Effect of Futures Trading on Spot Market Volatility: Evidence from Indian Commodity Derivatives Markets

Effect of Futures Trading on Spot Market Volatility: Evidence from Indian Commodity Derivatives Markets Author Mr. Brajesh Kumar Fellow, Indian Institute of Management, Ahmedabad, India Dorm 22, Room No. 27 Indian Institute of Management, Ahmedabad Vastrapur, Gujarat, India, 38005 Email: brajeshk@iimahd.ernet.in. Phone: +97966326277

Effect of Futures Trading on Spot Market Volatility: Evidence from Indian Commodity Derivatives Markets Abstract This study investigates the relationship between futures trading activity and spot market volatility for agricultural, metal, precious metals and energy commodities in Indian commodity derivatives market. This article contributes to the debate whether the futures trading in Indian commodity futures market stabilizes or destabilizes spot market. We explore this issue by modeling contemporaneous as well as dynamic relationship between spot volatility and futures trading activity including trading volume (speculative/day trading) and open interest (hedging). Following Bessembinder and Senguin (992), we examine contemporaneous relationship through augmented GARCH model in which spot volatility is modeled as GARCH (,) process and trading activity is used as explanatory variable. We also decompose futures trading volume and open interest series into expected and unexpected component. The lead-lag relationship between spot price volatility and futures trading volume and open interest is investigated through VAR model. Granger causality tests, forecast error variance decompositions and impulse response function are used to understand the dynamic relationship between these variables. We found that both expected and unexpected futures trading volume affects contemporaneous spot volatility positively. However, in case of agricultural commodities only unexpected volume affects the contemporaneous spot volatility. Granger causality tests, forecast error variance decompositions and impulse response function confirm that the lagged unexpected volatility causes spot price volatility for all commodities. The effect of speculative/day trading activity measured by trading volume on spot market volatility is positive. However, hedging activity measured by open interest does not show significant effect on spot market volatility. We do not find any effect of spot volatility on futures trading activity for most of the commodities. - 2 -

Effect of Futures trading on Spot Market Volatility: Evidence from Indian commodity derivatives Markets. Introduction The stabilizing/de-stabilizing effect of derivatives market on spot market has long been discussed in literature. Theoretical as well empirical studies on this are not conclusive. On theoretical grounds, it has been argued that futures trading may de-stabilize the spot volatility (Cox, 979; Figlewski, 98; Clifton; 985; Grammatikos and Saunders, 986; McCarthy and Najand, 993; Chatrath et al., 996). The argument is based on the fact that futures markets have low transaction cost and provide high leverage as compared to spot market. As a result, It is likely that the uninformed speculative investors will shift from the spot market to futures market. The shift of uninformed trader from spot market to futures market will decrease the market depth and de-stabilize the market by increasing spot volatility. Furthermore, spot and futures prices are linked by arbitrages. The effect of manipulation in the futures markets, if any, may trickle down (spillover effect) to spot market and destabilize the market through high spot volatility. On the other hand, some studies have argued that the derivatives market stabilize the spot market (Danthine, 978; Kyle's, 985, and Froot and Perold, 99). Futures markets are supposed to provide a medium for hedging, facilitate price discovery, and improve market efficiency. These studies argue that futures markets provide market depth, platform for hedging and play as a leading role in information dissemination. Any expected demand and supply shock in the spot market can be captured and hedged through futures market and hence will stabilize spot market volatility. The debate of stabilizing/de-stabilizing effect of derivatives market on spot market is also not conclusive in empirical literature. Empirical work related to the stabilizing and destabilizing effect of futures trading on spot volatility can be divided into two categories: a) investigating the spot volatility before and after the introduction of futures (Antoniou et al, 998; Lee & Ohk, 992) b) examining the interactions between futures trading activity (trading volume and open interest) and spot volatility (Bessemhinder & Seguin, 992; Gulen and Mayhew, 2000; Board et al, 200). A number of previous studies in equity, commodity and currency futures markets provide the empirical evidence of the de-stabilizing effect of futures trading on spot market. In equity market, many studies (Harris, 989; Damodaran, 990; Lockwood and Linn, 990; Schwert, 990; Chang, Cheng, and Pinegar, 999; Kyriacou and Sarno, 999) found the positive relationship between trading activity and spot volatility in US, UK and other developed markets. On the contrary, some studies (Santoni, 987; Bessembinder and Seguin, 992; Brown-Hruska and Kuserk; 995) found that futures trading negatively affect the spot volatility. Bessembinder and Seguin (992) divided the trading volume and open interest into expected and unexpected component and found that stock market volatility is positively related to unexpected trading activity, but negatively to expected trading activity. Chang et al. (2000) decomposed volatility estimates into expected and unexpected components and found that hedging activity in futures increases when unexpected volatility increases but speculative activity is not affected by volatility. In currency market, Clifton (985), Chatrath (996), Grammatikos and Saunders (986) and McCarthy and Najand (993) found positive correlation between spot price

variability and volume of futures trading. However, Adrangi and Chatrath (998) and Sarwar (2003) found stabilizing effect of futures trading on currency market. In commodity markets, very few studies have addressed this issue. Pashigian (986) and Weaver and Banerjee (990) found that futures trading activity destabilize the spot volatility of agricultural commodity. In a review paper, Kamara (982) explained that most of the empirical studies in agricultural derivatives market have supported that the introduction of futures trading generally reduced or at least did not increase the spot price volatility. On the other hand, Yang et al. (2005) examined the dynamic relationship between futures trading activity and spot volatility for agricultural commodities and found that an unexpected increase in futures trading volume unidirectionally causes an increase in spot price volatility for most agricultural commodities however, they found weak causal relationship between open interest and spot volatility. Methodologically, most of the studies pertaining to the empirical test of stabilizing/de-stabilizing effect of derivatives market on spot market generally focus on the paradigm of introducing futures trading. They compare spot market volatility before and after the introduction of futures trading. Recently, studies in this area model time varying volatility as GARCH process. Another approach used in the literature is estimation of contemporaneous relationship between trading activity (volume and open interest) and spot volatility. In this method spot price volatility is estimated as GARCH process. The effect of futures trading activity on spot volatility is investigated through augmented GARCH model in which trading activity is used as exogenous variable. Furthermore, futures trading volume and open interest are divided into expected and unexpected components and effect of expected and unexpected part on spot volatility is examined. The dynamic relationship between spot volatility and futures trading activity is also examined through Vector Autoregressive specifications. Most of the studies in this area have looked at the equity (index futures) and currency market, and very few studies have expanded this issue empirically in commodity futures market context. Recently, Yang et al. (2005) investigated this issue for agricultural commodity traded at developed commodity derivatives exchanges (CBOT, KCBT, CME etc). Studies on the commodity derivatives in the emerging market context are sparse. In emerging commodity derivatives markets context, where markets are generally very thin in terms of volume and number of derivatives products and participations is supposed to be limited, the effect of trading activity on spot volatility has not been investigated extensively. Emerging commodity markets are generally criticized for speculative activity and destabilizing role of derivatives in spot market through increased price volatility. For example, in Indian context, many a time agricultural commodities have been banned for their alleged destabilizing effect on spot market. The commodity derivatives markets in India also face increased regulation on futures trading despite any reliable statistical evidence. This paper adds to literature by investigating the issues of stabilizing/de-stabilizing of futures activity on spot market in an emerging commodity derivative market context. Further, in this paper, agricultural commodities, precious metals, metals and energy commodities are compared, which will hopefully help in understand the relationships for a broad range of commodities In this paper, the issues are addressed through both contemporaneous as well as dynamic relationship between spot volatility and futures trading activity. Following Bessemhinder & - 4 -

Seguin, (992), trading activity is divided into expected and unexpected part. The contemporaneous relationship is tested by GARCH model in which trading activity (expected and unexpected) variable is used as exogenous variable. The dynamic relationship across spot volatility, unexpected volume and unexpected open interest is also modeled through Vector auto regressive (VAR) specification. We use Granger causality test, variance decomposition technique and impulse response function to understand the issues more correctly. We specifically analyzed the following issues related to relationship among spot volatility, and futures trading activity (volatility and open interest). Do trading volume and open interest affect conditional volatility? Do expected and unexpected component of trading volume (open interest) behave in similar way? Is there any lead-lag relationship between conditional spot volatility and unexpected trading volume (open interest)? If yes, what is direction and sign of relationship on each other? In other words, does lagged trading activity increase/decrease spot volatility? Also, does spot volatility induce trading activity in the derivatives markets? Is the relationship convergent among different commodities or do different commodities (especially storable and non-storable, or local verses global) possess different dynamics? 2. Data and Time series Characteristics of Returns This study examines the effect of futures trading activity on spot volatility for Indian commodity futures market. The dynamic relationship across spot volatility, futures trading volume and open interest is analyzed. Our data consist of four agricultural commodities: Soybean, Maize, Castor seed, and Guar seed, three metals: Aluminum, Copper and Zinc, two precious metals: Gold and Silver, and two energy commodities: Crude oil and Natural gas. Details of the data period and source of data are given in Table. For agricultural commodities, data of volume and open interest are collected from NCDEX and for non-agricultural commodities MCX data is used. The selection of exchange for selecting the futures contract is based on comparatively higher trading volume of a commodity at exchange. These exchanges also report the data of spot prices of a particular delivery center (Table ) which has been used as data for spot market price. Three daily spot returns are constructed from the spot price data: as log(p s,t /P s,t- ), where P s,t is the spot price at time t. Table 2 presents basic statistics of spot returns of all commodities. The daily mean returns of agricultural commodities are positive and small except maize where negative returns are found. Precious metals and energy commodities also show positive return. In case of metals, copper has positive mean return where as aluminium and zinc have negative mean return. In terms of standard deviation, non-agricultural commodities show lower deviation in returns as compared to non-agricultural commodities, with highest deviation in metals. We also found positive as well as negative skewness in returns. Soybean, maize and silver returns show very high kurtosis (>5) as compared to other commodities. Autocorrelation functions of spot returns are also estimated up to 40 lags (Table 3 and Figure ). First order autocorrelation of most of the commodities are more than 5 %( except castor seed, guar seed, crude oil and zinc). Maize returns show highest positive autocorrelation even at higher lag. This market shows highest friction or over reaction of traders to new information which persists for more than 0 days. We also find some significant peaks (positive/negative) at higher lags. It may be because of pricing error or friction in the market. Stationarity of the spot returns - 5 -

is also tested using Augmented Dickey and Fuller (979) unit root test. It is found that all series are stationary. Results of the unit root test are presented in Table 4. Agricultural Bullion Metals Energy Table : Details of Commodity, Data Period and Source Commodities Data-Periods Future Market Spot Market Soy Bean 9//2004 to 0/20/2008 NCDEX Indore Maize /5/2005 to 0/20/2008 NCDEX Nizamabad Castor Seed 9/2/2004 to 0/20/2008 NCDEX Disa Guar Seed 4/2/2004 to 9/9/2008 NCDEX Jodhapur Gold 5/2/2005to 9/30/2008 MCX Ahmedabad Silver 5/2/2005to 9/30/2008 MCX Ahmedabad Aluminium 2//2006 to 9/30/2008 MCX Mumbai Copper 7/4/2005 to /20/2008 MCX Mumbai Zinc 4/3/2006 to 9/30/2008 MCX Mumbai Crude Oil 5/2/2005to 9/30/2008 MCX Mumbai Natural Gas 7/2/2006to 9/30/2008 MCX Mumbai Table 2: Summary Statistics: Spot Returns Commodity N Mean STD Skewness Kurtosis Soy Bean 970 6 0.875-7.624 43.472 Maize 709-0.3 2.34-0.686 23.244 Castor Seed 97 3 0.962 0.08.983 Agricultural Guar Seed 288 0.033.66 0.520 5.747 Gold 969 0.079.92-0.069 6.202 Bullion Silver 966 0.068.865 -.578 6.654 Aluminium 723-8.52 4.88 Copper 947 8 2.320-0.289 3.50 Metals Zinc 709-0.3 2.34-0.249 0.79 Crude Oil 984 6.904-0.028.936 Energy Natural Gas 97 3 0.962 0.08.983 Table 3: Autocorrelation Function of Different Commodities at Different Lag Lags Commodity 2 3 4 5 0 5 20 Agricultural Soy Bean 0.6 0.0-0.023 6 5-0.026 Maize 0.304 0.235 0.227 0.95 0.48-0.024-0.026-0.030 Castor Seed 0.038-0.034 9-0.04 5-0.028 0.033-0.022 Guar Seed 0.037-0.025-9 0.4 0.09 0.03 0.03-0.09 Bullion Gold -0.063 0.078 0.0 7 0.07 0.068-0.08 0.026 Silver -0.42-7 0.39-3 0.03-0.033 0.075 Metals Aluminium -0.22 7-0.04-0.02-0.03 0.08 3 0.023 Copper -0.2-0.02-8 0.067-0.039 0.038 7 0.08 Zinc -0.042-0.09 0.06-3 -0.085-0.072-0.03 0.063 Energy Crude Oil -0.04-0.03-2 -0.022 0.075-9 -7 0.00 Natural Gas -0.26-0.02-0.05-0.049 0.068-0.044 0.26-0. - 6 -

A utocorrelation A u t o c o r r e la t io n 0.5 0. - 0 3 5 7 9 3 5 7 9 2 23 25 27 29 3 33 35 37 39-0. 0. 0.08 0.06 0.04 0.02-0.04-0.06-0.08-0. Lags A utocorrelation 0.35 0.3 0.25 0.2 0.5 0. 0-3 5 7 9 3 5 7 9 2 23 25 27 29 3 33 35 37 39-0. -0.5 Lags A utocorrelation 0. 0.08 0.06 0.04 0.02 0-0.02 3 5 7 9 3 5 7 9 2 23 25 27 29 3 33 35 37 39-0.04-0.06-0.08-0. Lags A utocorrelation 0.5 0. 0 3 5 7 9 3 5 7 9 2 23 25 27 29 3 33 35 37 39 - a) Soybean b) Maize c) Castor seed d) Guar seed 0-0.02 3 5 7 9 3 5 7 9 2 23 25 27 29 3 33 35 37 39 Lags A u t o c o r r e la t i o n 0.2 0.5 0. 0-3 5 7 9 3 5 7 9 2 23 25 27 29 3 33 35 37 39-0. -0.5-0.2 Lags A u t o c o r r e la t io n 0. - -0. -0.5 0 3 5 7 9 3 5 7 9 2 23 25 27 29 3 33 35 37 39 Lags A u t o c o r r e l a t io n -0. -0.5 0. - -0. -0.5 Lags 0 3 5 7 9 3 5 7 9 2 23 25 27 29 3 33 35 37 39 e) Gold f) Silver g) Aluminium h) Copper A utocorrelation 0. - -0. -0.5 0 3 5 7 9 3 5 7 9 2 23 25 27 29 3 33 35 37 39 Lags A utocorrelation 0. 0.08 0.06 0.04 0.02 0-0.02 3 5 7 9 3 5 7 9 2 23 25 27 29 3 33 35 37 39-0.04-0.06-0.08 Lags 0 3 5 7 9 3 5 7 9 2 23 25 27 29 3 33 35 37 39 - i) Zinc j) Crude oil k) Natural gas Figure : Autocorrelation Function of Spot Returns A u toc orre latio n 0.5 0. -0. -0.5-0.2 Lags Lags 7

Table 4: ADF Test of Unit Root Commodity Type Rho Pr < Rho Tau Pr < Tau F Pr > F Soy Bean Zero Mean -224.659 0-7.8 <.000 Single Mean -225.837 0-7.8 <.000 30.44 Trend -225.765 0-7.79 <.000 30.4 Maize Zero Mean -434.862 0-8.4 <.000 Single Mean -459.876 0-8.47 <.000 35.88 Trend -460.467 0-8.47 <.000 35.85 Castor Seed Zero Mean -596.425 0-8.75 <.000 Single Mean -692.074 0-8.89 <.000 39.53 Trend -889.303 0-9.09 <.000 4.35 Guar Seed Zero Mean -783.365 0-0.36 <.000 Single Mean -789.27 0-0.36 <.000 53.65 Trend -80.674 0-0.37 <.000 53.85 Gold Zero Mean -556.804 0-8.65 <.000 Single Mean -672.384 0-8.88 <.000 39.45 Trend -675.94 0-8.88 <.000 39.42 Silver Zero Mean -668.726 0-8.89 <.000 Single Mean -723.282 0-8.98 <.000 40.3 Trend -790.25 0-9.06 <.000 4.02 Aluminium Zero Mean -547.239 0-7.76 <.000 Single Mean -549.409 0-7.76 <.000 30.08 Trend -552.99 0-7.75 <.000 30.06 Copper Zero Mean -252.375 0-7.52 <.000 Single Mean -252.344 0-7.52 <.000 28.28 Trend -38.027 0-7.88 <.000 3.05 Zinc Zero Mean -2.08 0-8.58 <.000 Single Mean -499.64 0-8.65 <.000 37.42 Trend -640.98 0-8.7 <.000 38.07 Crude Oil Zero Mean -42.48 0-9.43 <.000 Single Mean -252.85 0-9.48 <.000 44.99 Trend -272.04 0-9.49 <.000 45.03 Natural Gas Zero Mean -660.30 0-7.52 <.000 Single Mean -682.55 0-7.52 <.000 28.3 Trend -727.729 0-7.54 <.000 28.44 As described in Table, the data of daily futures trading volume and open interest are also collected. The basic statistics of trading volume and open interest are presented in Table 5. It is interesting to note that in agricultural commodities, mean open interest is higher than the mean volume. It is not the case with other non-agricultural commodities. In agricultural commodities, Guar seed has highest volume (both in quantities and value term) followed by Soybean, Castor seed and Maize in decreasing orders. In non-agricultural commodities, Gold has maximum turnover followed by Silver, Crude oil, Copper, Zinc, Natural gas and Aluminium in decreasing order. 8

Table 5: Summary Statistics: Volume, Open Interest, and Turnover Turnover Volume Open Interest (in millions rupees) Commodity N Mean Std. Dev Mean Std. Dev Mean Std. Dev Agricultural* Soy Bean 89 43.63 35.96 38.92 58.26 747.48 75.59 Maize 970 8.83 4.86 28.78 9.65 66.20 6.35 Castor Seed 97 2.46.94 9.97 4.4 46.56 35.73 Guar Seed 288 209.66 36.4 05.0 34.35 3720.45 2952.43 Bullion+ Gold 969 30.36 2.8.5 0.30 30642.48 24976.20 Silver 966 0.93 0.55 0.38 0.7 737.77 665.46 Metals$ Aluminium 723.38.3 2.40.57 64.24 66.64 Copper 947 34.8 28.63.39 6.6 0360.65 8330.49 Zinc 709 45.33 33.63 28.87 7.3 5479.62 4233.6 Energy@ Crude Oil 984 3.64 2.89.25 0.56 3379.43 3960.44 Natural Gas 608 2.97 2.60.2 0.83 0.7 904.43 * For agricultural commodities volume and open interest are in 0,000 MT + In case of bullions, volume and open interest for gold are in,000 Kg and for silver they are in 000MT $ In case of metals, volume and open interest are in 000MT @ In case of energy commodities, volume and open interest for crude oil are in 000,000 BBL and for Natural gas they are in 000,000 mmbtu The correlation among spot return, futures trading volume (quantity and value) and open interest is also analyzed and represented in Table 6. All commodities, except soybean, gold, copper and crude oil have positive correlation between trading volume and spot returns. The correlation between open interest and spot returns is negative for all agricultural commodities except maize and guar seed. All metals and natural gas also show negative correlation. However, bullions and crude oil show positive correlation. It is important to note that all commodities show positive and very high (20-60%) correlation between open interest and trading volume except silver. In case of silver this correlation is negative. Following Kim (2005), Kim et al. (2004), Fung and Patterson (999) and Campbell et al. (993), in order to eliminate any secular volume growth and to form a stationary time series of trading volume, we transform the trading volume as well as open interest series through incorporating 50-day backward moving average as follows Vt V t = [] 50 V t i 50 i= where, V t is the trading volume or open interest at time t. This transformed series produces a stationary time series that captures the change in the long run movement in trading volume or open interest. However, we also perform Augmented Dickey and Fuller (979) unit root test on de-trended volume and open interest and find that volume and open interest are stationary. Autocorrelation functions and results of unit root test on de-trended volume and open interest can be obtained from author on request. 9

Table 6: Correlation Coefficients between Spot Return Volume Open Interest and Turnover Spot return Volume Open Interest Turnover Spot return.00 Agricultural Bullion Metals Energy Soy Bean Maize Castor Seed Guar Seed Gold Silver Aluminium Copper Zinc Crude Oil Natural Gas Volume -0.03.00 Open Interest 0.48.00 Turnover -0.02 0.96 0.4.00 Spot return.00 Volume 0.3.00 Open Interest 0.0 0.63.00 Turnover 0.5 0.99 0.63.00 Spot return.00 Volume 0.06.00 Open Interest -0.03 0.4.00 Turnover 0.08 0.96 0.30.00 Spot return.00 Volume 0.07.00 Open Interest 0.5.00 Turnover 0.97 0.45.00 Spot return.00 Volume -0.02.00 Open Interest 0.02 0.4.00 Turnover 0.97.00 Spot return.00 Volume.00 Open Interest 0.03-0.04.00 Turnover 0.0 0.95-0.22.00 Spot return.00 Volume 0..00 Open Interest -0.07 0.3.00 Turnover 0. 0.99 0..00 Spot return.00 Volume -0.08.00 Open Interest -0.2 0.56.00 Turnover -0.06 0.97 0.50.00 Spot return.00 Volume 0.08.00 Open Interest -0.06 0.49.00 Turnover 0.07 0.9 0.25.00 Spot return.00 Volume -0.03.00 Open Interest 0.04 0.38.00 Turnover -0.02 0.93 0.32.00 Spot return.00 Volume 0.07.00 Open Interest -0.03 0.35.00 Turnover 0.07 0.96 0.20.00 0

3. Spot Volatility, Trading Volume and Open Interest: An empirical analysis We first represent the results of contemporaneous relationship between spot volatility and futures trading activity in Indian commodity market. We model the spot volatility as GARCH process and then GARCH equation is augmented with expected and unexpected futures trading volume and open interest. The dynamic relationship across spot volatility, unexpected volume and unexpected, modeled through VAR model, is also subsequently represented. 3.. Contemporaneous Relationship between Conditional Volatility and Futures Trading Activity: A GARCH Model Approach As suggested by Antoniou and Foster, (992) and Gulen and Mayhew (2000), we model the spot price volatility as a GARCH (,) process, which captures the well-known time-varying pattern of commodity spot volatility. The futures trading volume and open interest are also decomposed into expected and unexpected part using ARIMA (5,0,5) 2 model as in Bessembinder and Seguin (992, 993). The predicted part from ARIMA model gives the expected volume (open interest) and residuals are used as unexpected volume (open interest). We augment the volatility equation with expected and unexpected futures trading volume (open interest). The expected and unexpected contemporaneous volume and open interest are used as explanatory variable. The GARCH equation containing trading volume and open interest as exogenous variable in the volatility equation (Lamoureux and Lastrapes, 990) is as follows rt = a+ b r k i= i t i + ε t 2 ε t ψ t ~ N (0, σ t ) and [2] 2 t q p 2 2 = α0+ αε i t i + β Jσ t j + γv t + ηµ v, t + λoi t ϕµ OI, t. i= j= σ + Where, r t is the spot return at time t, V t is the expected volume at time t, µ vt is the unexpected volume at time t, O It is the expected open interest at time t and µ OIt is the unexpected open interest. The augmented GARCH model is used to estimate contemporaneous relationship between conditional volatility and trading activity (expected and unexpected). The conditional spot volatility is modeled as GARCH (,) process and we find that in all commodities, spot volatility follow a GARCH process with high degree of persistence (α +β >0.9). Results are presented in Table 7. Following Bessembinder and Seguin (992), first the trading volume and open interest are divided into expected and unexpected component and then spot volatility equation, which is modeled as GARCH process, is augmented with expected and unexpected futures trading volume and open interest as exogenous variable (equation 2). Parameters estimates of the GARCH model of spot volatility with expected and unexpected trading volume and open interest are shown in Table 8. 2 Similar approaches are employed by Board, et al (200), Kim, et al (2004) to consider the volume decomposed into predictable and unpredictable components. We find the correct model for the series depending upon AIC and SBC criteria. The augmented Dickey-Fuller test reveals that the detrended volume series and open interest are stationary, and thus the series is modeled without differencing.

Table 7: GARCH Parameters of Return Volatility Sum Of AR α α β α +β Soy Bean -0.258 0.60* 0.849* 0.33*.80 Agricultural Maize -0.448 0.02* 0.37* 0.846* 0.984 Castor Se -0.064 0.03* 0.06* 0.90* 0.968 Guar Seed -4 0.06* 0.3* 0.84* 0.978 Bullion Gold -0.032 3 * 0.95*.002 Silver 0.80 0.475* 0.258* 0.68* 0.876 Metals Aluminium 0.8 5# 0.06* 0.9* 0.976 Copper 0.29 0.22* 0.06* 0.88* 0.956 Zinc 0.08 0.894 0.077# 0.754* 0.83 Energy Crude Oil 0.075 0.094 0.04* 0.93* 0.975 Natural Gas 0.225 0.223+ 0.067* 0.9* 0.983 *(#) significant at % (5%) Table 8: GARCH Model parameters of spot Volatility with Expected and Unexpected Trading Volume and Open Interest Sum Of AR α α β α +β γ A λ C Soy Bean 0.237 0.72* 0.288* 0.569* 0.858 2 0.476* 4 0.072 Agricultural Maize 0.263 0.042* jd/-c* 0.803* 0.93 0.45* 0 0 Castor Seed 0.093 0.90* 0.090* 0.64* 0.73 7 0.69* 5 0 Guar Seed 0.92 0.294* 0.228* 0.642* 0.870 0.504* 0.075 0 Bullion Gold -0.39 0.638* 0.227* 0.43* 0.64 0 0.823* 0 0.037 Silver 0.5 0 0.88* 0.684* 0.872 0.395* 0.674* 0 0 Aluminium 0.424 0.627* 0.8* 0 0.8.3* 0.62* 0 0 Metals Copper 0.82 0.029 8* 0.739* 0.846 0.624*.569* 0 0 Zinc -0.282.28* 0.023 0.492* 0.54.27# 3.004* 0 0 Energy Crude Oil -0.456 0.06 0.099* 0.442* 0.54.523* 0.609* 0 2.656 Natural Gas 0.20 2 0.072* 0.94* 0.986 0.36 0.84# 0 0 *(#) significant at % (5%) It is found that for non-agricultural commodities the coefficient estimates of expected trading volume is positive and significant except natural gas. This result is contradictory to the results of Bessembinder and Seguin (992) on equity market. They found that an increase in trading expected volume reduces the spot price volatility. In agricultural commodities the coefficients are positive but insignificant. The coefficient estimates of unexpected trading volume are positive and significant for all commodities. This evidence is consistent with Bessembinder and Seguin (992). This result is not unanticipated because futures and spot markets are interlinked and any information shock should generate volume in both the markets. Unlike the results of expected and unexpected volume, the coefficient estimates of both expected and unexpected open interest are insignificant. These results also contradict the findings of Bessembinder and Seguin (992) that expected open interest reduces the spot volatility and unexpected increases it. The evidence indicates that the commodity spot volatility is positively related to the expected futures trading volume for non-agricultural commodities. This result suggests that the spot 2

volatility is increased when futures trading volume is high for non-agricultural commodities. The finding of destabilizing effect of futures trading volume on spot volatility is in line with the large number of empirical studies who report positive relationship between trading volume and spot volatility. Our results also highlight that in Indian commodity derivatives market open interest has no role in information dissemination and in stabilization/destabilization effect on spot volatility. The GARCH model estimates the contemporaneous relationship between spot volatility and futures trading activity. It does not consider the dynamic relationship among these variables. We investigate the lead-lag relationship between spot volatility and trading activity through VAR model. The Granger causality test, variance decomposition and impulse response functions are used. The contemporaneous relationship modeled through GARCH doesn t test the dynamic relationship among spot volatility and futures trading activities. However, the relationship among spot volatility, futures trading volume and open interest may be dynamic in nature (Yang, 2005), which can be explained through VAR model. In VAR model, the unexpected component of futures trading activity is used for examining the dynamic relationship among spot volatility, trading futures volume and open interest. 3.2 Dynamic Relationship between Spot Volatility, Futures Volume and Open Interest: A VAR Modeling Approach This paper uses a vector autoregressive (VAR) model to investigate the interactions among spot volatility, futures trading volume and open interest as follows: σ µ µ 2 t v, t = OI, t = a =, t a 2, t a + 3, t + i= + k k i= b k i=, t b 2, t b σ 3, t 2 t i σ σ + 2 t i 2 t i + k i= + k c i= k i=, t c 2, t c µ 3, t v, t i µ + v, t i µ v, t i i= + k + k d i= k i=, t d 2, t d µ OI, t i µ 3, t OI, t i µ OI, t i [3] 2 Where, σ t is conditional volatility estimate of spot returns from GARCH (,) model, µ v, t is the unexpected futures volume obtained from ARIMA model, and µ OI, t is unexpected open interest obtained from ARIMA model. Sims (972, 980) and Abdullah and Rangazas, (988) suggested that the variance decomposition and orthogonalized Impulse response function of the forecast error is advisable while analyzing the dynamic relationship between variables because it may be misleading to rely solely on the statistical significance of economic variables as determined by VAR model or Granger causality test. Therefore, we also estimate the variance decomposition of the forecast error and orthogonalized Impulse response function of the forecast error for each endogenous variable. The variance decomposition of the forecast error gives the percentage of variation in each variable (e.g. spot volatility) that is explained by other variables in the system (e.g. volume 3

or open interest). The orthogonalized Impulse response functions estimates the response of one variable (e.g. volatility) after shock in each variable (e.g. open, interest or volume) at a particular time for a specified length in future, which is serially and contemporaneously uncorrelated. 3.2.: Granger Causality Test The bivariate causality tests between spot volatility and unexpected trading volume, and spot volatility and unexpected open interest are performed. In this study, the five lagged length of each variables are used consistently. The choice of five lag is to control for any weekly effect these time series may have. The Chi- square statistics of the granger causality test are presented in Table 9. In panel A, the bi-drectional causality between spot volatility and futures trading volume is shown whereas in panel B bi-drectional causality is presented for spot volatility and open interest. If p-value associated with Chi- square statistics is greater than 0% significance level, the null hypothesis of no causality in the given direction can not be rejected. The first column of panel-a presents the Granger causality running from trading volume to spot volatility and second column represent the causality from spot volatility to trading volume. The results of the lead-lag relationship between spot volatility and trading volume suggest that for all commodities, unexpected trading volume causes spot price volatility. This result is consistent with the findings of Yang (2005) on agricultural commodity futures, Chatrath, Ramchander and Song (996) on currency futures but contradictory to the findings of Darrat and Rahman (995) on S&P 500 index futures. However, the spot price volatility causes the unexpected trading volume for four of the eleven commodities in studies. For maize, Guar seed, Aluminium and crude oil, bi-directional relationship exists between futures trading activity and spot price volatility. Our result is similar to the findings of Yang (2005) on sugar futures but stand in sharp contrast with the findings of Darrat, Rahman and Zhong (2002) on S&P 500 index futures. Table 9: Granger Causality Test of Effect of Volume and Open Interest on Spot Volatility by Applying trivariate VAR Model for Causal Relations A. Spot volatility and Volume B. Spot volatility and Open Interest Volume--> Spot volatility Spot volatility--> Volume Open Interest--> Spot volatility Spot volatility--> Open Interest Agricultural Soy Bean 3.77# 5.5 24.67* 0.63 Maize 22.62* 0.6$ 0.55$ 5.82 Castor Seed 39.2* 2.4.49 5.48 Guar Seed 70.33* 8.07* 5.79 9.94* Bullion Gold 55.52* 8.35 37.32* 3.79 Silver 49.73* 8.09.43# 3.76 Metals Aluminium 35.37* 8.97* 9.57$ 3.63# Copper 8.96* 7.04 22.62* 3.4 Zinc 86.42* 5.22 5.49 4.76 Energy Crude Oil 39.33* 9.94$ 0.84$ 8.84 Natural Gas 5.5* 7.5 3.42 9.78* *, #, $ significant at, 5 and 0% level respectively The results of the lead-lag relationship between spot volatility and open interest indicate that unexpected open interest also cause spot volatility of all commodities except castor seed, guar seed, zinc and natural gas while spot price volatility causes open interest in only three commodities namely guar seed, aluminium. These results support findings of Yang (2005) on 4

agricultural commodities but do not support the argument in Chen, Cuny and Haugen (995) that a change in S&P 500 spot price volatility may cause a change in futures open interest. The sign results of the Granger causality test is presented in Table 0. Panel-A of Table 0 reports the spot volatility equation whereas Panel-B and Panel-C report trading volume and open interest equation respectively. As shown in Panel-A of the Table 0, it is found that in most of the cases unexpected trading volume cause an increase in spot price volatility expect natural gas where effect is negative. In the cases where unexpected open interest causes spot price volatility, we find the mixed results. The unexpected open interest causes an increase the spot volatility for soybean, crude and copper, whereas for maize, gold, silver, and aluminium, open interest has negative effect on spot volatility. In the cases where causality runs from spot volatility to unexpected trading volume or from spot volatility to unexpected open interest, the sign in not clear because of multiple alternative positive and negative significant lagged parameters of volatility in volume or open interest equations. As suggested by Sims (972, 980) and Abdullah and Rangazas (988), we also performe forecast error variance decomposition and impulse response analysis to produce some insights beyond the statistical significance of Granger causality tests and VAR estimates. 3.2.2 Variance Decomposition Under VAR specification, the variance decomposition explains the relative impact of one variable on another variable. This analysis measures the percentage of the forecast error of one endogenous variable that is explained by other variables. Based on estimated VARs, the variance decomposition and orthogonalized Impulse response function of forecast error are estimated for trivariate case with spot volatility, unexpected trading volume and unexpected open interest as variable of interest. Results of the variance decomposition for volatility, volume and open interest are shown in Table. Panel-A of the Table explains the percentage variation in spot volatility explained by trading activity including unexpected volume and open interest whereas Panel-B reports the variation in trading activity explained by spot volatility. As shown in Panel-A of Table, the percentage of variation in spot price volatility explained by unexpected trading volume is more than 5% for eight out of eleven commodities [e.g. maize 20%, castor seed 7%, Guar seed 30%, gold 8%, silver 2%, Aluminum 0%, copper 2%, and zinc oil 3%]. In case of soybean, and energy commodities, trading volume (unexpected) explains only % of the variation in spot volatility. These results are consistent with the findings of Granger causality test. Contrarily, spot price volatility only explains about -2% variation in the unexpected trading volatility for all commodities. Our results support the findings of Yang (2005) on agricultural commodities. The results in panel B of Table show that the unexpected open interest is not able to explain more than -5% variation in spot volatility with the highest explanatory power of 5% for Guar seed. Also, spot price volatility accounts for little of the variation (-2%) in unexpected open interest. Similar results are found by Yang (2005) on agricultural commodities. These results indicate that the evidence of causality running from unexpected open interest to spot price volatility and spot price volatility to unexpected open interest may be misleading because of lack of economic significance. 5

Table 0: VAR Parameter Estimates of Spot Volatility, Unexpected Trading Volume and Unexpected Open Interest. A) Volatility Commodity CON A_ A2_ A3_ A4_ A5_ A_2 A2_2 A3_2 A4_2 A5_2 A_3 A2_3 A3_3 A4_3 A5_3 Soy Bean.240* 0.403* -0.06 0.07 8 0.08 2.332* 0.840-0.55 0.489 0.74 0.223.976* -3.296-0.906 -.626 Maize 4* 0.864* 0.032 0.044 0.20* -0.3* 0.48* 7* -0.023$ 0.023-0.025# -0.98$ -0.025-0.278* 0.25# 0.048 Castor Seed 0* 0.962* -0.06-0.09 0.0 9 0.030* 0.023* 8-0.0 0.02-0.037-0.036-0.033 Agricultural Guar Seed 0.208* 0.983* -0.072$ -0.045 0.30* -0.062# 0.676* 0.47* 0.023 0.332* 0.264* -0.709* -0.535* -0.39$ -0.457# -0.687* Gold 0.022#.004# -0.06 0.9# -0.082$ 8 0.047* 0.072* 0.0 0.00-9 -0.399* 0.96# -0.042-4 -0.096 Bullion Silver 0.943* 0.769* -0.3 0.69* -0.26* 0.037 0.379.20* 0.584# -0.326 0.50# -3.78* 0.243-0.374-0.69 0.090 Aluminium 5* 0.996* -0.6# 0.036 0.225* -0.67* 3* 0.062* 0.024# 2 0.06$ -0.44-0.035-0.230* 0.089-0.037 Copper 0.0# 0.909* 0.09-0.023-7 0.08# 0.023 0.323* 0.6* 0.069$ 0.05-0.070 0.507* -0.504* -0.00 0.075 Metals Zinc 0.804* 0.764* 0.# -0.065 0.92* -0.55* 0.20* 0.34* -0.04-0.068-0.039-0.255-4 -0.400-0.42 0.343 Crude Oil 0.49*.037* -0.08-0.096# 0.040-5 7 0.09* -0.036# -0.03$ 0.06-6 0.49# 0.024-0.07-0.8$ Energy Natural Gas 0.429#.99* -0.323* 0.049 0 0.038-0.32 0.48-0.448# -0.208-0.024 0.305 0.524 0.869-0.248 0.635 *, #, $ significant at, 5 and 0% level respectively B) Volume Commodity CON A_ A2_ A3_ A4_ A5_ A_2 A2_2 A3_2 A4_2 A5_2 A_3 A2_3 A3_3 A4_3 A5_3 Soy Bean -4 0 - - 2# -0.0 0 7 7 0.020 0.377* 0.7-0.278* -0.6-0.079 Maize 0.023 0.47-0.088-0.286# 0.24$ -0.02 0-0.028-0.035 0.026 7.02* -0.28 5 0.283 - Castor Seed 0.080 0.203-0.275 9-8 -0.07-0.022-0.039-0.025-0.03 0.05 0.659* 0.46* 0.598* 0.090 0.295$ Agricultural Guar Seed -0.026-6* # 0.0-0.08 0.03-0.25* 0.042 0.04 0.046-0.03 0.62* 0.42-0.038-0.27-0.6$ Gold 0.03 0.074-0.20 0.244$ -0.63 0.030-0.045 0.046-7$ 0.067# -0.088*.037* 0.242 0.98 0.455# 0.32 Bullion Silver 0.03 8$ -8-3 -5 4 2 0.087* 5* 0.8 0.46* 0.205 0.045 0.063 0.48 0.24 Aluminium 0.5$ -0.240$ 0.42-0.42 0.250-0.077-0.205* -5-8 -0.080-0.063$ -0.28 0.96 0.505$ 0.528# 0.408* Copper 0.042 0.03 0.020-0.025 3-0.038-0.026 0.024-0.075# 0.035-0.4* 0.472* 0.34# 0.33# -0.08-0.088 Metals Zinc 0.69-0.028 0.026 0 0.027-5 0.07-4 0.69* -0.096# -0.42* 0.09 0.482# -0.34-0.28-0.066 Crude Oil 0.49# -0.04 0.072 8-0.00-0.075-0.036 0.043-0.092* 5* -0.49* 0.063 0.42* -0.074 0.27# 0.099 Energy Natural Gas 0.025-0.05$ 0.05-0.08 0.02 4 5 9-0.034 0.04-0.068 0.086-0.230-0.276 0.296$ *, #, $ significant at, 5 and 0% level respectively C) Open Interest Commodity CON A_ A2_ A3_ A4_ A5_ A_2 A2_2 A3_2 A4_2 A5_2 A_3 A2_3 A3_3 A4_3 A5_3 Soy Bean -4 0 0 0 0 0 0.05$ -8 9-3 0.03-4 -0.049$ -0.028-0.09-7 Maize 6 0-7 -6-2 0.06 0.0* 0.03* 0.04* 7$ 4-0.040-0.028-0.020-0.09-0.026 Castor Seed -0.05 0.030-0.048-9 0.047 0.04# 0.05# 7 0-8 -0.02-0.05-0.02-0.06-0.07 Agricultural Guar Seed 4-0.0# 8 6 2-0.00# -0.078* 4-0.023$ -3-0.035* 0.095* 0.09 0.027-0.025 8$ Gold 8$ -4 0.02-0.023 7-5 -6-2 -7 6 0.05-3 7 0.07-0.06 Bullion Silver 4 0 0-0 -6 2-0.02# -3 - -3 0.00-0.04-0.09 Aluminium 0.05-0.04-0.039$ 0.043# -0.020 0.022-8$ -6 7 5 6 0.077 0.358* 0.088# 0.08 0.03 Copper 0.02-9 6 6-6 2-0.09* 0.02-6 0.00 7 8-6 0.09-0.02-0.0 Metals Zinc 0.0 5 - -6-0.022* -6 2 0.0-3 - 5 0.029 7-0.033 Crude Oil 5-0.025 5# -0.0-0.020 - -5-0.05$ -9-0.020# -0.08-0.02 9-7 3 Energy Natural Gas 0.02# 3-4 6$ -3-3 -0.030 6 0.08$ 5-2 -0.09-9 -0.023-0.022-0.09 *, #, $ significant at, 5 and 0% level respectively 6

Table : Forecast Error Variance Decompositions Trivariate VAR A. Spot volatility Explained By Futures Trading activity B. Futures Trading activity Explained By Spot volatility Volume Open Interest Volume Open Interest 5 0 5 20 5 0 5 20 5 0 5 20 5 0 5 20 Soy Bean 0% % % % % 0% 2% 2% 2% 2% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% Maize 0% 9% 2% 22% 22% 0% 0% 0% 0% 0% % 2% 2% 2% 2% 0% 0% 0% 0% 0% Castor Se 0% 7% 8% 9% 9% 0% 0% 0% 0% 0% 0% % % % % 0% 0% 0% 0% 0% Guar Seed 0% 7% 30% 34% 35% 0% 2% 4% 5% 6% % 2% 2% 2% 2% % % % % 2% Gold 0% 7% 9% 9% 9% 0% % % % % 0% 0% 0% 0% 0% 0% % % % % Silver 0% 5% % 3% 3% 0% % % % % 0% % % % % 0% 0% 0% 0% 0% Aluminium 0% 8% 0% 0% 0% 0% % 2% 2% 3% 0% % % % % 0% 2% 3% 3% 4% Copper 0% 9% 2% 3% 3% 0% 0% % % % 0% 0% % % % 0% % % % % Zinc 0% 3% 3% 3% 3% 0% 0% 0% 0% 0% % % % % % 0% 0% % % % Crude Oil 0% % % % % 0% 0% 0% 0% 0% 0% 0% 0% % % 0% % % % % Natural Gas 0% 0% % % % 0% 0% % % % 0% % % % % 0% % 2% 2% 3% 7

To conclude, the forecast error variance decomposition results between unexpected trading activity and spot price volatility suggest that the unexpected trading volume has considerable influence on spot price for all agricultural and non-agricultural commodities. However, the unexpected open interest has negligible or very less effect on spot price volatility. By contrast, the spot price volatility has no influence on futures trading activity including unexpected trading volume and open interest. 3.2.3 Impulse Analysis The impulse-response function simulates the effect of a shock to one variable in the system on the conditional forecast of other variables. We use the impulse-response function to analyze the impact of change in trading activity on spot volatility and vice versa. As shown in figure 2 (a), it is found that the speed of adjustment in spot volatility to its own volatility shock is high. For agricultural commodities the speed of adjustment is faster than nonagricultural commodities and stabilizes earlier. Spot volatility adjustment to shock in the unexpected volume is positive for all commodities except natural gas. In case of guar seed, copper, silver and natural gas, the external shock in the unexpected volume on spot volatility do not stabilize after 20 days. The response of spot volatility to open interest shock is mostly negative for all commodities except, soy bean, copper, and natural gas. In case of gold and crude oil the responses are minimal. Results of the impulse-response function are consistent with the Granger causality result and are conclusive in terms of sign effect which is not obvious from the results of VAR model. Interestingly, volatility shock in most of the commodities produces both positive and negative adjustment in trading activity including unexpected volume and unexpected open interest. In all commodities, the volatility shock to volume or open interest stabilizes after 4-5 days. It is important to note that it takes longer time (2-5 days) for volume shock on spot volatility to stabilize while it takes few days (4-5) for the spot volatility shock on volume to die down. The results of the impulse-response function support the findings of Granger causality test and variance decomposition results. We find that in (almost) all commodities, the spot volatility is positively affected by lagged unexpected volatility and this effect is persistence in nature. The effect of open interest is mostly very minimal and short lived. The effect of spot volatility affects on trading volatility is reciprocal. It supports the alternate overreaction and under reaction of participants on volatility shock. 4. Conclusions This study investigates, the contemporaneous and dynamic relationship between spot market volatility and futures trading activity including futures trading volume and open interest in an emerging commodity derivatives market context with an example of Indian commodity derivatives market. We examine this issue for wide variety of commodity including agricultural commodities, metals, precious metals and energy commodities. For contemporaneous relationship between spot volatility and futures trading activity, the GARCH method is applied and VAR modeling approach is used for lead-lag relationship. 8

3.90.90 9.90 7.90 5.90 3.90.90 - Soy Bean Maize Guar Seed Castor Seed.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 Aluminium Copper Zinc 4.0 3.5 3.0 2.5 2.0.5.0 0.5 0.0 Gold Silver 3.50 3.00 2.50 2.00.50.00 0.50 Crude Oil Natural Gas a) Volatility 2.00.50.00 0.50-0.50 Soy Bean Maize Guar Seed Castor Seed 0.50 0.40 0.30 0.20 Aluminium Copper Zinc.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0. 0.0-0. Gold Silver - -0.20-0.30-0.40-0.50 Crude Oil Natural Gas b) Volume 2.00.50.00 0.50-0.50 -.00 Soy Bean Maize Guar Seed Castor Seed R esponse - - Aluminium Copper Zinc 0.2 0. 0.0-0. -0.2-0.3-0.4-0.5 Gold Silver 0.50 0.40 0.30 0.20 - Crude Oil Natural Gas c) Open Interest Figure 2 (A): Impulse Function for Spot Volatility to Shock in a) Volatility, b) Volume and c) Open Interest Shock 9

- -.00 0.80 0.60 0.40 0.20-0.20 0.20 0.5 - Soy Bean Maize Guar Seed Castor Seed Soy Bean Maize Guar Seed Castor Seed Soy Bean Maize Guar Seed Castor Seed R esponse - Aluminium Copper - Zinc 0.90 0.70 0.50 0.30-0.30 - - Aluminium Copper Zinc Aluminium Copper Zinc - - a) Volatility.00 0.80 0.60 0.40 0.20 b) Volume Gold Silver Gold Silver -0.20 - c) Open interest Gold Silver - -.00 0.80 0.60 0.40 0.20 Crude Oil Natural Gas Crude Oil Natural Gas -0.20 - Crude Oil Natural Gas Figure 2 (B): Impulse Function for Futures Volume to Shock in a) Spot Volatility, b) Volume And c) Open Interest Shock 20

0.02 0.0 Soy Bean Maize Guar Seed Castor Seed 0.02 0.0 Aluminium Copper Zinc 0.02 0.0 Gold Silver 0.02 0.0 Crude Oil Natural Gas -0.0-0.0-0.0-0.0-0.02-0.02-0.02-0.02 a) Volatility 0.08 0.06 0.04 0.02-0.02-0.04 Soy Bean Maize Guar Seed Castor Seed 0.03 0.02 0.0-0.0-0.02 Aluminium Copper Zinc 0.02 0.0-0.0-0.02 Gold Silver 0.03 0.02 0.0-0.0-0.02-0.03 Crude Oil Natural Gas b) Volume 0.5 Soy Bean Maize Guar Seed Castor Seed 0.5 - Aluminium Copper Zinc 0.5 - Gold Silver 0.5 Crude Oil Natural Gas - - c) Open interest Figure 2 (C): Impulse Function for Futures Open Interest to Shock in a) Spot Volatility, b) Volume And c) Open Interest Shock 2

We find that there exists a positive contemporaneous correlation between both the expected and unexpected trading volatility for non-agricultural commodities. In case of agricultural commodities only unexpected volume is positively related with spot volatility. The effect of both expected and unexpected open interest on contemporaneous spot volatility is insignificant for most of the commodities. Our results are contradictory to the results of Bessembinder and Seguin (992) on equity market. They found that the expected trading activity, both futures trading volume and open interest stabilize the spot market by reducing the spot market volatility. Results of the lead lag relationship between spot volatility and futures trading activity (unexpected) suggest that in most of the commodities the unexpected futures trading volume causes spot price volatility. This result is confirmed through Granger causality test, forecast variance decomposition and impulse response function. The effect of unexpected volume on spot volatility is positive and persists for many days. In case of unexpected open interest, we find weak causal relationship with spot volatility. Our results are consistent with findings of Yang (2005) on agricultural commodities. In some cases, we do find significant causality running from spot volatility to unexpected trading volume or spot volatility to unexpected open interest. However, they lack economic significance as analyzed through forecast variance decomposition and impulse response function. Our results are consistent with findings of Chatrath, Ramchander and Song (996), and Adrangi and Chatrath (998) on currency futures but are contradictory to studies of Chen, Cuny and Haugen (995) Darrat and Rahman (995) and Darrat, Rahman and Zhong (2002) on stock index futures markets. Our findings on commodity derivatives markets are similar with others, who found that the futures trading activity increase spot market volatility. This, however, begs the question as to whether it is the case merely because the presence of markets results in flow of information taking place in developed futures market or because of trading activity of uninformed speculation in the futures market.

References Abdullah, D.A. and Rangazas, P.C. (988). Money and the Business Cycle: Another Look, Review of Economics and Statistics, 70, 680 85. Adrangi, B., Chatrath, A. (998). Futures commitments and exchange rate volatility, Journal of Business Finance and Accounting, 25, 50 520. Antoniou, A., Holmes, P., and Priestley, R. (998). The effects of stock index futures trading on stock index volatility: An analysis of the asymmetric response of volatility to news. Journal of Futures Markets, 8, 5-66. Bessembinder, H., Seguin, P.J., (992). Futures trading activity and stock price volatility, Journal of Finance 47, 205 2034. Bessembinder, H. and Seguin, P. J. (993). Price volatility, trading volume, and market depth: evidence from futures markets, Journal of Financial and Quantitative Analysis, 28, 2 39. Board, J., Sandmann, G., and Sutcliffe, C. (200). The effect of futures market volume on spot market volatility, Journal of Business finance and accounting, 28, 799 89. Brown-Hruska, S., & Kuserk, G. (995). Volatility, volume and the notion of balance in the S&P 500 cash and futures markets, Journal of Futures Markets, 5, 677 689. Chang, E. C., Cheng, J. W., and Pinegar, J. M. (999). Does future trading increase stock market volatility? The case of the Nikkei stock index futures markets, Journal of Banking and Finance, 23, 727 753. Chang, E., Chou, R.Y., Nelling, E.F. (2000). Market volatility and the demand for hedging in Stock Index Futures, The Journal of Futures Markets, 20, 05 25. Chatrath, A., Ramchander, S., Song, F. (996). The role of futures trading activity in exchange rate volatility, The Journal of Futures Markets, 6, 56 584. Chen, N., Cuny, C.J. and Haugen, R. A. (995), Stock Volatility and the Levels of the Basis and Open Interest in Futures Contracts Journal of Finance, 50, 28 300. Clifton, E.V. (985). The currency futures market and interbank foreign exchange trading, The Journal of Futures Markets, 5, 375 384. Cox, C. C. (976). Futures trading and market information, Journal of Political Economy, 84, 25 237. Damodaran, A. (990). Index futures and stock market volatility, Review of Futures Markets, 9, 442 457. - 23 -