Evidence of informed trading in Single stock futures market in the national stock exchange of India Ltd.
This paper examines the role of Single Stock Futures (SSF) market in the National Stock Exchange(NSE) of India ltd. in facilitating trading for investors with better information with respect to the fundamentals of securities traded. Given the leverages of trading in the derivatives market, the informed traders are expected to migrate to the futures market. In that case, the direction of causality would be from futures market to spot market. This causal relationship between spot and futures market is analyzed by using the Granger causality test. The results, in general, show that SSF market leads spot market and trading volume of futures market contains useful information for market participants.
It has been suggested by Black (1975) that investors acting upon the fundamental values of the stocks may shift their investment avenue from the stock market to the derivatives market due to economic incentives provided by reduced transaction costs, capital requirements, and trading restrictions. Copeland (1976) considers volume as a proxy for information arrival. He assumes existence of an initial equilibrium where they all possess an identical set of information. Then allow a single piece of news to be generated. As each individual receives it, he reacts by shifting his demand curve. Finally, when all individuals have received the news, they once again possess an identical set of information and a new equilibrium is established. Anthony (1988) hypothesizes that if information trading takes place predominantly in the option market, one would expect to observe a lead-lag relation for the options/shares trading volumes. This proposition can be easily extended to the SSF market as the SSF is based on the common shares like stock options.
If there are alternative markets in which informed traders can profit from their information, then where informed traders choose to trade may have important implications not only for security price movements, but for the behavior of related prices as well. This suggests that transactions in derivative markets may be an important predictor of future security price movements. Purchasing a futures contract is futures trade that would be used by traders informed of good news, and it can be called as positive futures trade. Selling a futures contract could be termed as negative futures trade, as they would be entered by traders acting on the bad news. In a pooling equilibrium a positive option trade provides a positive signal to all market makers, who then would increase their bid and ask prices. Similarly, a negative option trade depresses quotes. Because market participants can learn from trades in both markets, option trades would affect subsequent behavior of the stock market. Easley et al says that it is because of the option trading behavior of informed traders that the volumes of particular types of option trades have predictive power. If, instead, options are used only as hedging vehicles, then all option trades would be liquidity-based (i.e., un-informative) and we would not expect to find any link between option volumes and stock prices.
The asymmetric information model in market microstructure elaborates that the migration of informed trader causes a reduction in the proportion of informed traders in the underlying security market. Consequently, the competitive risk neutral market maker will reduce bid-ask spread due to lower adverse selection costs in trading with informed trading. Narrower spread means declining volatility since transaction price lies within the spread.
This paper investigates the preference of the informed traders for trading in the SSF market by analyzing the trading volume in the spot and futures market. It is well known that the information asymmetry exists in the financial market. Therefore, following the sequential information flow hypothesis, stock futures volume increases should precede volume increase in the spot market, surrounding information release. In this sense, SSF trading is expected to lead trading in spot market.
The rest of the study is organized as follows. Section 2 gives a brief note on the Sequential Information Arrival Hypothesis. Section 3 reviews the previous literature related to this topic. Nature of the data and research methodology are explained in section 4. Section 5 presents the empirical results followed by the conclusion.
II. SEQUENTIAL INFORMATION ARRIVAL HYPOTHESIS
The sequential information arrival model developed by Copeland (1976) offers a framework for the study of this dynamic adjustment process. The sequential information arrival process begins with the asset market in equilibrium. A single item of information then arrives at the market. In previous informational studies using equilibrium analysis, all market participants are assumed to become informed simultaneously. The sequential information arrival model assumes that only one trader observes the information initially. This trader interprets the news, revises his beliefs, and trades to arrive at a new optimal position. The outcome of this series of events is the generation of transaction volume and a new equilibrium price. After the market arrives at this. new equilibrium, the next investor becomes informed and, after a similar sequence of events, a second temporary equilibrium is achieved. This process continues until all traders are informed and results in a series of momentary equilibria. When the last trader receives the information, the market reaches a final equilibrium. The sequential process allows one to observe the path of trades, prices, and volume. In addition this model provides a more realistic model for most information events. The sequential information arrival model does not change the capital asset pricing literature in any way. Instead, it adds to it by giving a better understanding of the parameters which affect volume as well as its relationship with price changes.
III. PREVIOUS WORK
Previous researches on the informed trading are mainly based on the option markets. Anthony (1988) had analyzed the interrelation of stock and option market trading volume in the context of shares listed in the NYSE and options in the CBEO. By using Granger causality methodology, he had identified that informed trading takes place in option market as trading volume of call options leads trading in the underlying shares, with a one-day lag. Easley et al. (1998) had attempted to examine the informational links between options markets and equity markets by using option volume and stock prices. Their empirical testing reveals that stock prices lead option volume implying that options market is dominated by hedgers. They also found that particular option volumes lead stock prices. This result is strongly consistent with option markets being a venue foe information-based trading. Thus, their work recognizes volume playing a key role in the process in which markets become efficient.
Chakravarty et al (2004) have applied the Hasbrouk's (1995) methodology to time series of stock prices and implied option prices to measure the relative share of price discovery occurring in the stock and option markets. They found a significant price discovery in the option market. They find evidence that the proportion of information revealed first in the option market varies across stocks. Option markets tend to be more informative on average when option trading volume is high and when stock volume is low, when option effective spreads are narrow, and when stock spreads are wide. Manster and Rendleman (1982) used Granger causality test to identify the lead-lag structure between derivative prices and stock prices. They hypothesized that if informed trader migrated to derivatives markets, the derivative prices should predict the underlying stock prices. They have empirically found that informed trading takes place in the derivative markets. Stephan and Whaley (1990) investigated the interrelationship between option and stock prices using intra-day data and found no causal relationship between them. The evidence that derivative prices lead the stock prices does not always imply superior information in derivative market over the stock market. There can be noise trading in the derivative markets. Khanthavit (1996) separated noise trading from the informed trading by observing the path of prices after trading. Information based trading will cause permanent price changes whereas the effect of noise trading on prices will be temporary. Vijh (1990) tested the informed trading by investigating option volume and option price. He did not find any significant relationship between the two variables and concluded that option trading is not information based. Boluch et al (1997), in their study based on the selected CBOE option volume and underlying stock prices, found feedback trading in both markets. Adjustments in one market is quickly reflected in the other market and it appears that neither market can be used as a benchmark to predict activity in the other
IV. DATA AND METHODOLOGY
The required data of daily frequency for this study is collected from the official website of NSE. The data set consists of 28 stocks on which SSF contracts started trading on November 9, 2001. Daily volume of both underlying stocks and SSF contracts are used to analyse the causal relationship between them. Daily turnover in both markets are considered as a proxy for volume. The study covers the period from November 9, 2001 through the end of May, 2007.
This study uses the methodology adopted by Anthony (1988) who analysed the relationship between stock and options volume listed in the Chicago Board of Option Exchange (CBOE) by using Granger causality test. The choice of the lag length of each series in the test is selected based on the Akaike information criterion. If there is no informed trading in the Single Stock Futures markets in the NSE, then trading in the futures market would not cause the trading activities in the cash market. However, according to Granger and Newbold (1986) if two time series are taken at the actual levels, such time series may not be white noise. This may cause large cross-correlations, thereby leading to spurious results. To avoid this possibility, Granger and Newbold suggested that the time series be initially identified and estimated. Following Anthony (1988), the stock and option volume time series are fitted with an autoregressive integrated moving average [ARIMA (p, d, q)] model and used the residuals from these models as inputs in the Granger causality model. In doing so, it is ensured that the series are uncorrelated since the residuals indicate only unexplained variations in the data. The next stage in the ARIMA (p, d, q) modeling is the estimation of the just identified model from the available data. After having estimated the tentative model, as final step in the modeling, the diagnostic checking of the estimated model is undertaken so as to ensure the adequacy of the model. A model is considered to be accurately specified if the residuals from the estimated model are serially uncorrelated or stationary. If they are found to be non-stationary, it means that the estimated model does not fit the data properly so that a new model has to be specified. Thus, ARIMA modeling is an iterative process of identifying the underlying process in which a particular data series has been generated or evolved.
V. EMPIRICAL RESULTS
As noted earlier, the empirical analysis begins with identification and estimation of appropriate ARIMA models for the stock and futures volume. This step is necessary to extract residuals from the models which form the inputs for the Granger causality test. Table 1 presents the values of p, d, q for the ARIMA models for the sample companies.
It is shown that most of the time series on stock volume and futures volume are generated by an MA process of varying order. However, most of the time series follows an MA (1) order. There are five time series (companies) following ARIMA process with low order. The order of integration in all time series is unity implying that all of' them are non-stationary in the level form with high serial correlation and first differencing will render them stationary. The adequacy of the series identification and estimation was tested by analyzing the serial autocorrelation of the residuals from the estimated models. The results show that residuals are serially uncorrelated so that the estimated models fit the data well.
The next step in the empirical analysis is to establish the direction of the causality between stock and futures market so as to corroborate the preference of informed traders. The causality test is conducted by using pre-whitened residuals from the models fitted to the data at the series identification stage. The results of the Granger causality test are given in table 2. In column 2 of the table, the upper value of the cell refers to the hypothesis that the futures volume do not Granger cause the stock volume. The lower value refers to the hypothesis that the stock do not Granger cause the futures volume.
The most important thing to be noted in the result is the rejection of the null-hypothesis that stock futures volume do not Granger cause the stock volume in the case of 20 companies out of 28 sample companies. It is quiet evident that F-statistics of these 20 are statistically significant at varying levels of confidence. Therefore, it is concluded that the direction of causality runs from the single stock futures market to the stock market and informed trading takes place in the futures market. This result is consistent with the findings of Easley et al (1998). In other words, the trading activity in *,he stock futures market leads the trading activity in the stock market. The remaining stocks in the sample stocks show either feed-back trading or existence of no causality. Feedback relationship is surprising even after the pre-whitening of the data series. However, it shows the existence of instantaneous causality.
This study investigates the informational link between stock and single stock futures markets in the NSE of India. The informed traders may take positions in the futures market due to the leverages available there so that stock market is expected to follow it. This study used the daily trading volume in both stock and stock futures markets as proxy for trading activity following Copeland's (1976) proposition. Following Anthony (1988, Granger causality methodology is adopted to determine the preference of informed traders. The overall results show that the informed trading takes place in the single stock futures market as the causality runs from the futures volume to the stock volume in the case of majority of stocks. The most important implication of this study is that volume plays a key role in making the market efficient as it contains information useful for investors. Therefore, the relevance of volume as an informant in the capital market with various investment avenues must be recognized.
Anthony, J. (1988), "The Interrelationship of Stock and Option Market Trading Volume Data," The Journal of Finance, Vol.43, No.4, 949-964
Ang, J., and Y. Cheng (2005), "Single Stock Futures: Listing Selection and Trading Volume," Finance Research Letters, Vol. 2, No. 1, 30-40.
Black, F. (1975), "Fact and Fantasy in the use of Options," Financial Analysts Journal, Vol. 31, No. 4, 36-41.
Boluch, M., and T. Chamberlain (1997), "Option Volume and Stock Price Behavior: Some Evidence from the Chicago Board Options Exchange," Atlantic Economic Journal, %7ol. 25, Issue 4,358-370.
Chan, K., Y. Chung and W. Fong (2002), "The Information Role of Stock and Option Volume," The Review of Financial Studies, Vol. 15, Issue 4, 1049-1075.
Cox, C. (1976), "Futures Trading and Market Information," The Journal of Political Economy, Vol.84, No.6, 1215-37.
Easley, D., M. O' Hara and P. Srinivas (1998), "Option Volume and Stock Prices: Evidence on Where Informed Traders Trade," The Journal of Finance, Vol. 53, Issue. 2,431-65.
Granger, C. (1969), Investigating Causal Relations by Ecolmmetric Models and Cross Spectral Models, Econometrica, Vol. 37, Issue 3, pp. 424-38.
Granger, C. W. J and P. Newbold (1986), Forecasting Economic Time Series, New York: Academic press.
Hasbrouck, J. (1991), "Measuring the Information Content of Stock Trades," The Journal of Finance, Vol. 46, Issue. 1, 179-207.
Hasbouch, J. (1991), "The Summary of Informativeness of Stock Trades: An Econometric Analysis," The Review of Financial Studies, Vol. 4, Issue. 3, 571-95.
Jennings, R., L. Starks, and J. Fellingham (1991), "An Equilibrium Model of Asset Trading with Sequential Information Arrival," The Journal of Finance, Vol. 36, Issue. 1, 143-61.
Khanthavit, A. (1996), "Information Structure, Trading Behavior and Stock Prices in Thai Market," working paper, Thammasat University.
Manster, S. and R. J. Rendlemen, Jr (1982), "Option Prices as Predictors of Equilibrium Stock Prices," The Journal of Finance, Vol. 37, Issue. 4, 1043-1057.
Stephan, J., and R, Whaley (1990), "Intraday Price Change and Trading Volume Relations in the Stock and Option Markets," The Journal of Finance, Vol. 45.No. 1, 191-220.
Copeland, T. (1976), "A Model of Asset Trading Under the Assumption of Sequential Information Arrival," The Journal of Finance, Vol. 31, Issue. 4, 1149-68.
Vijh, A. (1990), "Liquidity of the CBOE Equity Options," The Journal of Finance, Vol. 45, Issue. 4, 1157-1179.
ANVER SADATH AND BANDI KAMAIAH
University of Hyderabad, Hyderabad
Table 1 Identification of Arima (p, d, q) for Stock and Futures Volumes Stock Futures Volume Volume Company p d q p d q ACC 0 1 1 0 1 1 BAJAJAUTO 0 1 1 0 1 1 BHEL 0 1 2 2 1 2 BPCL 0 1 1 1 1 1 BSES 0 1 2 0 1 2 CIPLA 0 1 2 0 1 1 DRREDDY 1 1 1 1 1 1 GRASIM 0 1 2 0 1 1 GUJAMBCEM 0 1 1 0 1 1 HDFC 0 1 1 1 1 1 HINDPETRO 0 1 3 0 1 3 HINDLEVER 0 1 1 0 1 1 INFOSYS 0 1 1 0 1 1 ITC 1 1 1 1 1 1 L&T 0 1 1 0 1 1 M&M 0 1 1 0 1 1 MTNL 0 1 2 0 1 2 RANBAXY 0 1 2 0 1 1 RELIANCE 0 1 3 0 1 1 RELPETRO 0 1 1 0 1 1 SATYAMCOM 0 1 1 0 1 1 SBIN 0 1 1 0 1 1 STROPTICAL 0 1 2 0 1 1 TATAPOWER 0 1 1 0 1 1 TATATEA 0 1 2 0 1 2 TELCO 1 1 1 1 1 1 TISCO 0 1 1 0 1 1 VSNL 0 1 1 0 1 1 Table 2 Causality Between Stock and Futures Volumes Direction Company F-Statistic of Causality ACC 5.20 ** Futures to stock 2.90 BAJAJ 3.39 *** Futures to stock 1.56 BHEL 3.30 *** Futures to stock 2.58 BPCL 3.84 ** Futures to stock 1.47 BSES 3.17 *** Futures to stock 1.46 CIPLA 14.41 * Feed back 5.26 ** DRREDDY 3.38 *** Futures to stock 2.00 GRASIM 4.09 ** Futures to stock 1.16 GUJAMBCEM 3.87 ** Futures to stock 2.65 HDFC 4.24 ** Feed back 5.18 ** HINDPETRO 3.36 *** Futures to stock 2.65 HINDLEVER 4.52 ** Futures to stock 1.55 INFOSYS 1.27 No causality 2.11 ITC 4.83 ** Futures to stock 2.22 L&T 6.36 ** Futures to stock 0.87 M&M 3.14 *** Futures to stock 1.90 MTNL 3.02 *** Futures to stock 2.39 RANBAXY 1.65 No causality 1.85 RELIANCE 4.67 ** Feed back 18.59 * RELPETRO 4.02 ** Futures to stock 0.64 SATYAMCOM 3.05 *** Futures to stock 2.52 SBIN 1.68 No causality 1.38 STROPTICAL 3.10 *** Futures to stock 1.76 TATAPOWER 4.26 ** Futures to stock 1.50 TATATEA 14.91 * Futures to stock 2.02 TELCO 5.81 ** Feed back 7.14 * TISCO 0.56 No causality 0.73 VSNL 2.92 *** Futures to stock 0.64 '*', '**' and '***' respectively imply significance at 1%, 5% and 10% levels