The driving forces of option trading: an empirical analysis on European data

Transcription

1 UNIVERSITEIT GENT FACULTEIT ECONOMIE EN BEDRIJFSKUNDE ACADEMIEJAAR The driving forces of option trading: an empirical analysis on European data Masterproef voorgedragen tot het bekomen van de graad van Master in de Toegepaste Economische Wetenschappen: Handelsingenieur Ineke de Boer onder leiding van Dr. Dries Heyman

2

3 UNIVERSITEIT GENT FACULTEIT ECONOMIE EN BEDRIJFSKUNDE ACADEMIEJAAR The driving forces of option trading: an empirical analysis on European data Masterproef voorgedragen tot het bekomen van de graad van Master in de Toegepaste Economische Wetenschappen: Handelsingenieur Ineke de Boer onder leiding van Dr. Dries Heyman

4

5 PERMISSION The author of this master thesis declares that the content of this document can be consulted and/or reproduced, but only if the source is mentioned. Ineke de Boer Ghent, 20 may

6 ACKNOWLEDGMENTS I would like to acknowledge the guidance of my promoter, Dries Heyman, throughout this year. Clear analysis, answering questions, selecting the right path forward allowed to efficiently come to results. I also would like to thank my mother, Line Alderweireldt, for her patience and her never ending continuous synergy, as well as for her extensive reading check both for minor spelling issues as for overall content. Sincere thanks for the special Word-expertise of Lien Deleu, who certainly contributed to the lay-out and graphics of this thesis. Furthermore, I appreciated my friends and family s support and relativation capabilities in a motivating surrounding. Finally, I sincerely hope that reading through this thesis evokes your appreciation for this important endeavour. 2

7 INDEX PERMISSION... 1 ACKNOWLEDGMENTS... 2 INDEX... 3 LIST OF ABBREVIATIONS... 5 LIST OF TABLES... 6 LIST OF FIGURES... 7 INTRODUCTION... 8 PART I: LITERATURE INTRODUCTION REDUNDANCY OF OPTIONS THE ADDED-VALUE OF THE OPTION MARKET Informational efficiency due to completeness of the market Price & volatility effect on underlying asset OPTION MARKET LIQUIDITY Order based liquidity Transaction based liquidity CONCLUSION PART II: EMPIRICAL RESEARCH OVERVIEW VARIABLES Option Liquidity measures Potential determinants of option liquidity Option specific elements Firm specific elements Macro-economic elements Sentiment measures Summary of the expected relations

8 3. DATA ANALYSIS Methodology Summary statistics Liquidity measures Summary statistics of potential determinants of option liquidity Empirical evidence on determinant selection Unconditional correlations Endogeinity of stock return volatility Conditional relations Final set of relevant variables Model estimation: regression results Analysis beyond the model Are option markets growing? Year to year analysis CONCLUSION Overall conclusion Imperfections of the empirical research Suggestions for further research PART III : APPENDICES... I APPENDIX I: VARIABLES DESCRIPTION... II A. Liquidity measures... II B. Independent variables: option specific elements... III C. Independent variables: firm specific elements... IV D. Independent variables: market specific measures... VI E. Independent variables: sentiment measures... VII APPENDIX II: TABLES... VIII List of Tables... VIII PART IV: REFERENCES... XXI LIST OF USED LITERATURE... XXII 4

9 LIST OF ABBREVIATIONS C CC CDZ CORV COVC COVV CPI CS DOVC DOVV DSV DY DZ G G IS MV N OM ORV OVC OVV PDZ PORV POVC POVV PSR PYD ROA SR SRV SV SZ T UK UP ZEW Constant term Consumer confidence Call option delta Call option return volatility Daily call option volume in number of contracts Daily call option value in Euro Inflation Common shares outstanding Daily change in option value in Euro Daily change in option volume in number of contracts Daily change in stock volume Dividend yield Option delta Growth Germany Investor sentiment Market volatility The Netherlands The nearest to the money option price Option return volatility Daily option volume in number of contracts Daily option value in Euro Put option delta Put option return volatility Daily put option volume in number of contracts Daily put option value in Euro Past stock returns Dividend payment date Return on assets Present stock return Stock return volatility Daily stock volume Firm size Linear time trend United Kingdom Underlying stock price Analysts forecast 5

10 LIST OF TABLES TABLE 1: SUMMARY OF THE EXPECTED RELATIONS TABLE 2: SUMMARY OF THE DATA QUALITY TABLE 3: SUMMARY STATISTICS OPTION LIQUIDITY MEASURES TABLE 4: CORRELATION MATRIX OF LIQUIDITY MEASURES TABLE 5: SUMMARY STATISTICS POTENTIAL SET OF DETERMINANTS TABLE 6: EXTRACT FROM THE PEARSON CORRELATION MATRIX IN TABLE C, APPENDIX II TABLE 7: TEST FOR THE ENDOGENEITY OF STOCK RETURN VOLATILITY TABLE 8: RELEVANT DETERMINANT SELECTION, CONDITIONAL RELATIONS FOR OPTION VOLUME TABLE 9: RELEVANT DETERMINANT SELECTION, CONDITIONAL RELATIONS FOR OPTION VALUE IN EUROS TABLE 10: SCORING OF POTENTIAL DETERMINANTS TABLE 11: THE FINAL SET OF RELEVANT DETERMINANTS TABLE 12: MODEL ESTIMATION WITH FINAL SET OF RELEVANT VARIABLES TABLE 13: MODEL ESTIMATION INCLUDING TIME TREND TABLE 14: THE RELEVANT DRIVING FORCES OF OPTION LIQUIDITY TABLE 15: OVERVIEW OF THE MODEL ESTIMATION RESULTS

11 LIST OF FIGURES FIGURE 1: SUMMARY OF THE ADDED-VALUE OF EQUITY OPTIONS FIGURE 2: SUMMARY OF THE EXISTING LITERATURE ON OPTION MARKET LIQUIDITY FIGURE 3: OVERVIEW OF THE SET OF POTENTIAL DETERMINANTS OF OPTION LIQUIDITY FIGURE 4: OPTION VOLUME AND VALUE OVER TIME PER COUNTRY FIGURE 5: THE SIGNIFICANCE OF COEFFICIENTS OF STOCK RETURN VOLATILITY FOR LAGGED AND ANTICIPATED VALUES FIGURE 7: THE SIGNIFICANCE OF PRESENT RETURN COEFFICIENT ESTIMATES FOR CALL OPTION VOLUME FIGURE 8: THE SIGNIFICANCE OF PAST STOCK RETURN COEFFICIENT ESTIMATES FOR CALL OPTION VOLUME

12 INTRODUCTION Equity options are derivative financial products that entitle a buyer to buy or sell a stock on a predetermined date range for a predetermined price. Equity options are used for three objectives: to hedge an existing position in stocks, to speculate on a change in the underlying stock price and to speculate on the change in volatility of the underlying stock price (Birkner). In Europe most options are traded on LIFFE CONNECT, which is known as the world leading electronic derivatives trading system since The platform was developed in by the London International Financial Futures and Options Exchange (LIFFE) to replace its open outcry trading floor (Euronext 2004). In 2001 LIFFE was bought by Euronext. In 2003 the Brussels and Paris market transferred their derivatives business to the Euronext, and in 2004 Amsterdam and Lisbon markets followed. Alltogether leading to a single European market for derivatives (Euronext 2004). This rapid growth of the European option market requests further research regarding the fundaments of option trading activity in Europe. The last decade the popularity of option trading has grown, both in number of underlying assets as when measured in the trading volume. Obtaining a clear idea on the option market dynamism and more specifically its liquidity is highly desirable. Although the rising importance of option trades, little research is conducted on the determining elements of the option market liquidity. The purpose of this thesis is to analyse which variables are significantly affecting the European option market liquidity. The thesis contains two main chapters, the first chapter gives an overview of the literature written about option liquidity. The second chapter explains the empirical analysis conducted. The literature overview respectively discusses the potential redundancy of options, the actual added value of options and finally the few papers that do discuss the determinants of option liquidity in some way. 8

13 The empirical analysis contains five parts. First, the list of variables that might affect option liquidity as extracted from the literature are defined and the expected relation with option liquidity is clarified. Second, summary statistics of both the option liquidity measures as the set of potential variables are provided. Third, the potential list of variables is reduced to a final set of relevant variables by analysing the significance of the unconditional and conditional relations with option liquidity measures. Fourth, based on the final list of variables an analysis is conducted to assess in which direction and to what extent the variables determine the option liquidity. Last, this final regression is checked for robustness and model specification. 9

14 PART I: LITERATURE 10

15 1. INTRODUCTION This extensive examination of the literature summarizes the work written on the existence and added-value of equity options in the financial market. Three main themes are discussed. The first chapter describes the circumstances for which options appear to be redundant. Since those circumstances are not valid for the real financial market options are not redundant in reality. Therefore, the added-value and the impact of option trading activity is discussed in the second chapter, a separation is made between the impact of option introduction on the informational efficiency and the impact of option trade on the underlying stock market. A third chapter summarizes the few papers that actually did research upon the determinants of the option market liquidity; the research on order based liquidity and transaction based liquidity is treated separately. The last chapter contains the main findings and contradictions of the literature on option trading activity in several representative figures. 11

16 2. REDUNDANCY OF OPTIONS The pricing model of Black and Scholes assumes that an option is redundant because options can be constructed by other corporate liabilities, therefore options can be valued with noarbitrage arguments. They argue that under ideal conditions, assuming a complete competitive and frictionless world and if options are correctly priced that obtaining certain profits by using arbitrage is impossible. Furthermore Black and Scholes pricing model tells that in an ideal, complete market the underlying stock price follows a stochastic process, a geometric Brownian motion, and does not depend on the introduction of options. This imposed the theory that options are redundant and do not influence the prices of underlying asset or their allocation (Black and Scholes 1973). However, their empirical tests on call options revealed that in reality option buyers pay option prices that are constantly higher than the price obtained by their pricing model while option writers pay more or less the option price obtained by using non-arbitrage arguments. This phenomenon is exemplified with the existence of large transaction costs paid by the option buyers (Black and Scholes 1972). Connolly confirmed the correctness of the Black and Scholes model, he declared options on currency futures as redundant because the future market is already gifted with an ample range of derivatives (Connolly 1996). 12

17 3. THE ADDED-VALUE OF THE OPTION MARKET 3.1 INFORMATIONAL EFFICIENCY DUE TO COMPLETENESS OF THE MARKET Under the assumption that the creation of derivatives is not too expensive, Ross argues that introducing options can complete the market because this triggers the expansion of the opportunity set available to investors, and thus increases the market efficiency. He states that the market exists as a number of states but that there are more states than available assets. To create an efficient competitive equilibrium the remaining states can be constructed less costly with options (Ross 1976). Aliprantis and Tourky come across with another motivation to relax the assumptions in the Black and Scholes pricing model, they find that if the number of securities is smaller than half the number of states not a single option can be replicated by traded securities. Furthermore, they state that incompleteness of security markets affects portfolio choices, the more incomplete the market the more popular hedging strategies like super-replication of contingent claims (Aliprantis and Tourky 2002). If markets are incomplete, financial innovation affects the information flow in the market, this will be reflected in the characteristics of new derivatives contracts (Strike price etc.). This is only true if there is enough diversity among traders, because then derivative and asset markets interact. This results in the dependency of the security prices on the characteristics of the option contract (Detemple and Selden 1991). Anthony was the first to prove that the option volume is an indication for the information arrival rate (Anthony 1988). Cao states that the presence of options enhances the incentive and the ability to trade on private information about the underlying asset, and thus attracts more informed traders. This boost of informed trading results in a more informative price and thus a higher level of informational efficiency (Cao 1999). That informed traders are attracted after an option introduction is confirmed by Danielsen and Sorescu, they state that 13

18 the disappearance of short-sales constraints in comparison with the security market leads to an increase of the level of short interest in stocks. (Danielsen and Sorescu 2001). Chan and Donald found evidence for the improved feedback between the spot and the currency futures market after option introduction, by examining the Geweke feedback measure before and after the option introduction. They conclude in their paper that options play a leading role for information transmission, most likely due to their low transaction cost (Chan and Donald 2006). There is no consensus concerning the direction of the information flow between the stock and the option marked, it is not clear where informed investors prefer to trade. Anthony was pioneering when he proved with call option data that option volume was leading stock volume in most of the cases (Anthony 1988). Chen et al. expand the research concerning the information flow between equity and option markets by examining the relation between stock returns and option trading value ratios 1. They are the first to control for the variables moneyness, information asymmetry and option liquidity. They discover the leading role of stock returns with respect to the option value ratio of OTM options, this relation has not been revealed with the other two types of moneyness. This is consistent with Pan and Poteshmans earlier finding that the effect of option volume on stock prices is higher with a higher degree of leverage (Pan and Poteshman 2006). Chen et al. suggest that OTM options attract informed investors because of their higher liquidity, lower premiums and higher delta to premium ratios. The revealed feedback relation is stronger with a higher informational asymmetry. For more liquid options this causality reverses, in this case they find that OTM option trading value directs stock returns. (Chen, Lung, and Tay 2004). 1 Option trading value ratio = (call volume x call premium)/(put volume x put premium). This measure can be seen as a measure for information embedded in option trades. (Chen, Lung, and Tay 2004) 14

19 Easley et al. find that informed traders trade in both markets if the leverage of options is high, the stock market is not that liquid and the concentration of informed traders is high. Under those three conditions, the option volume will drive the underlying stocks. If those prerequisites are not valid, informed traders only trade in the stock market (Easley, O'Hara, and Srinivas 1998). Pan and Poteshman work deepens this issue out by testing the validity of the assumptions designed by Easley et al. They find that option trading volume ratios are good predictors of future stock returns. They argue that the level of predictability is stimulated by non-public information brought by informed investors and by the leverage of the option. This predictability fades away with time, resulting in the fact that the information carried by option volume is incorporated in the underlying stock prices. (Pan and Poteshman 2006). Consistent with Pan and Poteshman, Cao and Wei discover a higher level of information asymmetry in option markets in comparison with the underlying asset markets, indicating that informed traders direct themselves more to the option market. (Cao and Wei 2010). 15

20 3.2 PRICE & VOLATILITY EFFECT ON UNDERLYING ASSET A higher level of informational efficiency leads to changes in price and volatility of the underlying security. However, major divergence exists in both the direction as in the reason of change in price and volatility of the underlying security. Both Detemple and Selden as Cao conclude that the introduction of derivatives for a certain asset increase the expected price of the underlying asset and decrease the volatility of the stock returns. However, they believe that this effect is initiated by following mechanisms. Detemple and Selden argue that if investors disagree on the volatility of future prices and agree on the expected value of the future prices, then the introduction of options increases the demand of the stock, thus leading to an augmentation of the stock prices and a decline of the volatility of stock returns. The aggregation of those effects leads to a more stabilized market after option introduction. This dynamic is proven with Detemple and Selden s model describing the interaction of two investors with a different mind-set of the market: one focuses on speculation and thus sees both markets as substitutes, and the other focuses on hedging and thus sees the option market as a complement of the stock market. The more heterogeneous the beliefs of those investors, the larger the interaction between the option and the stock market resulting in a more stabilized market (Detemple and Selden 1991). Cao put forward another reason that drives this process: he believes that the increase in the stock price is indirectly driven by the asymmetric information and the incentive to collect information after an option introduction (Cao 1999). However Cao s study reveals that the reduction of stock return volatility is the most influenced by the earliest option introductions. As the market becomes more and more complete, additional new option introductions will have less effect on the underlying asset, since the price is more informative and thus less sensitive to a non informational supply shock such as earnings announcements. This fading phenomenon is confirmed by Detemple and Jorion (Detemple and Jorion 1990). 16

21 The alternate stream of opinions states that the introduction of options on the price is negative. Danielsen and Sorescu examine the effect of the disappearance of short-sales constraints because of an option introduction, on the price of the underlying security. They conclude that the removal of short-sales constraints increases underlying short interest and diminish the underlying stock price. (Danielsen and Sorescu 2001) Faff and Hillier conclude that no predictable pattern can be determined in the stock price after option introduction. They state that the direction of the change in the stock price depends on the expected future returns of the underlying asset. They assume that option introduction drives informed traders to the option market because of the higher leverage 2. This assumed migration of informed traders decreases the level of informational asymmetry in the asset market and increases the level of asymmetry in the option market. This evolution in information asymmetry and the expected future value of the stocks drive the change in stock price and result in unpredictable patterns of abnormal returns after an option introduction (Faff and Hillier 2005). Roll, Schwartz and Subrahmanyam examined the effect of option trading on firm valuation. The increased level of informational efficiency in the stock market (because of option introduction) leads to a more informative price of the shares of a company. Thus option introduction positively affects the firm value, measured by Tobin s Q. This effect was more explicit for situations where information asymmetry is higher, because then the likelyness to produce public information based on investment analysis is higher. The impact of option trading on firm valuation was negligible if option trading was only driven by speculation objectives or for less active option markets (Roll, Schwartz, and Subrahmanyam 2009). 2 As shown in the earlier paragraph there is no consensus about the validity of this general assumption. Easley et al. find that informed traders can trade in both option as the stock market if leverage is high, liquidity of the stock market is low and if the concentration of informed traders is high; in the other case they state that informed traders stay in the stock market (Easley, O'Hara, and Srinivas 1998). 17

22 Focusing on the effect of options on the volatility of the underlying assets, several opinions are observed throughout the literature. In an incomplete market the introduction of options may stabilize the asset market because of more efficient risk allocation or the reverse because of an increased interest in speculation. Back reveals that an unintroduced option can be redundant, but once the option is introduced in the market, it is not redundant anymore because its price contains information about the value of the underlying asset. He concludes that options are difficult to price following the Black and Scholes pricing model because of the existence of asymmetric information. The model shows that the richer information signals released by the introduction of options are asymmetrically received by the market. This asymmetry leads to the stochastic volatility of the underlying asset implying that the option price doesn t equal the Black and Sholes price, because a stochastic volatility of the underlying makes the construction of synthetic options impossible (Back 1993). In an attempt to answer the puzzle whether option introduction stabilizes the asset market, Hwang and Satchell prove that the fundamental volatility 3 decreases after a European option introduction. This decrease in fundamental volatility is caused by the stochastic information arrival and stabilizes the market. (Hwang and Satchell 2000). 3 Volatility is defined as the sum of the transitory noise and the unobserved fundamental volatility (Hwang and Satchell 2000) 18

23 4. OPTION MARKET LIQUIDITY The literature about option market liquidity is rather thin, despite the rising importance of option trading. Kalodera and Schlag contribute to the literature by making a clear difference between transaction based liquidity measures and order based liquidity measures. Transaction based liquidity 4 is related to volume and trading frequency while order based liquidity refers to spreads and depth. This clear distinction is used throughout the further research work represented in this document. Also, several papers exist that encapsulate both liquidity measures. Cao and Wei found strong evidence that the lower the size of the firm and the higher the stock return volatility, the higher the degree of commonality between the different liquidity measures. When screening the different option liquidity measures (bid-ask spread, proportional bid-ask spread, option volume in shares and in Euros) they revealed the highest correlation between option volume in shares and in Euros. However they proved the existence of a negative and quite low correlation for the percentage bid-ask spread and volume (Cao and Wei 2010). George and Longstaff examine the cross-sectional distribution of bid-ask spreads and trading activity for S&P 100 index options, across exercise prices and maturities. They find that cross-sectional differences are linked to market-making costs and trading frequency. They study the determinants of the spread and the relation between spread and trading activity, measured by the number of trades and trading volume. Their paper reveals that trading activity and spreads are jointly determined and inversely related to each other. By examining the relation between spreads and trading activity across options they conclude that calls and puts with the same strike price and the same time to maturity are seen as close substitutes (George and Longstaff 1993). 4 Frequently named trading activity in literature 19

24 4.1 ORDER BASED LIQUIDITY The literature around order based option market liquidity measures is more extensive than transaction based liquidity. However, the finding of both Cao and Wei as George and Longstaff that bid-ask spread and option volume show an inverse relation and are instantaneously determined, makes it worthwhile to study the determinants discovered in the papers covering order based liquidity measures. Using a sample of Australian options written on only nine shares, Pinder examines the determinants of the bid-ask spread. He uncovers a positive relation between the bid-ask spread and the option value, time to maturity, options delta and the stock return volatility. A negative relation is calculated between the bid-ask spread and option trading activity (Pinder 2003). Chong et al. discover a negative relation between time to maturity and the bid-ask spread for ATM over the counter currency options after controlling for price risk, competition and trading activity. They reason that a higher theta and gamma risk for shorter time to maturity options causes this inversion (Chong, Ding, and Tan 2003). Wei and Zheng refine the research on option liquidity by studying the proportional bid-ask spread 5 as a liquidity measure on a sample of equity options. They contribute with three innovative findings. First, they stress the importance of option return volatility 6 in explaining the spread differences. Second, they discover that traders use short term options as substitutes for midterm options when the latter are not available due to expiration cycles, this increases the option volume in an inorganic way. Last, they note that more OTM options are traded when stock return volatility increases (Wei and Zheng 2010). 5 Proportional bid-ask spread = spread/average bid-ask quotes. The advantage of using the proportional bid-ask spread is that it eliminates the bias incorporated by the higher market-making cost associated with more leveraged options (Wei and Zheng 2010) 6 Option return volatility = option price elasticity * stock return volatility (Wei and Zheng 2010) 20

25 4.2 TRANSACTION BASED LIQUIDITY Mayhew, Sarin and Shastri are the first to investigate the impact of the stock market on the order based liquidity of the option market. They examine option liquidity expressed as trading volume, i.e. the number of option transactions and the dollar value of option trading volume. They find a positive and significant impact of stock price and stock volume on option liquidity, they are not able to prove a significant impact of stock return volatility on the option liquidity. Also, the percentage ratio of the stock s bid-ask spread over market depth negatively influences option liquidity (Mayhew, Sarin, and Shastri 1999). However the methodology used is heavily criticised in the literature 7. Kalodera and Schlag extend Mayhew, Sarin and Shastri s work by using a small sample of German DAX index stock and options. Their empirical analysis regresses stock volume, stock return volatility, stock returns and their lagged values on different measures of option liquidity. The six withheld measures of option liquidity are: the number of option contracts traded, the number of option transactions, option volume measured in Euros, option spread, the option spread in percent of the midpoint quote and the depth. They are the first to reveal clear differentiation between the reactive behaviour of calls and puts on the stock market activity. They note a significant and positive relation between stock volume, present and past stock returns and option liquidity. Furthermore, they reject the intuitive idea that stock return volatility has a significant impact on option liquidity, not being consistent with the earlier finding of Mayhew, Sarin and Shastri. In addition Kalodera and Schlagg find a higher explanatory power for the regressions of transactions based liquidity versus the regressions of order based liquidity (Kalodera and Schlag 2004). 7 First, they run cross-sectional regressions but average the coefficients across time so that time-series variation in trading activity is smoothed out. Second, they suggest differences across companies and do not introduce this in their regressions. And last, they do not run different regressions for calls and puts. (Kalodera and Schlag 2004) 21

26 Lakonishok et al. investigate option trading behaviour among different classes of investors. They find that option market activity measured by purchased and written call and put option volume is positively related to past stock returns, stock return volatility and negatively related with book-to-market ratio which is a measure of growth of the firm. Evidence is provided that option market volume is driven by speculation and hedging of the direction of the underlying stock price and not by volatility traders. They investigate the option activity during the stock market bubble of and find that sophisticated investors do not change their behaviour while less sophisticated investors significantly increased their positions in options, especially on growth stocks during that period. Furthermore they illustrate that on average there is more written than purchased open interest. This is due to the more sophisticated traders who hold their written positions longer than their purchased positions, most likely because the purchased positions are motivated by short-term speculation (Lakonishok et al. 2007). In their more recent empirical analysis Roll, Schwartz and Subrahmanyam investigate the trading activity of options relative to the stock trading volume. Using a sample of almost 3000 trading days they find a strong cross-sectional relation between the options/stock trading volume ratio and determinants firm size, implied volatility, delta, trading cost, institutional holdings, earnings announcements. Furthermore, analyzing the time series of the option/stock trading volume ratio they notice day to day swings, possibly indicating that on some days traders seek to exploit private information. After conducting an event study they conclude that option/stock trading volume ratio significantly increases a few days around an earnings announcement. Analysis of the cumulative abnormal return suggests that option trading activity leads to a higher degree of market efficiency (Roll, Schwartz, and Subrahmanyam 2010). 22

27 5. CONCLUSION A summary of the main findings of the literature review of is given below. Figure 1 reveals that in an incomplete market option trading activity has an impact on the financial market. Literature agrees that option introduction increases informational efficiency due to the increased completeness of the market. Concerning the impact of option trading activity on the stock market, the literature is not able to find a consensus. Figure 1: Summary of the added-value of equity options Are financial markets complete? YES NO Options are redundant Options are not redundant Stock price Effect on the stock market No consensus about: Stock volatility The direction of the influence The reason of the relation Whether stock markets lead option markets or the other way around Options complete the market: opportunity set of available for investors Incentive to trade on private information Information flow Market efficiency An overview of the literature on option market liquidity is provided in figure 2. A distinction is made between order and transaction based liquidity measures, between which a high degree of commonality exists. For both classes an evolution of the research is provided. Each researcher clearly builds on the work of their predecessors by adding an innovative element to the research. 23

28 Figure 2: Summary of the existing literature on option market liquidity Option market liquidity Order based liquidity: Spreads Market depth Research evolution: 2003: First research on bid-ask spreads 2004: put and call options are treated seperatly in the analysis of bid-ask spreads and market depth 2010: The proportional bid-ask spread and its determinants is examined Degree of commonality Transaction based liquidity: Volume Transaction frequency Research evolution: 1999: First study on the impact of the stock market 2004: put and call options are treated seperatly 2007: behaviour of different classes of investors is examined 2010: the option/stock volume ratio and its determinants is studied The examination of the literature revealed several important elements. First, the research on the impact of options on the financial market and the option liquidity is very thin. Second, the literature does not find a consensus on all aspects. Third, there remain many black holes in the literature on option market liquidity, such as: what the relevant determinants of option market liquidity in a European context are. This specific topic was consciously chosen as the key theme for the second part of this work: transaction-based option liquidity measures in a European context. 24

29 PART II: EMPIRICAL RESEARCH 25

30 1. OVERVIEW The scope of the empirical research is to examine if the variables extracted from the literature are relevant to explain the option market liquidity and to determine the direction with which the relevant variables influence the option market liquidity. The empirical analysis builds through five key parts. In the first part the liquidity measures and a potential set of determinants is defined together with the expected relation between the determinants and the liquidity measures. The second part documents the basis of the data sample together with the quality of the data. The third part contains the overall analysis. In this chapter the methodology is clarified. General summary statistics of both the liquidity measures as the potential drivers are provided. The relevant determinants are extracted from the list of potential determinants. For this set of relevant variables the direction and the significance of the relation with option liquidity measures is examined. The fourth part of the empirical analysis contains analyzes the final model for growth trends and for stability of the coefficient estimates over time. The fifth part of the empirical analysis summarises and concludes on the relevant findings and suggests potential research extensions referring to this interesting but yet little explored topic. 26

31 2. VARIABLES This paragraph zooms in on the selection process for the main relevant drivers used in the later analysis together with their expected relation with the independent variables. Note that only the variables for which data was available are further examined, other variables that were mentioned in the literature are out of scope in this empirical analysis. First the option liquidity measures are defined. In a second step, the potential set of option liquidity drivers are defined and the expected relation with option liquidity is clarified. These potential determinants are classified into four groups with respective characteristics: option specific elements, firm specific element, market specific elements and sentiment measures. The groups are reflected in figure 3. Figure 3: Overview of the set of potential determinants of option liquidity A summary of the expected relation between the potential set of determining variables and independent variables is given in the conclusion of this paragraph. 27

32 2.1 OPTION LIQUIDITY MEASURES The research is based upon two liquidity measures: the Daily option volume in number of contracts (OVC) of a lot size of options per contract and the Daily option value in Euros 9 (OVV) constructed by the multiplication of the option volume in number of contracts with the option market price. Put and call option series are treated separately in this analysis, as different behaviours occur for calls and puts. First, call and put option series can be seen as close substitutes (George and Longstaff 1993). Second, several researchers find that calls dominate puts, meaning that the open interest in calls is more than twice as large as the put open interest (Lakonishok et al. 2007; Kalodera and Schlag 2004). To examine these two statements a differentiation is imperative between call and put option series. The used option price is the option price of the option series closest to the money, this eliminates outliers and is relevant because among other Lakonishok et al. find that option volume is concentrated in near-the-money options (Lakonishok et al. 2007). To reduce the effect of potential outliers the square root of the independent variable is used, however for convenience there will be referred to the squared variables as OVC and OVV. 8 Since the English option contracts have a lot size of 1000 options per contract, those volumes are multiplied by 10 to make comparison possible. 9 The British option price is brought back to Euros using the European Central Bank reference exchange rate. 28

33 2.2 POTENTIAL DETERMINANTS OF OPTION LIQUIDITY In this section potential determinants of the option liquidity are preselected. Respectively option specific elements, firm specific elements, macro-economic elements and sentiment measures are described. Option specific elements evolve with the type of option series. Firm specific elements relate to the situation of the specific ticker symbol. Macro-economic elements represent the macro-economic climate where the firms operate in. Sentiment measures summarize the elements that might stimulate irrational trading behaviour. Note that only the variables for which actual data was found are examined, the literature however depicts more variables that fall out of scope of this empirical research. For a more technical and mathematical definition of the considered variables refer to Appendix I OPTION SPECIFIC ELEMENTS The delta (DZ) of an option series is included as an approximation for the hedging-related demand. A higher call option delta indicates more sensitivity to changes in the price of the underlying asset. If the call delta decreases, more call option contracts per share have to be bought to create a delta-neutral portfolio (Hull 2008). Hence, a downward relation is expected between call delta and call option volume in number of contracts. Since delta of put options is always negative the inverse relation can be expected between put option volume in number of contracts and put delta. The relation with call option value in Euros is ambiguous since lower delta firms have lower option prices. For put options the inverse is true. Roll et al. find that the call (put) option delta is strongly negatively (positively) influencing the option/stock volume ratio, the impact on the dollar value of the option/stock volume ratio is less negative and even turns positive in some cases. (Roll, Schwartz, and Subrahmanyam 2010). George and Longstaff find that the delta has a negative and significant impact on the bid-ask spread, which is inconsistent with the hypothesis that options with greater deltas impose greater inventory risk on market makers. In fact, spreads are greater for options that are less risky, thus having a lower delta. A potential explanation 29

34 is that options with the smallest deltas are the options with returns more sensitive to changes in the underlying asset (George and Longstaff 1993). Note that the delta used in the empirical research is the option price elasticity, this is a good construct to approximate the option delta derived from the Black and Scholes pricing model. Thus, similar relations with option market liquidity can be expected using this construct. Option return volatility (ORV) is defined as the multiplication of the absolute value of the option price elasticity and stock return volatility. The choice to include the option return volatility over the implied volatility is made, because the implied volatility represents the anticipated future volatility which is not likely to be jointly determined with the current option trading activity, while the option return volatility is. An increase in option return volatility can stimulate two effects: on one hand, a higher volatility leads to a higher option price, thus to a higher option value. On the other hand, a higher volatility may attract more informed traders, because the more volatile an option the higher the potential leverage when trading upon their private information (Roll, Schwartz, and Subrahmanyam 2010). Taking these two effects into account a positive relation is expected between option return volatility and both option liquidity measures. The importance of option return volatility in the explanation of option liquidity is illustrated by Wei and Zheng who uncover that this determinant has a higher power in explaining the bid-ask spread difference than common liquidity determinants such as stock return volatility. They conclude that option return volatility explains over 45% of the proportional bid-ask spreads, indicating that it is the most significant determinant of option liquidity measured by the proportional bid-ask spread. In one of their robustness checks option return volatility is replaced by implied volatility in the regression, in all their regressions different volatility measures result in similar levels of statistical significance (Wei and Zheng 2010). 30

35 Roll et al. find a strong positive relation between implied volatility and the option/stock volume ratio. They notice that the t-statistics are larger for dollar value of the option/stock volume ratio. This finding can be brought back to the close connection between implied volatility and option prices (Roll, Schwartz, and Subrahmanyam 2010). In conclusion a positive relation can be expected between option volume and option return volatility, this effect is supposed to be higher and more significant for option value than for option volume FIRM SPECIFIC ELEMENTS Literature points out that option liquidity varies considerably across firms (Mayhew, Sarin, and Shastri 1999; Kalodera and Schlag 2004; Lakonishok et al. 2007). Therefore it is reasonable to include several characteristics of firms and their stocks in the set of relevant variables. The Growth of a firm (G) as measured by the annual growth of the book value per share, is incorporated to distinguish between growth stocks and options on value stocks. High growth stocks have a higher call option trading volume, because of their higher future leverage potential. Logically firms that have a high growth have a higher future stock price (thus a lower Book-to-market value). If traders want to capture this stock price augment they have to buy call options. On the other hand, stocks with a low or even negative growth will have a more stable or smaller future stock price, capturing this decrease in stock price can easily be done with the purchase of put options. So we would expect a negative relation between annual growth of the book value per share and call option trading activity and a positive relation for put option trading activity. Lakonishok et al. find that more but smaller option positions are opened on growth stocks. Furthermore they find little difference between growth stocks and value stocks while evaluating the option volume, with exception of the stock bubble of ; during that period the option activity on growth stocks increased. They believe that this effect is due to 31

36 the speculation of less sophisticated investors who speculate on the continuing rise of the stock price (Lakonishok et al. 2007). Indicating that growth stocks attract more traders driven by speculation and stable stocks attract more traders driven by hedging purposes. Return on assets (ROA) of a firm is included as a measure of profitability and management effectiveness. A higher ROA can imply that the firm is in a mature phase, having more favourable investment opportunities (Roll, Schwartz, and Subrahmanyam 2009). Higher ROA means that the management is more effective in transforming liabilities into stock leverage. This augment in leverage attracts traders, both traders in the underlying stock market as traders in the derivative market. Thus one can expect a positive relation between ROA and option liquidity. The stock volume (SV) is included as a determining variable, the literature agrees upon a positive significant relation between stock volume and option volume. The literature does not depict any strong significant relation between stock volume and option value, however expectations are for a positive relation. Two effects can take place: first, while speculating stocks and options can be seen as substitutes resulting in a negative relation between stock and option volume. The reason why options may replace stocks is that one can easily replicate the stock by trading in options and a risk-free investment. This trading strategy comes along with lower transaction costs and the avoidance of the short-sales constrains in the option market (Danielsen and Sorescu 2001). Secondly, a stock volume increase encourages an increase in hedging-related option demand. The discussion whether options and stocks are substitutes or complements is clarified by Kalodera and Shlagg who reveal a significant and strongly positive relation between stock and option volume indicating that the second effect is larger than the first. However they are not able to find a clear relation between stock volume and the option value in Euros 32

37 (Kalodera and Schlag 2004). This finding is consistent with Mayhew et al. who uncover that options are more liquid if the stock volume is higher (Mayhew, Sarin, and Shastri 1999). These conclusions imply that traders who aim at hedging have a higher impact on the option market, a positive relation is expected between stock volume and option market liquidity. Stock return volatility (SRV) is mentioned in the literature as a potential explanatory variable. However the literature is disagreeing upon the direction of its impact on option liquidity. A minority of researchers defend a negative relation: if the stock return volatility decreases, the hedging-related demand is lower, thus option volume will decrease (I.-Duon and Jueh-Neng 2009). The majority of researchers defend a positive relation: when stock return volatility increases the tendency to speculate on the volatility increases, stimulating the option volume of OTM options (Wei and Zheng 2010; Lakonishok et al. 2007; Mayhew, Sarin, and Shastri 1999). Although Lakonishok et al. find that strategies like straddles and strangles, which are the most common ways to speculate on the stock return volatility, only account for a small fraction of the option trading activity (Lakonishok et al. 2007). A small group of researchers find no consistent significant impact between volatility of the underlying asset and option trading activity (Kalodera and Schlag 2004). This inconsistency reveals that a relation between stock return volatility is proven but only under certain conditions. This dependency is further examined in this research. A teasing question is whether derivative trading activity stabilizes the stock market, and thus decreases the stock return volatility. Hwang and Satchell decompose 4 FTSE100 stock index related volatilities into transitory noise and unobserved fundamental volatility. They uncover that the introduction of options reduces fundamental volatility measured by the stock return volatility, indicating that option volume drives the stock return volatility and not the other way around. (Hwang and Satchell 2000). To conclude, the introduction of stock return volatility as an explanatory variable may induce problems of endogeneity. This variable is therefore treated in a separate section to test for endogeneity. 33

38 Dividend yield (DY) is a potential explanatory variable. Dividends reduce the stock price on the ex-dividend day, the higher the dividend the higher the stock price reduction. The call option value will be negatively affected by high dividends and thus a lower ex-dividend stock price and the put option value will be positively affected by high dividend yields. (Hull 2008). The relation with option volume in number of shares is ambiguous since high dividend firms may be larger and thus have a higher option volume. Lakonishok et al. illustrate a positive relation between dividend yield and call option trading volume. They include the interaction term dividend yield multiplied by the dividend payment date in their regression and find that there is less activity in high dividend stocks shortly before the dividend date in comparison with low dividend stocks, however this negative relation is not always significant (Lakonishok et al. 2007). Dividend payment date (PYD) represents the date of payment of a cash income dividend payment. For this variable the same behaviour as the dividend yield (DY) is expected for the above mentioned reasons (Lakonishok et al. 2007). This variable complements the dividend yield (DY) variable by bringing the timing of the dividends to the set of potential determinants, whereas the variable dividend yield (DY) focuses on the amount of the dividends. The firm size (SZ) expressed as market capitalisation is included in order to control for firm size effects. A positive relation is expected, since the larger the firm the more liquid the option market and the more potential informed trading (Roll, Schwartz, and Subrahmanyam 2010). In contradiction, Mayhew et al. define firm size as the total book value of the assets and find that options are more liquid for smaller firms (Mayhew, Sarin, and Shastri 1999). 34