the spread between the average

    weekly return for the third-Friday and the average weekly return for the other Fridays is sizable and

    statistically significant. Further, we find that the third-Friday spread behavior is reliably stronger

    in the large-firm portfolios, as compared to the smaller-firm portfolios. This behavior is consistent

    with the notion that option-related activity influences the prices of large firms that have more

    actively traded stock options.

    5. Weekly S&P 500 Index Returns and Option Expiration

    5.1. Discussion of Main Results

    Given the evidence that large-cap firms with actively traded stock options exhibit particular strong

    returns during the option expiration weeks, we inquire if such a trading pattern is limited only to

    these particular stocks or do they spill over to the broad market in general? In particular, we are

    interested in the the performance of the S&P 500 index during option expiration weeks for two

    major reasons. First, it is the most widely used benchmark index for many mutual funds as well as

    hedge funds. Second, some of the most actively traded futures and option contracts are based on

    the prices the S&P 500 index. Therefore if we find a similar trading pattern in the S&P 500 index,

    this indicates that our empirical findings carry broad-based economic significance to all investors.

    We collect daily S&P 500 index prices from the Center for Research in Security Prices (CRSP)

    and convert them to weekly returns. The data is available from 1962 to 2008. Similar to what

    we did in previous sections, we focus on the sample period from July 1983 to December 2008 for

    our main results. It should be noted that the CBOE introduces the trading of S&P 500 index

    options in July 1983. We calculate weekly S&P 500 index returns using Friday-to-Friday prices. If

    the stock market closes on a Friday and therefore the price on Friday is not available, then we use

    15

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    18/33

    Thursdays prices instead, and so on.

    To contrast with our main sample period, we also include the results calculated from an earlier

    sample period that spans from July 1962 to the week ended on July 1st, 1983 (which is a Friday).

    It should be noted that during the first half of this earlier sample period (1962 to 1972), there were

    no listed option contracts available for trading. In the second half (1973 to 1983), listed option

    contracts were available for some individual equities but the overall size of the option market is

    quite small compared to the cash market.

    Panel A of Table 7 reports the mean return, standard deviation, and return spread for the S&P

    500 index during the third Friday week of a calendar month. To calculate the return spread, we

    run the same regression as shown in equation (1), but for the S&P 500 index return. The estimated

    coefficient 1 captures the return spread between the option expiration week and other weeks of

    a calendar month. For comparison, we also include the results for the full sample (both option

    expiration week and other weeks of a calendar month).

    We find that the mean return for the third Friday week is 37 basis points, more than double the

    average return for the full sample average of 15 basis points for all weeks. Meanwhile the volatility

    during the third Friday week is about the same as the full sample average. The return spread

    between third Friday week and other weeks is also significantly positive at 28 basis points.

    Further, we find the results are strikingly different for the pre-1983 sample. Here the third

    Friday weekly returns do not appear to be different from other weeks. The mean return for third

    Friday weeks is 11 basis points, nearly identical to the full sample average of 12 basis points. The

    return spread between option expiration weeks and non-expiration weeks is -1 basis points and

    statistically insignificant. In terms of return standard deviations, option expiration weeks and

    other weeks are also very similar (1.95% vs. 1.99%).

    Hence our results indicate that there appear to have been a change in the markets trading

    activity during the option expiration weeks in the two sample periods. When trading in option

    contracts are either not available or relatively unimportant due to a small market size, returns in

    the third Friday weeks are not dissimilar to other weeks. However as options market become more

    active in the more recent sample, returns in the third Friday weeks become pronouncedly more

    16

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    19/33

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    20/33

    value-weighted and equal-weighted indexes are very strong around the turn of a month, defined

    as the last trading day of a month plus the first three trading days of a month, and weak during

    other trading days of a month. Lakonishok and Smidt (1988) also confirms this turn of month

    effect using 90 years of data on the Dow Jones Industrial Average. Since the third-Friday weeks

    do not coincide with turn of the month days, our results are unlikely to be driven by this effect.

    Nevertheless it is interesting to know what our results will look like after controlling for this effect.

    To this end, we exclude the weekly S&P 500 returns that do not belong to the third-Friday week

    or the two weeks adjacent to it. In other words, for every calendar month, we only examine the

    returns from the three weeks around option expiration. In Panel D of Table 7, we find that by

    excluding the weeks not neighboring the option expiration, the sample mean return is reduced by

    about two thirds from 0.15% to only 0.05% in our 1983 to 2008 sample. This shows that the turn

    of month effect continues to be significant in our sample period. Meanwhile excluding the turn of

    month effect also strengthens our main result. We find the spread between third Friday weeks and

    other weeks now increases to 48 basis points with a t-statistic of 3.

    Overall we find the third-Friday returns for the S&P 500 exhibit similar patterns as those stocks

    with actively traded options and large-cap stock portfolios. The pattern is consistent across the

    subperiod samples and robust after accounting for the 1987 market crash, September 11, 2001

    terrorist attack, as well as the turn of the month effect. Further the lack of a similar pattern in the

    pre-1983 sample indicates that it likely to be attributable to option market trading activities.

    6. Conclusions and Additional Evidence towards Interpretation

    We document that large-cap stocks with actively traded options tend to have higher average weekly

    returns during third-Friday weeks of the month over 1983 to 2008, with consistent subperiod results.

    This empirical regularity is also evident in several large-cap-dominated stock portfolios that we

    examine. Further, over the 1996 to 2008 period with available Option Metrics data, we find that

    the average weekly stock return for option-expiration weeks tends to be appreciably larger for

    option-expiration weeks that also experience a relatively high option trading volume, relative to

    the underlying stock trading volume.

    18

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    21/33

    In contrast, for our sample of large-cap stocks, the average weekly stock return for the third-

    Friday of a calendar month is not different than other weeks for a pre-option-market sample over

    1948 to 1972. Further, over our recent sample period of 1983 to 2008, the third-Friday average

    weekly returns are not much different than other weeks for smaller-cap stock portfolios, which

    contains stocks with relatively less option trading activity.

    In our view, this collective evidence provides solid circumstantial evidence that suggests option

    activity tends to contribute to higher average returns during the option-expiration week, at least

    for stocks with relatively high option trading activity. Further, the much higher Sharpe ratios for

    option-expiration weeks suggest that the high average returns during the option-expiration weeks

    cannot be rationally attributed to an ex ante risk-return tradeoff.

    However, it is unclear what types of option-related activity might contribute to understanding

    the documented effect in option expiration weeks. While our aggregated option volume data limits

    us on what we can determine in this area, we do discuss some limited exploratory analysis in this

    direction in this concluding section.

    With access to a unique option data set, Lakonishok, Lee, Pearson, and Poteshman (2007)

    document that nonmarket maker investors (in aggregate) have more written than purchased open

    interest, with written calls making up the majority of this written open interest. Further, they

    note (page 841) that call writing is the most important category of option trade, based on open

    interest, and the second most important, based on open volume. In our view, their summary

    of option market activity suggests three non-fundamental-value avenues that could promote the

    observed option-expiration spread in weekly mean returns.

    First, retail investors could be purchasing the stock to cover their uncovered written calls or be

    purchasing stock so that they do not have to lower their position in the underlying stock in the

    event their covered calls expires in the money. Such stock demand could tend to promote a stock

    price increase during the option-expiration week. Presumably, such activity would be amplified if

    the stock price were increasing during the option-expiration week for fundamentals reasons.

    Second, especially during option-expiration weeks with stock price increases for fundamentals

    reasons, retail investors may be generating sizable option volume by offsetting their existing written

    19

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    22/33

    call positions. Presumably, such offsetting would serve to lower the net long call position of market

    makers. This, in turn, could result in market makers lowering their short stock positions that they

    may have in place to delta hedge their long call positions. Market makers covering their short stock

    positions could promote further stock price increases during the option-expiration week.

    Third, since the market makers tend to have a net long call position, then they would benefit

    from stock price increases during the option-expiration week for their long call positions that are not

    delta hedged. Thus, market-makers might have an incentive to promote stock price appreciation

    during the option-expiration week.

    If upward stock price movements during option-expiration weeks tend to have a partial non-

    fundamental-value component, then one would presumably expect to see lower average returns

    during the week following option expiration, as prices revert to their fundamental value. Accord-

    ingly, we also examine the average weekly returns for the week following the option expiration. For

    our sample of 28 large individual stocks over 1996 to 2008, the average return for fourth-Friday

    weeks is -2.9 basis points (with a median of -4.8 basis points) versus an average return of 18.1 basis

    points (with a median of 12.8 basis points) for the other weeks that are not option-expiration weeks

    (referring to a calendar months first-Friday, second-Friday, and fifth-Friday weekly return). Recall

    that option-expiration weeks have a mean of 45.0 basis points (with a median of 48.4 basis points

    across our 28 stocks), so this result suggests a return reversal in the week following the option

    expiration, which is consistent with the explanation suggested in the prior paragraph.6

    Next, recall that the average weekly stock returns for option-expiration weeks are especially

    large when the option-expiration week also experiences a relatively high O/S volume ratio. As we

    pointed out in our introduction, we acknowledge that it is possible that option activity responds

    to an increasing stock price (rather than the option activity contributing to the observed high

    average weekly returns for the option-expiration weeks). If so, it seems likely that other non-

    option-expiration weeks that have a high O/S value would also be associated with higher average

    weekly stock returns. We investigate this possibility for our sample of 28 large stock over 1996

    6The same analysis over the earlier 1983 to 1995 period yield consistent results. For this earlier period, the average

    weekly stock return for fourth-Friday weeks is -9.2 basis points versus an average weekly stock return of 55.9 basis

    points for the option-expiration weeks.

    20

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    23/33

    to 2008. For non-option-expiration weeks that also have a relatively high O/S volume ratio, the

    average weekly stock return is 38.7 basis points (with a median of 32.1 basis points for the 28

    stocks) across our 28 stocks.7 While sizably positive, this average weekly return is only about 47%

    as large as the average weekly returns for the option-expiration weeks that also have a high O/S

    ratio (which have an average weekly return of 82.5 basis points with a median of 86.6 basis points

    for the 28 stocks).

    Finally, our above discussion and the supplementary results suggest that written calls held by

    non-market makers may have a role in understanding our primary findings. If so, then activity

    related to written calls (such as offsetting written calls that may otherwise expire in the money)

    should be stronger following periods of up movement in the underlying stock prices. If so, then

    option-related influences on the underlying stock price during option-expiration weeks may be

    stronger following recent stock price appreciation. To investigate this possibility, we calculate the

    average returns for third-Friday weeks following positive prior three-week returns and negative

    prior three-week returns. We conjecture that average option-expiration week stock returns may be

    higher when the prior three-week returns are positive (as contrasted to the case when the prior

    three-week returns are non-positive). The evidence in Panel D of Table 7 supports this conjecture.

    We find that for the S&P 500 index, the third-Friday weekly average return following positive

    prior three-week returns is almost twice as much as the mean return conditional on negative prior

    three-week returns. In addition, volatility is lower and the proportions of positive returns are also

    higher following positive prior three-week returns.

    Overall, we interpret this sections additional evidence as supportive of the notion that option-

    related activity contributes to the high average stock returns that we document for the option-

    expiration week. In later revisions to this study, we hope to further probe the potential explanations

    that may be behind our intriguing empirical findings tied to the option-expiration weeks.

    7Recall that we define a high O/S value as an O/S realization above the 75th percentile value of the firms respective

    O/S distribution.

    21

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    24/33

    REFERENCES

    Ariel, R. A., 1987, A monthly effect in stock returns, Journal of Financial Economics17, 161174.

    Cao, H. Henry, 1999, The effect of derivative assets on information acquisition and price behavior

    in a rational expectations equilibrium, Review of Financial Studies12, 131163.

    Cinar, E. Mine, and Joseph Vu, 1987, Evidence on the effect of option expirations on stock prices,

    Financial Analysts Journal43, 5557.

    Conrad, Jennifer, 1989, The price effect of option introduction, The Journal of Finance44, 487499.

    Dennis, Patrick, Stewart Mayhew, and Chris Stivers, 2006, Stock returns, implied volatility in-

    novations, and the asymmetric volatility phenomenon, Journal of Financial and Quantitative

    Analysis41, 381406.

    Detemple, Jerome, and Philippe Jorion, 1990, Option listing and stock returns: An empirical

    analysis, Journal of Banking and Finance14, 781801.

    Detemple, Jerome, and Larry Selden, 1991, A general equilibrium analysis of option and stock

    market interactions, International Economic Review32, 279303.

    Klemkosky, Robert C., 1978, The impact of option expiration on stock prices, Journal of Financial

    and Quantitative Analysis13, 507518.

    Lakonishok, Josef, Inmoo Lee, Neil D. Pearson, and Allen M. Poteshman, 2007, Option market

    activity,Review of Financial Studies20, 813857.

    Lakonishok, Josef, and Seymour Smidt, 1988, Are seasonal anomalies real? a ninety-year perspec-

    tive, Review of Financial Studies1, 403425.

    Mayhew, Stewart, 2000, The impact of derivatives on cash markets: What have we learned?,

    Working paper, Unversity of Georgia.

    Ni, Sophie Xiaoyan, Neil D. Pearson, and Allen M. Poteshman, 2005, Stock price clustering on

    option expiration dates, Journal of Financial Economics78, 4987.

    22

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    25/33

    Roll, Richard, Eduardo Schwartz, and Avanidhar Subrahmanyam, 2010, O/S: The relative trading

    activity in options and stock, Journal of Financial Economics96, 117.

    Sorescu, Sorin M., 2000, The effect of options on stock prices: 1973 to 1995,The Journal of Finance

    55, 487514.

    Stoll, Hans R., and Robert E. Whaley, 1987, Program trading and expiration-day effects,Financial

    Analysts Journal43, 1628.

    , 1991, Expiration-day effects: What has changed?, Financial Analysts Journal47, 5872.

    23

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    26/33

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    27/33

    Table 2: Average 3rd Friday Weekly Returns for 28 Large Individual Stocks

    This table reports the average weekly returns for 3rd-Friday weeks (the option-expiration week) for

    our sample of 28 large-cap stock with active option trading. The weekly returns are five-day Monday-

    through-Friday returns, ending on a Friday. The sample period is 1996 to 2008, corresponding to the data

    availability from Option Metrics. Column one reports the ticker for each individual large stock. Column

    two reports the average return on the 3rd-Friday weeks when options expire. Column three reports the

    average return for the remaining weekly returns. Column four reports the spread in the mean returns

    between columns two and three, with t-statistics in parenthesis that are calculated with heteroskedastic

    and autocorrelation consistent standard errors. For the t-statistics in parentheses for column four, 1, 2,

    3, indicate 1%, 5%, and 10% p-values, respectively.

    1. Ticker 2. Mean 3. Mean 4. Spread of

    3rd

    Friday Other Weeks Means(Col. 2 - Col 3.)

    IBM 0.758% 0.161% 0.597% (1.30)

    TXN 0.553% 0.288% 0.266% (0.43)

    MMM 0.496% 0.092% 0.404% (1.16)

    EK 0.141% -0.271% 0.412% (0.84)

    HPQ 1.080% 0.047% 1.033% (1.87)3

    KO 0.596% -0.013% 0.608% (1.65)3

    BA 0.274% 0.108% 0.166% (0.35)

    GE 0.747% -0.011% 0.758% (1.91)3

    WMT 0.471% 0.296% 0.175% (0.43)

    MRK 0.708% -0.039% 0.747% (1.84)

    3

    HAL 0.342% 0.292% 0.050% (0.08)

    SLB 0.633% 0.183% 0.450% (0.91)

    AA 0.113% 0.163% -0.050% (-0.10)

    BMY 0.131% 0.181% -0.050% (-0.12)

    JNJ 0.746% 0.080% 0.665% (1.97)2

    MCD 0.061% 0.295% -0.234% (-0.66)

    IP 0.606% -0.210% 0.817% (1.86)3

    OXY 0.662% 0.317% 0.346% (0.86)

    HON 0.371% 0.143% 0.228% (0.44)

    XRX 0.470% -0.015% 0.485% (0.66)

    F 0.184% 0.000% 0.184% (0.30)

    PEP 0.157% 0.212% -0.055% (-0.17)

    BDK 0.657% 0.016% 0.641% (1.39)

    DOW -0.007% 0.172% -0.179% (-0.44)

    GPS 0.296% 0.301% -0.005% (-0.01)

    BAX 0.532% 0.193% 0.339% (0.81)

    AVP 0.312% 0.280% 0.032% (0.08)

    HNZ 0.513% 0.031% 0.482% (1.79)3

    average(all 28) 0.450% 0.118% 0.333%

    average(1st 10) 0.583% 0.066% 0.517%

    average (last 10) 0.348% 0.133% 0.215%

    25

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    28/33

    Table 3: Average 3rd Friday Weekly Returns when Option Trading is High: Large Individual Stocks

    This table reports the average weekly returns for 3rd-Friday weeks (the option-expiration week) when

    the option trading volume is relatively high for our sample of 28 large-cap stocks. Weeks are categorized as

    having a relatively high option trading volume when the option volume is high relative to the underlying

    stock trading volume, as explained in Table 1. The sample period is 1996 to 2008. Column one reports

    the ticker for each individual large stock. Column two reports the average return on the 3rd-Friday weeks

    that have a relatively high O/S option-stock volume ratio. Column three reports the average return for

    the remaining weekly returns. Column four reports the spread in the mean returns between columns two

    and three, with t-statistics in parenthesis that are calculated with heteroskedastic and autocorrelation

    consistent standard errors. For the t-statistics in column four, 1, 2, 3, indicate 1%, 5%, and 10% p-values,

    respectively. Column five reports the number of weekly returns that meet the column-two conditions for

    each stock.

    1. Ticker 2. Mean 3. Mean 4. Spread of 5. Number

    3rd Friday Other Weeks Means of Weeks

    High Opt. Vol. (Col. 2 - Col 3.) Col. 2

    IBM 1.403% 0.177% 1.226% (1.70)3 67

    TXN 1.267% 0.266% 1.000% (1.02) 56

    MMM 0.532% 0.158% 0.373% (0.72) 49

    EK 1.551% -0.296% 1.847% (2.61)1 44

    HPQ 1.341% 0.184% 1.157% (1.37) 59

    KO 1.430% -0.004% 1.434% (2.85)1 62

    BA 0.803% 0.097% 0.706% (1.07) 47

    GE 1.756% 0.017% 1.738% (2.90)1

    57WMT 1.044% 0.278% 0.765% (1.38) 51

    MRK 1.779% -0.018% 1.797% (3.21)1 57

    HAL 0.006% 0.325% -0.319% (-0.38) 47

    SLB 0.905% 0.229% 0.676% (0.84) 58

    AA -0.071% 0.165% -0.237% (-0.34) 40

    BMY 0.564% 0.139% 0.425% (0.67) 48

    JNJ 1.370% 0.121% 1.249% (2.76)1 61

    MCD 1.057% 0.180% 0.876% (1.72)3 47

    IP 1.005% -0.103% 1.108% (1.45) 49

    OXY 0.688% 0.376% 0.311% (0.40) 43

    HON 0.153% 0.199% -0.046% (-0.05) 52

    XRX 0.387% 0.074% 0.313% (0.20) 49

    F 0.554% 0.006% 0.548% (0.62) 45

    PEP 0.309% 0.191% 0.118% (0.21) 44

    BDK 0.068% 0.171% -0.103% (-0.19) 50

    DOW 0.317% 0.117% 0.200% (0.32) 46

    GPS 0.373% 0.294% 0.079% (0.09) 51

    BAX 1.253% 0.198% 1.055% (2.10)2 47

    AVP 0.104% 0.300% -0.196% (-0.26) 43

    HNZ 1.152% 0.062% 1.090% (2.70)1 50

    average(all 28) 0.825% 0.140% 0.685%

    average(1st 10) 1.291% 0.086% 1.204%average (last 10) 0.467% 0.161% 0.306%

    26

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    29/33

    Table 4: 3rd Friday Weekly Returns: One-Half Subperiod Results

    This table reports the comparable results from Tables 2 and 3 for one-half subperiod analysis for our

    sample of 28 large individual stocks. Panel A reports on the same analysis as for Table 2 for the 3rd

    -Fridaymean weekly returns. Panel B reports on the same analysis as for Table 3 for the 3 rd-Friday mean weekly

    returns for weeks with a relatively high option volume (high O/S ratio). The two subperiods run from

    January 1996 to June 2002 and from July 2002 to December 2008, with 339 weeks in each subperiod.

    Column four reports the spread in the mean returns between columns two and three, with the number

    in parenthesis, (), indicating the number of stocks with a positive spread, and the number in brackets, [],

    indicating the number of stocks where the spread is positive and statistically significant at a 10% p-value

    or better. As in Tables 2 and 3, the individual stocks are ranked from 1 to 28, based on each stocks O/S

    volume ratios 75th

    percentile value.

    Panel A: 3rd Friday Weekly Returns - Subperiods

    1. Ticker 2. Mean 3. Mean 4. Spread of

    3rd Friday Other Weeks Means

    (Col. 2 - Col 3.)

    First Half: Jan:1996 to Jun:2002

    average(all 28) 0.488% 0.241% 0.247% (19) [1]

    average(1st 10) 0.719% 0.175% 0.544% (9) [1]

    average (last 10) 0.351% 0.326% 0.025% (6) [0]

    Second Half: Jul:2002 to Dec:2008

    average(all 28) 0.412% -0.006% 0.418% (22) [4]

    average(1st 10) 0.368% 0.029% 0.339% (7) [3]

    average (last 10) 0.364% -0.062% 0.426% (9) [0]

    Panel B: 3rd Friday Weekly Returns with High Option Volume- Subperiods

    1. Ticker 2. Mean 3. Mean 4. Spread of

    3rd Friday Other Weeks Means

    High Opt. Vol. (Col. 2 - Col 3.)First Half: Jan:1996 to Jun:2002

    average(all 28) 0.913% 0.234% 0.679% (22) [6]

    average(1st 10) 1.248% 0.198% 1.050% (9) [3]

    average (last 10) 0.619% 0.302% 0.317% (7) [2]

    Second Half: Jul:2002 to Dec:2008

    average(all 28) 0.742% 0.035% 0.707% (23) [11]

    average(1st 10) 0.741% 0.053% 0.668% (9) [4]

    average (last 10) 0.446% -0.001% 0.448% (6) [3]

    27

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    30/33

    Table 5: Average 3

    rd

    Friday Weekly Returns for 28 Large Individual Stocks: Other Periods

    This table reports how the mean weekly returns ending on the 3rd Friday of a calendar month are

    different than other Friday weekly returns for a sample of large individual stocks. This table reports

    results for the following periods: 1983 to 2008, 1985 to 1995, 1996 to 2008, and 1948 to 1972. Over 1983

    to 2008, our sample includes the following 28 large individual stocks: AA, AVP, BA, BAX, BDK, BMY,

    DOW, EK, F, GE, GPS, HAL, HNZ, HON, HPQ, IBM, IP, JNJ, KO, MCD, MMM, MRK, OXY, PEP,

    SLB, TXN, WMT, and XRX (by ticker). Over 1948 to 1972, our sample includes the following 15 large

    firms, which are the firms from our 1983 to 2008 sample that have return data over the entire 1948 to 1972

    period: BA, BDK, BMY, DOW, EK, GE, HNZ, HON, IBM, IP, JNJ, KO, MMM, MRK, and PEP (by

    ticker). T-statistics are in parentheses, based on heteroskedastic- and autocorrelation-consistent standard

    errors. For row 6 (8), we report the number of the estimated coefficients that are positive (negative) and

    statistically significant at a 10% p-value, or better.

    1. 2. 3. 4.

    Sample Period: 1983 - 2008 1983 - 1995 1996 - 2008 1948 - 1972

    1. Average Weekly Return, 0.505% 0.559% 0.450% 0.341%

    3rd

    Friday Weeks

    2. Average Weekly Return, 0.206% 0.295% 0.118% 0.337%

    Other Weeks

    3. Average Spread, 0.298% 0.264% 0.333% 0.004%

    3rd Friday less Others

    4. Median Spread, 0.267% 0.254% 0.342% 0.039%

    3rd Friday less Others

    5. Positive Spreads 26 of 28 22 of 28 22 of 28 8 of 15

    6. Positive & Significant Spreads 8 of 28 4 of 28 7 of 28 0 of 15

    7. Negative Spreads 2 of 28 6 of 28 6 of 28 7 of 15

    8. Negative & Significant Spreads 0 of 28 0 of 28 0 of 28 0 of 15

    28

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    31/33

    Table 6: Option-Expiration Weekly Returns for Size-based Portfolios

    This table reports how the mean weekly returns ending on the 3rd Friday of a calendar month are

    different than other Friday weekly returns for size-based portfolios returns. The weekly returns are

    five-day Monday-through-Friday returns, ending on a Friday. To calculate the spreads, we estimate the

    same model as in Table ??. The size-based decile portfolios in rows 1 through 10 are the value-weighted

    portfolios from the French data library. In row 11, the [10 minus (Mean of 1 to 5)] is the difference

    between the weekly return of the largest decile-10 portfolio and the average of the weekly returns of

    the smallest five size-based deciles. The sample period is 1983 to 2008. T-statistics are in parentheses,

    based on heteroskedastic- and autocorrelation-consistent standard errors.

    Portfolios: 1. Mean 2. Mean 3. Spread 4. T-stat.

    Size Deciles of 3rd Friday of Other between Col. 1 on

    & Other Weekly Returns Weekly Returns & Col. 2 the Spread

    1. Decile-1 0.185% 0.168% 0.017% (0.13)

    2. Decile-2 0.234% 0.181% 0.054% (0.36)

    3. Decile-3 0.269% 0.192% 0.078% (0.50)

    4. Decile-4 0.259% 0.179% 0.080% (0.51)

    5. Decile-5 0.249% 0.202% 0.047% (0.30)

    6. Decile-6 0.254% 0.200% 0.054% (0.36)

    7. Decile-7 0.278% 0.208% 0.069% (0.46)

    8. Decile-8 0.280% 0.196% 0.085% (0.54)

    9. Decile-9 0.317% 0.195% 0.123% (0.82)

    10. Decile-10 0.417% 0.142% 0.275% (1.81)

    11. 10 minus 0.177% -0.042% 0.220% (2.09)

    (Mean of 1 to 5)

    29

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    32/33

    Table 7: 3rd Friday Weekly Returns: S&P 500 Index

    This table reports how the mean weekly returns ending on the 3 rd Friday of a calendar month are

    different from other weekly returns for the S&P 500 index. We report the mean return, standard

    deviation, and return spread for various sample periods. Panel A shows the main results from July

    1983 to December 2008. To contrast, we also include the earlier sample results from 1962 to 1983.

    Panel B report the results for two half-sample subperiods from July 1983 to April 1996 and April

    1996 to December 2008. Panel C shows the results after excluding the weeks during October 1987

    market crash and the September 11, 2001 terrorist attack. Panel D accounts for the turn of the

    month effect and the triple-witching weeks. Panel E reports the results conditional on positive or

    non-positive prior three-week returns. The results are shown in percentage form. The return spreads

    are calculated from equation (1). T-statistics are reported in parentheses.

    Panel A: S&P 500 Index Weekly Returns and Option Expirations

    1983 to 2008 1962 to 1983

    Mean Return Std Dev Spread Mean Return Std Dev Spread

    3rd Friday Weeks 0.3702 2.3415 0.2838 0.1121 1.9512 -0.0108

    (2.77) (1.94) (0.91) (-0.08)

    All Weeks 0.1517 2.2494 0.1204 1.9925

    (2.46) (2.00)

    Panel B: Subperiod Evidence

    July 1983 to April 1996 April 1996 to December 2008

    Mean Return Std Dev Spread Mean Return Std Dev Spread

    3rd Friday Weeks 0.3844 1.8807 0.2110 0.3560 2.7253 0.3568

    (2.53) (1.21) (1.62) (1.51)

    All Weeks 0.2219 1.8902 0.0813 2.5571

    (3.03) (0.82)

    30

  • 8/10/2019 Options - Stock Returns during Option Expiration Weeks and the Option-Stock Volume Ratio.pdf

    33/33

    Table 7: (continued)

    Panel C: 1987 Market Crash and September 11, 2001

    Excluding 1987 Market Crash Excluding September 11, 2001

    Mean Return Std Dev Spread Mean Return Std Dev Spread

    3rd Friday Weeks 0.3702 2.3415 0.2718 0.4094 2.2425 0.3236

    (2.77) (1.88) (3.19) (2.23)

    Full Sample 0.1609 2.2246 0.1601 2.2278

    (2.64) (2.62)

    Panel D: Third Friday Week Returns Conditional on Prior 3-Week Returns

    Positive Prior 3-Week Returns Non-Positive Prior 3-Week Returns

    Mean Return Std Dev %> 0 Mean Return Std Dev %> 0

    3rd Friday Weeks 0.5154 1.5992 64.34 0.2644 2.7546 55.93

    (3.66) (1.27)


Recommended