FINANCIAL VOLATILITY AND DERIVATIVES PRODUCTS: A BIDIRECTIONAL RELATIONSHIP

Transcription

1 FINANCIAL VOLATILITY AND DERIVATIVES PRODUCTS: A BIDIRECTIONAL RELATIONSHIP Claudiu Tiberiu ALBULESCU Politehnia University of Timişoara Timisoara, Romania laudiu.albulesu@t.upt.ro Daniel GOYEAU University of Poitiers Poitiers, Frane daniel.goyeau@univ-poitiers.fr Abstrat This paper studies the dynamis of the relationship between the volume of transations with derivative produts and pries volatility of their underlying asset. This relation was widely approahed, but mainly from the perspetive of the impat of derivative produts on the volatility of their underlying assets. The fat that hedging as well as speulative operations with derivative produts are based on the prie volatility of their underlying asset leaves a priori room to the idea aording to whih the volume of ativity related to derivative produts has to follow in a unidiretional manner the prie volatility of the underlying assets. However, more reently, the possibility of a bidiretional relationship was put forward, supported by a ertain markets imperfetion and by an informational asymmetry between the traders. We look into this ausality relationship onsidering the equity index produts (futures and options) and the stok exhange markets whih are members of the Euronext.liffe, exept for the Lisbon. We ompute a VAR and we perform ausality tests in the sense of Granger. In general, it seems diffiult to formulate a firm onlusion on the informational ontent of the derivative markets and on the objet (hedging or speulation) of the dominant operations, in the ontext in whih the ausality relationships whih our differ onsiderably between one produt and another and between one ountry and another. Keywords: finanial volatility, derivative produts, VAR, Granger ausality, Euronext.liffe JEL lassifiation: G12, G15, C22 1. INTRODUCTION The onsequenes of an exessive finanial volatility are well reognized. Finanial volatility is generally assoiated with the inertitude related to the future finanial flows and to the disount rate [BIS, 2006] and there are multiple fators whih influene it. Finanial volatility first results from the real volatility (of the GDP), but other fators suh as the eonomi and finanial liberalization [see Milles, 2002], the arhiteture of the markets, the

2 58 Claudiu Tiberiu ALBULESCU, Daniel GOYEAU distribution of assets owned by investors, the osts of the transation, the finanial innovations,. were put forward. As Antoniou and Holmes [1995] observe, the worries related to the impat of ontrats futures upon the spot markets manifested sine the very first transations with futures on the Chiago Board of Trade, in Analyses foused on the responsibility of speulators in amplifying the volatility and they highlight the fat that an inrease of the volatility is undesirable. The impat of the derivative produts introdution on the volatility of the underlying assets was largely srutinized after the Asian risis and it still stands for a subjet of interest nowadays, in the ontext of the present worldwide risis. From the empirial point of view, the influene of derivative produts on the volatility of the underlying assets was approahed in two different manners. The first approah ompares finanial volatility before and after the introdution of derivative produts (in partiular the futures ontrats). The seond approah, more reent, investigates the impat of derivatives on the behaviour of the underlying assets, inluding on their volatility. This latter approah reahes two different sets of results [Bandivadekar and Ghosh, 2003]. The first one alleges that the introdution of derivative produts will lead to an inrease of the volatility on the spot markets, thus destabilizing these markets [Figlewski, 1981; Stein, 1987]. The seond set of results, unlike the first one, supports the idea that the introdution of derivative produts ontributed to reduing the volatility [Powers, 1970; Shwarz and Laatsh, 1991]. Nevertheless, most of the researhes promote ontraditory arguments in respet of the effets of derivative produts on the prie volatility of the underlying assets [Dennisa and Sim, 1999]. We have to note that most of the studies apture the impat of the introdution of futures and options on the stok prie volatility. However, few researhes [Zapatero, 1998] look to other finanial assets, suh as for example the interest rate. The fat that both hedging and speulative operations with derivative produts are based on the prie volatility of their underlying asset a priori leaves room to the idea aording to whih the volume of ativity related to derivative produts has to follow in a unidiretional manner the prie volatility of the underlying assets. Therefore, if a more powerful assets volatility is antiipated, risk managers inhoate hedging operations while speulators take or strengthen their position by means of derivative produts. At the same time, this unidiretional onnetion, from the prie volatility of the underlying assets towards the volume of transations with derivative produts, suppose the existene of perfet markets with homogeneous information, being also required that the volume of transations does not provide information to the operators in respet of the future volatility of the underlying assets. Nevertheless, if there are traders better informed than others and if they are not apable of exatly antiipating the prie volatility, the ausality from the prie volatility of the underlying assets towards the volume of transations with derivative produts is no longer neessarily a unidiretional relation. Pratially, so far as the derivative produts provide lower transation osts and higher leverage effets, the traders who are better informed are suseptible of being more attrated by the derivative produts than by their underlying assets. In this ase, if the better informed traders lead the markets, the volume of transations with derivative produts has to forego the prie volatility of the underlying assets. In general, the sense of the ausality between the volume of transations with derivative produts and the prie volatility of the underlying assets would depend on the level of markets perfetion of the as well as on the nature of the operations (hedging or speulation) whih prevail on the derivatives markets. The distintion between the hedging and the spe-

3 Finanial Volatility and Derivatives Produts: A Bidiretional Relationship 59 ulative operations is not well-marked. Nevertheless, we may onsider that the speulative operations prevail in ase the growth of the derivative produts volume amplifies the volatility of the underlying assets, and, the reverse situation, that the hedging operations are more frequent if the inrease of the derivative produts volume results in a drop of the volatility. The analysis of this double ausality between the volume of transations with derivative produts and the finanial volatility is of reent interest in literature. Most of the researhes interrogate the role played by derivatives in amplifying finanial volatility and they make use of daily of intra-daily data. An exeption in this respet is the paper developed by Jeanneau and Miu [2003] whih shows, based on monthly data, that a high volatility on the finanial markets entails an augmentation of the transations on the derivative produts markets. More reently, Sarwar [2005] and Buhr et al. [2010] have tested the double potential ausality between volume of transations on options and the volatility of the S&P 500 index, respetively that of the ASX 200 index. In this paper, we intend to analyse the dynamis of the ausality relationship between the volume of transations with equity index produts (futures and options) and the pries volatility of their underlying asset, on the finanial markets members of the Euronext.liffe. We have eliminated from the alulations the transations performed on the Lisbon stok exhange, beause of the limited number of data available for the analysed period and of the strong volatility registered with number of traded ontrats. The index volatility is estimated based on the histori volatility, measured by means of the standard deviation of the stok market index returns, alulated for a rolling window of twelve months, and on the onditional volatility, omputed based on a GARCH(1,1) model. Our analysis distinguishes from the previous literature and mainly from the researhes of Sarwar (2005) and Buhr et al. (2010), in respet of two main aspets. On the one hand, our paper proeeds to an analysis of the relationship stability, onurrently srutinizing the markets of several ountries, the United Kingdom, Frane, Netherlands and Belgium. On the other hand, it simultaneously onsiders the ontrats futures and options. It is possible to meet ausality differenes in as far as the ontrats options are the only finanial produts allowing to bet diretly on the prie volatility of the underlying assets. The rest of the paper is strutured as follows. A first setion proposes a review of the literature on the relationship finanial volatility derivative produts. The seond part desribes the data and the applied methodologies and it briefs the results of the eonometrial estimations. Finally, the last setion summarizes the main findings. 2. REVIEW OF THE LITERATURE The literature approahing the topi of the onnetion between finanial volatility and derivative produts was for a long time developed in one single diretion and its endeavour was to emphasize the impat of derivative produts on the prie volatility of the underlying assets. From the theoretial point of view, there are two main antagonizing approahes. The first approah supports the idea that the transations with derivative produts lead to an inrease of the volatility on the spot markets. The derivatives markets, through the leverage effet they provide, attrat the investors, situation whih is suseptible to generate an augmentation of the volatility on the spot markets. Moreover, the liquidity existing on this latter market direts towards the derivative produts and this liquidity redution might amplify the volatility on the spot markets.

4 60 Claudiu Tiberiu ALBULESCU, Daniel GOYEAU The other approah onsiders that the introdution of derivatives diminishes the prie volatility of the underlying assets. Thus, several arguments are put forward to support this theory. Skinner [1989] alleges that there are ertain onditions to be met in respet of the underlying asset in order for a ontrat with these produts to be listed. These seletion riteria give onfidene to the investors who expet a redution of the volatility of the assets returns after the introdution of derivatives. Other voies argue that the derivative produts markets, due to their omplexity, generally draw in better informed investors. Fedenia and Grammatikos [1992] also rekon that the bid-ask spreads on the stok market an diminish after the introdution of derivative ontrats on these stoks and, if so, the volatility of the stok market would derease. From the empirial point of view, the outomes are very ontraditory [Charupat, 2006]. They are influened by the market type, assets type, volatility alulation method and seleted periods. We have to note that most of the papers analyze the impat of the introdution of equity index futures on the volatility of the underlying assets. Amongst these researhes, there are some showing that the introdution of derivatives amplified the volatility of the underlying assets (see for example the researhes of Robinson [1994] on the FTSE 100 index, of Reyes (1996) on the CAC 40 index, of Antoniou et al. [1998] on DAX 100 and Swiss MI indies and those of Antoniou and Holmes [1995] on the FTSE 100 index). Few studies illustrate, from the empirial perspetive that the intensifiation of derivatives transations leads to a lessening of the underlying assets volatility. Using an EGARCH model, Kasman and Kasman [2008] reah the onlusion that the futures introdution lowers the onditional volatility of the ISE 30 index. Nevertheless, a onsiderable number of papers either do not find a signifiant effet in this respet [Edwards, 1988; Darrat and Rahman, 1995] or underline only a redued effet [Dennisa and Sim, 1999; Jeanneau and Miu, 2003]. These ontraditory results are mainly influened by the retained volatility onept. The first papers on this subjet measure the non-onditional variane while other studies estimate the onditional variane, depending on its past values. Reent researhes use the ARCH (Autoregressive Conditional Heteroskedastiity), GARCH (Generalized ARCH) or TGARCH (Threshold GARCH) models, whih take into aount the heteroskedastiity of assets returns. Pratially, in order to measure the volatility, it is neessary to make a distintion between long-range volatility (months, years) and the short-range volatility (hours, days). If the long term volatility is influened in partiular by the eonomi fundamentals and institutional hanges, on short-range, the volatility is generated by the pressure and turmoil asertained in respet of the transation proess, as well as by the noises. An important aent has lately fallen on this latter type of volatility whih depends on its past values. Daly [2008] depits an exhaustive presentation of the finanial volatility alulation tehniques and makes a distintion between statisti methods and time-varying measures of volatility. A ommon measure of stok market volatility is the standard deviation of returns. For example, Cushman [1983] makes use of a rolling window of four observations to measure the standard deviation of the exhange rate, as a proxy for the exhange rate volatility. These measures of volatility are referred to as time invariants measures. However, the variane of stok returns is widely aknowledged to be time-varying and, onsequently, the usefulness and effiieny of time invariant measures have been questioned. This resulted into inreasing attempts to develop time-varying methods to alulate the volatility (volatility depends on its past values). The best known models reahing this purpose are ARCH model

5 Finanial Volatility and Derivatives Produts: A Bidiretional Relationship 61 (developed by Engle [1982]) and GARCH model whih represents an extension of the ARCH model into a generalized one (developed by Bollerslev [1986]). GARCH is an intuitive model to predit volatility. Another onept of volatility is now often used, more speifially the implied volatility. This type of volatility is identified using a model of Blak and Sholes based on observed stok pries and its stands for a measure of volatility antiipated by the market. Thus, Jeanneau and Miu [2003] use the atual volatility, measured by means of the annualized standard deviation of the assets prie hange, and the implied volatility. In our paper, we do not retain the delineation between fundamental volatility and transitory noise volatility. Atually, in so far as our analysis is aimed at investigating the ausality between the volume of transations with derivative produts and the prie volatility of their underlying asset, it seems diffiult to explain why the operators hedging or speulative attitudes on the markets would differ, depending on the type of observed or predited volatility (either fundamental or noise volatility). At the same time, we leave aside, as most of the reent researhes do, the implied volatility. Pratially, in so muh as it depends on the in-the-money, at-the-money and out-of-the-money nature of the options (volatility smile), the relation between the volume of transations with derivatives and the volatility of the underlying assets is suseptible of being distorted by the evolution of the type of exhanged options (in-the-money, at-the-money and out-of-the-money). Besides, the results of Buhr et al. [2010] show that, on the Australian market, the ativity of all options near-themoney and in-the-money has a signifiant preditive ability for the prie volatility of underlying assets, while the market ativity of all options out-of the-money only has a poor or even no preditive ability. At the same time, beause we retain the aggregate volume of ontrats options traded on eah of the markets, it seems deliate to retain the onept of implied volatility to perform our analysis. Therefore, in order to approximate the spot markets volatility, we retain two onepts of volatility: histori volatility and onditional volatility. Thus, the GARCH type model is used to define the short term volatility, generated by market onditions and by noises, while the lassial statisti methods an be applied to measure the long term volatility. In this ontext, we resort to the standard deviation of equity index returns for a rolling window of twelve months (data available in the Yahoo.finanes database) to measure the histori finanial volatility, using at the same time a GARCH(1,1) model to alulate the onditional volatility equity index returns. Aiming at emphasizing the biunivoal relationship whih an our between finanial volatility and volume of transations with derivative produts, we estimate a VAR. High assets prie volatility an entail an inrease of the volume of transations with derivative produts and, inversely, an intensifiation of derivatives transations an result into an inrease of prie volatility of the underlying assets. To test this double ausality, the Granger method is retained. Thus, if the index volatility has a Granger ausality effet on the volume of derivatives, this leaves room to think that the hedging operations prevail on the derivatives markets. In exhange, if the volume of derivative produts has a Granger ausality effet on the indexes volatility, then the speulative operations prevail. 3. EMPIRICAL ANALYSIS We first present the data, then we proeed to the methodology and then we onlude with the results.

6 62 Claudiu Tiberiu ALBULESCU, Daniel GOYEAU 3.1. The data There are two main approahes to measure the derivative produts related ativity [Jeanneau and Miu, 2003]. The first possibility onsiders their turnover (or volume), and it refers to the number of ontrats traded in a speifi period. The other approah is the open interest, whih advert to the total number of ongoing ontrats. Taking into onsideration that there are few available data on the value of the volume or the fat that this information is of reent nature, we have preferred using an indiator mirroring the number of ontrats to measure the derivative produts related ativity. At the same time, the number of ontrats and the total value of the ontrats volume are highly orrelated in so far as, on the exhange markets, the ontrats value is mainly standardized. The analysis onsiders the number of futures and options equity index ontrats. These data are available on Euronext.liffe. The analysis overs the period This timeframe is large enough and it is haraterized by signifiant evolutions of index returns volatility, whih proved high during risis periods, in partiular at the beginning and at the end of the years 2000s, and appeared to reah very low levels between these milestones [BRI, 2006] The methodology There are two main possibilities to study the bidiretional relationship between the derivative produts and their underlying assets, both entailing advantages and inonvenienes: the vetor autoregression (VAR) and the simultaneous equations models (SEM). The VAR was ritiized on the one hand by Koh [1993], who argued that the VAR ignored the possibility of ontemporaneous interations and ould therefore lead to biased results and to inaurate onlusions. On the other hand, Chan and Chug [1995] have found that the VAR model an better reveal the underlying proess and that SEM ould be misleading and yield unreliable inferenes. We resort to the VAR method to assess the relationship between finanial volatility and the volume of traded derivative produts. All the variables in a VAR are treated symmetrially by inluding for eah variable an equation explaining its evolution based its own lags and the lags of all the other variables in the model. Beause there is a priori a biunivoal relationship between the seleted variables, we onsider this method as the most appropriate. In order to selet the number of lags and the onstant, we refer to the information riteria (the likelihood ratio test an be also used to hoose the number of lags). Eah equation of the VAR enables to dedut an AIC (Aikake) riterion and a SC (Shwartz) riterion and we have alulated a weighted mean of the two (depending on the degrees of freedom) whih resulted into a single AIC and SC riterion. Further on, we have retained the VAR whih minimized the two information riteria. Then, we have interrogated the stability of the VAR. The estimated VAR is stable if all roots have modules less than one and lie inside the unit irle. If the VAR is not stable, ertain results (suh as impulse response standard errors) are not valid. In our ase, all the VAR retained into the analysis are stable. Finally, we perform a Granger ausality test to identify the diretion of the ausality between finanial volatility and the number of negotiated derivatives ontrats. We repeat this type of analysis for eah ategory of ontrats (futures and options) and for eah stok market.

7 Finanial Volatility and Derivatives Produts: A Bidiretional Relationship 63 But, before testing our VAR, we have heked the stationarity of the data retained for the analysis (see Table 1). In order to do this, we have applied two types of tests: the Augmented Dikey-Fuller (ADF) and the Phillips-Perron (PP). Null hypothesisnon stationarity Table no.1 Unit root tests United Kingdom Frane Netherlands Belgium (level) ADF PP ADF PP ADF PP ADF PP feip *** -4.5*** -2.9** ** ** oeip -2.9** * ** -6.8*** -3.9*** sdir * -2.7* * vir -4.0*** -3.9*** -3.7*** -3.7*** -5.5*** -5.5*** -4.3*** -4.1 (*) (**) and (***) indiate the rejetion of the null hypothesis at 10%, 5%, respetively 1% (t-statisti) Note : feip number of futures equity index produts; oeip number of options equity index produts; sdir index returns histori volatility; vir index returns onditional volatility. The stationarity tests are required beause the presene of a unit root within the data had very important effets from the statisti point of view. First, the presene of regressors with a unit root within a regression an lead to the situation in whih apparently very good regressions are estimated amongst variables whih are totally independent one from another. Then, a series stationary in level and a series stationary in first differene show a radially opposed behavior on the long run. A stationary time series tends to reposition itself around its determinist trend after a mean reversion. A series stationary in differene does not return around its trend after a reversion, beause the ho also affets the stohasti trend of the series. As we an observe in Table 1, the hoie of the VAR is appropriate beause the variables are stationary in level aording to one test or to the other The results The results related to the relationship between index returns histori volatility and the number of futures equity index produts, are listed in Table 2 below. Table no. 2 Causality relation between index returns histori volatility (sdir) and number of futures equity index produts (feip) UNITED KINGDOM VAR sdir feip Granger ausality Probability of the null hypothesis sdir(-1) (125.4) sdir (-2) feip(-1) (0.709) (2.379) (21.18) feip(-2) R feip does not ause sdir sdir auses feip (0.47) (0.01)

8 64 Claudiu Tiberiu ALBULESCU, Daniel GOYEAU FRANCE VAR sdir feip Granger ausality Probability of the null hypothesis sdir(-1) (0.06) (7.86) feip auses sdir sdir does not ause feip (0.04) (0.24) sdir (-2) (0.06) (7.77) feip(-1) (0.00) (0.06) feip(-2) (0.00) (0.06) R NETHERLANDS VAR sdir feip Granger ausality Probability of the null hypothesis sdir(-1) (13.06) (- 1.60) feip does not ause sdir sdir does not ause feip (0.13) (0.22) sdir (-2) (1.405) (1.704) feip(-1) (0.233) (8.818) feip(-2) (- 0.91) (5.984) R BELGIUM VAR sdir feip Granger ausality Probability of the null hypothesis sdir(-1) (13.71) sdir (-2) (- 0.72) feip(-1) (0.704) feip(-2) (0.186) R *( ) t-statisti (- 0.52) (0.477) (9.601) (4.604) feip auses sdir sdir does not ause feip (0.02) (0.86) In respet of the index returns histori volatility, two situations an our. For the United Kingdom, it is the index volatility that determines the inrease of the transations with futures equity index produts (the hedging operations prevailing), whereas for Frane and Belgium, the outomes indiate an inverse relation the inrease of the traded futures equity index produts auses the finanial volatility inrease. For Netherlands, no relation is indiated. Exept for the United Kingdom, all the VAR retained into the analysis embody two lags. If we look at the lags oeffiients (and at the orresponding t-statisti), the interdependene relationship between the two variables appears rather of poor signifiane. At the same time, the Granger test provides additional information.

9 Finanial Volatility and Derivatives Produts: A Bidiretional Relationship 65 The results related to the relationship between index returns onditional volatility and number of futures equity index produts is presented in Table 3. Table no. 3 Causality relation between index returns onditional volatility (vir) and number of futures equity index produts (feip) UNITED KINGDOM VAR vir feip Granger ausality Probability of the null hypothesis vir(-1) (23.04) (2.625) feip does not ause vir vir auses feip (0.13) (0.01) vir (-2) feip(-1) (1.495) (20.22) feip(-2) (3.615) (-1.03) R FRANCE VAR vir feip Granger ausality Probability of the null hypothesis vir(-1) (0.06) (-1.98) feip auses vir (0.00) vir (-2) (0.06) (2.530) vir auses feip (0.01) feip(-1) (0.00) (8.662) feip(-2) (0.00) (6.275) R NETHERLANDS VAR vir feip Granger ausality Probability of the null hypothesis vir(-1) (9.726) (-1.03) feip auses vir vir does not ause feip (0.05) (0.57) vir(-1) vir (-2) feip(-1) (10.27) (1.199) (0.548) (0.478) (-1.31) (9.633) feip auses vir vir does not ause feip feip(-2) (0.281) (4.846) R *( ) t-statisti vir (-2) (0.587) (0.892) feip(-1) (1.351) (8.544) feip(-2) (-0.80) (6.197) R BELGIUM VAR vir feip Granger ausality Probability of the null hypothesis (0.01) (0.21) When approximating finanial volatility using onditional volatility, a general analogue fat an be asertained. As for the histori volatility, no ausality diretion learly evolves for the entire set of ountries. In general, having as starting point the analyzed futures European markets, it seems, on one hand, that the futures derivatives markets are not apable of providing reliable information on the future volatility of the stok markets and, on the other hand, that no type of operation (hedging or speulation) signifiantly dominates the other.

10 66 Claudiu Tiberiu ALBULESCU, Daniel GOYEAU For the options ontrats the situation is different. When approximating the volatility using histori volatility, a ausality relationship appears only for the United Kingdom and Belgium. Yet, if in ase of the United Kingdom the transations with options seem to preede the volatility showing thus a more speulative harater, for Belgium, the diretion of the ausality goes in the other sense and the transations with options mainly result from hedging operations (see Table 4). sdir(-1) sdir (-2) oeip(-1) oeip(-2) (16.26) (-1.27) (0.515) (-0.56) (-0.19) (0.397) (9.636) (5.386) oeip does not ause sdir sdir does not ause oeip R NETHERLANDS Table no. 4 Causality relation between index returns histori volatility (sdir) and number of options equity index produts (oeip) UNITED KINGDOM VAR sdir oeip Granger ausality Probability of the null hypothesis sdir(-1) (13.20) (-0.61) oeip auses sdir (0.04) sdir (-2) (-1.12) (0.657) sdir does not ause (0.80) oeip(-1) (-0.62) (8.080) oeip oeip(-2) (1.769) (4.654) (2.355) (0.378) R FRANCE VAR sdir oeip Granger ausality Probability of the null hypothesis (0.84) (0.22) VAR sdir oeip Granger ausality Probability of the null hypothesis sdir(-1) (11.23) (-1.27) oeip does not ause sdir sdir does not ause oeip (0.37) (0.34) sdir(-1) sdir (-2) oeip(-1) (74.72) (0.959) (2.185) (23.65) oeip does not ause sdir sdir auses oeip oeip(-2) R * ( ) t-statisti sdir (-2) (0.197) (1.405) oeip(-1) (0.790) (7.102) oeip(-2) (-0.35) (5.818) R BELGIUM VAR sdir oeip Granger ausality Probability of the null hypothesis (0.33) (0.02) When approximating the volatility using onditional volatility, the disparities amongst ountries are onfirmed. For the United Kingdom, a bidiretional relation ours. If we refer to Frane, a single univoal ausality relation emerges, from volatility towards the number

11 Finanial Volatility and Derivatives Produts: A Bidiretional Relationship 67 of options equity index produts. However, in all the other situations, no signifiant ausality relation appears (Table 5). vir(-1) vir (-2) oeip(-1) (13.43) (-0.45) (1.331) (-1.57) (2.286) (9.576) oeip does not ause vir vir auses oeip oeip(-2) (-1.14) (5.496) (3.379) (-0.56) R NETHERLANDS Table no. 5 Causality relation between index returns onditional volatility (vir) and number of options equity index produts (oeip) UNITED KINGDOM VAR vir oeip Granger ausality Probability of the null hypothesis vir(-1) (9.751) (-3.59) oeip auses vir (0.00) vir (-2) (0.893) (4.023) vir auses oeip (0.00) oeip(-1) (3.358) (8.058) oeip(-2) (-1.59) (5.305) (3.184) (0.303) R FRANCE VAR sdir oeip Granger ausality Probability of the null hypothesis (0.40) (0.04) VAR sdir oeip Granger ausality Probability of the null hypothesis vir(-1) (8.424) (-0.89) oeip does not ause vir vir does not ause oeip (0.15) (0.54) vir(-1) vir (-2) oeip(-1) oeip(-2) (18.82) (-0.82) (-1.00) (18.26) oeip does not ause vir vir does not ause oeip (2.946) (3.109) R *( ) t-statisti vir (-2) (0.495) (1.088) oeip(-1) (1.060) (6.884) oeip(-2) (-0.56) (6.035) R BELGIUM VAR sdir oeip Granger ausality Probability of the null hypothesis (0.41) (0.31) In general, the outomes of the eonometrial tests based on options transations onfirm the ambiguity of the results. It seems, on one hand, that options equity index produts are not apable of providing reliable information on the future volatility of stok market index and, on the other hand, that no type of operation signifiantly dominates the other. Table 6 provides a synthesis of results.

12 68 Claudiu Tiberiu ALBULESCU, Daniel GOYEAU Table no. 6 Synthesis of results Causality Volatility futures (hv) futures (v) options (hv) options (v) United Kingdom Volume Frane Netherlands Belgium Volume Volume Volume 4. CONCLUSIONS The literature on the relationship finanial volatility volume of transations with derivative produts is abundant, but, up to a reent date, it showed interest mainly for the impat of derivatives introdution on the underlying assets volatility. Nevertheless, at theoretial level, the sense of the ausality is ambiguous and depends on the level of market perfetion and on the nature of operations (hedging or speulation) prevailing on the derivatives markets. On one hand, high volatility an be an inentive for investors to over their positions resorting to derivative produts. On the other hand, an inrease of the number of traded ontrats an generate a more powerful volatility of the underlying assets. In this ase, the operations would show a speulative harater. These assumptions have been analyzed on the stok exhange markets members of Euronext.liffe, estimating a VAR model and proeeding to Granger ausality tests. The stok exhange index volatility is approximated using the histori volatility, measured by means of the stok exhange index standard deviation alulated for a rolling window of twelve months, and the onditional volatility, predited based on a GARCH(1,1) model. Unlike previous studies fousing on a single exhange market and on a single derivative produt, onsidering several exhange markets and several derivative produts enables to test the stability of identified ausalities. In general, our analysis does not allow removing the theoretial ambiguity in as far as no ausality onnetion, ommon to the different approahed exhange markets, emerges. Thus, it seems on one hand, that equity index European markets are not in ondition to provide reliable information on the future volatility of stok exhange markets and, on the other hand, that no type of operation (hedging or speulation) signifiantly dominates the other. Referenes [1] Antoniou, A., Holmes, O., Priestley, R., The effets of stok index futures on stok volatility: An analysis of the asymmetri response of volatility to news, Journal of Futures Markets, Vol. 18, 1998, pp [2] Antoniou, A., Holmes, P., Futures Trading, Information and Spot Prie Volatility: Evidene for the FTSE-100 Stok Index Futures Contrat Using GARCH, Journal of Banking & Finane, Vol. 19, 1995, pp [3] Bandivadekar, S., Ghosh, S., Derivatives and Volatility on Indian Stok Markets, Bank of India Oasional Papers, Vol. 24/3, 2003, pp [4] BIS, The Reent Behaviour of Finanial Market Volatility, BIS Paper 29, [5] Bollerslev, T., Generalised Autoregressive Conditional Heteroskedastiity, Journal of Eonometris, Vol. 31, 1986, pp

13 Finanial Volatility and Derivatives Produts: A Bidiretional Relationship 69 [6] Buhr, K.E., Li, X., Rose, L.C., The Informational Role of Options Trading Volume in the Australian Index Options Markets, Finane and Corporate Governane Conferene, La Trobe University, Australia, 7-9 April, [7] Chan, K., Chug, Y.P., Vetor autoregression or simultaneous equations model? The intraday relationship between index arbitrage and market volatility, Journal of Banking & Finane, Vol. 19, 1995, pp [8] Charupat, N., The effet of derivative trading on the underlying markets: Evidene from Canadian instalment reeipts trading, International Review of Eonomis and Finane, Vol. 15, 2006, pp [9] Cushman, D.O., The effets of real exhange rate risk on international trade, Journal of International Eonomis, Vol. 15, 1983, pp [10] Daly, K., Finanial volatility: Issues and measuring tehniques, Physia A, Vol. 387, 2008, pp [11] Darrat, A.F., Rahman, S., Has futures trading ativity auses stok prie volatility?, Journal of Futures Markets, Vol. 15, 1995, pp [12] Dennisa, S.A., Sim, A.B., Share prie volatility with the introdution of individual share futures on the Sydney Futures Exhange, International Review of Finanial Analysis, Vol. 8, 1999, pp [13] Edwards, F.R., Futures trading and ash market volatility: Stok index and interest rate futures, Journal of Futures Markets, Vol. 8, 1988, pp [14] Engle, R.F., Autoregressive onditional heterosedastiity with estimates of the variane of UK inflation, Eonometria, Vol. 50, 1982, pp [15] Fedenia, M., Grammatikos, T., Options trading and the bid-ask spread of the underlying Stoks, Journal of Business, Vol. 65, 1992, pp [16] Figlewski, S., Futures Trading and Volatility in the GNMA Market, Journal of Finane, Vol. 36, 1981, pp [17] Jeanneau, S., Miu, M., Volatility and derivatives turnover: a tenuous relationship, BIS Quarterly Review, Marh, 2003, pp [18] Kasman, A., Kasman, S., The impat of futures trading on volatility of the underlying asset in the Turkish stok market, Physia A, Vol. 387, 2008, pp [19] Koh, P.D., Reexamining intraday simultaneity in stok index futures markets, Journal of anking & Finane, Vol. 17, 1993, pp [20] Powers, M.J., Does Futures Trading Redue Prie Flutuations in the Cash Markets?, Amerian Eonomi Review, Vol. 60, 1970, pp [21] Reyes, M.G., Index futures trading and stok prie volatility: Evidene from Denmark and Frane, Journal of Eonomis and Finane,Vol.20, 1996, pp [22] Robinson, G., The effet of futures trading on ash market volatility: Evidene from the London stok exhange, Review of Futures Markets, Vol. 14, 1994, pp [23] Sarwar, G., The Informational Role of Option Trading Volume in Equity Index Options Markets, Review of Quantitative Finane and Aounting, Vol. 24, 2005, pp [24] Shwarz, T.V., Laatsh, F., Dynami Effiieny and Prie Leadership in Stok Index Cash and Futures Markets, Journal of Futures Markets, Vol. 11, 1991, pp [25] Skinner, D., Options markets and stok return volatility, Journal of Finanial Eonomis, Vol. 23, 1989, pp [26] Stein, J.C., Information Externalities and Welfare Reduing Speulation, Journal of Politial Eonomy, Vol. 95, 1987, pp [27] Zapatero, F., Effets of finanial innovations on market when beliefs are heterogeneous, Journal of Eonomi Dynamis and Control, Vol. 22, 1989, pp