ICRA (Chakraverty) - SSRN papers

1 downloads 0 Views 244KB Size Report
Abstract. In this paper we analyse whether the Indian Stock Index Futures market plays an important role in the assimilation of information and price discovery in ...

Contribution of Indian Index Futures to Price Formation in the Stock Market

I C R A B U L L E T I N

Money

&

Finance

F E B R U A R Y. 2 0 0 7

SUCHISMITA BOSE Abstract In this paper we analyse whether the Indian Stock Index Futures market plays an important role in the assimilation of information and price discovery in the stock market. Using Futures prices for the S&P CNX Nifty Index traded on the National Stock Exchange of India, we find that there is significant information flow from the futures to the spot market and futures prices/returns have predictive power for the spot prices. If we take account of the long run relation between the two price series we find clear bidirectional information flows or feedback between the markets. The contributions of the two markets to the price discovery process are also almost equal with the futures showing a marginal edge over the spot market, as the information flow into the stock prices from the futures is slightly higher than the price information flows to the futures market from the spot market. The futures market also readjusts faster to market-wide information and thus absorbs much of the volatility induced by flow of new information.

I. Introduction This paper examines whether prices in the Indian stock index futures market contribute to the pricing process in the stock market. In a previous article on the Indian derivatives markets we had described how financial derivatives trading has been formalised and modernised in various aspects through introduction of new instruments, electronic trading and clearing mechanisms, payment guarantees as well as close regulatory supervision (Bose, 2006). We also showed that the stock futures and options products have been witnessing increasingly large volumes of trading. This motivates us to study whether this evolving market is effectively playing its role in the price discovery in its underlying stock market. One of the core economic benefits of a market for derivative products of any asset class is the additional information content that may be extracted out of the prices evolving in this market. Thus it is said that besides the traditional role of risk sharing assigned to futures markets, these markets also play an important role in the aggregation of information and price discovery. Price discovery refers to the process through which markets converge towards the efficient price of the underlying asset. At any point in time there is a flow of new information into asset markets and market prices for the assets concerned

Using Futures prices for the S&P CNX Nifty Index traded on the National Stock Exchange of India, we find that there is significant information flow from the futures to the spot market and futures prices/ returns have predictive power for

Electronic copy available at: http://ssrn.com/abstract=1017477

the spot prices.

39

I C R A B U L L E T I N

Money

&

Finance

F E B R U A R Y. 2 0 0 7

Historically it has been observed that the futures market has a greater speed of assimilation of new information that comes to the market and hence has predictive power for future price movements in the underlying asset.

40

readjust to such news flows.1 Theoretically when two markets for the same asset are faced with the same information arriving simultaneously, the two markets should react at the same time in a similar fashion. If the two markets do not react at the same time, one market will then lead the other. When such a lead-lag relation appears in case of price adjustments, the leading market is viewed as contributing a price discovery function for that sector. Historically it has been observed that the futures market has a greater speed of assimilation of new information that comes to the market and hence has predictive power for future price movements in the underlying asset. Futures markets usually incorporate new information more quickly than the spot markets primarily because of their inherently high leverage and low transaction costs.2 The contribution of each market to price discovery depends, at least in part, on the microstructure of these markets including the level of transparency, the liquidity supply mechanism, the rules governing the priority of orders, the constraints on short sales and the settlement mechanism (Tse, 1999). A number of studies have empirically examined the temporal relationship between futures and spot markets involving well established US, European and Asian futures markets. These studies seek to find the lead-lag relations between the futures and the spot market for an asset class and the differential speed of adjustments to flow of new information. Some studies also estimate the Information-Share of the individual markets in the price discovery process. The informationshare of a market is its contribution to price discovery, which is derived from the premise that through price adjustments both futures and spot markets contribute to the discovery of a unique and common unobservable price that is the efficient price for that asset class. Most studies support the primacy of the futures market in the price discovery process and support the notion of a slightly higher information share for this market. Some studies, however, report clear bidirectional information flow between the two markets. Understanding the behaviour and price effects of futures and spot index trading in different world markets is of great significance to brokerage houses, investors, portfolio managers, regulators, legislators and the major global stock and futures exchanges. In Box 1, we have summarised the results of some of these international studies on the relation between the spot index and index futures prices.

1 Such news can be global or domestic economic information, sectoral information or individual corporate announcements, or any information that market participants feel is relevant to the pricing of the asset. 2 In a derivatives market investors can command large resources with the help of a much smaller amount of funds compared with the underlying asset market as buying a derivatives contract costs a fraction of the amount needed to buy the security.

Electronic copy available at: http://ssrn.com/abstract=1017477

I C R A B U L L E T I N

BOX 1 Index (Country)

Money

&

Time Period

Broad Conclusions

S&P 500 and Major Market Index (MMI) (US)

Intraday (10 minutes)

S&P 500 and MMI futures lead the cash market. This is attributed to faster dissemination of information into futures market.

S&P 500 and Nasdaq-100 (US)

Intraday (1 second) March 1 to May 31, 2000

The market for US equity indexes has traditionally comprised Hasbrouck floor-traded index futures contracts and the individual (2003) markets for the component stocks. This picture has been altered by the advent of exchange-traded funds (ETFs) that mirror the indexes and electronically-traded, smalldenomination (“E-mini”) futures contracts. For the S&P 500 and Nasdaq-100 indexes most of the price discovery occurs in the E-mini markets: price changes in the E-mini futures generally lead those in the regular futures contracts and the ETFs. Trading activity across the sector ETFs varies considerably and the most actively traded sectoral ETF (technology) contributes a modest amount to price discovery in the overall index.

DJIA 30 and S&P 500 (US)

Intraday ( Time stamped) May through July 2004

This paper explores the dynamics of price discovery Tse et al between the Dow Jones Industrial Average (DJIA) index (2006) and its three derivative products: the DIAMOND exchangetraded fund (ETF), the floor-traded regular futures, and the electronically traded mini futures. Even though the American Stock Exchange is the primary listing exchange for the ETF, the analysis indicates that the electronically traded ETF on the Archipelago (ArcaEx) electronic communications network dominates the price discovery process for DIAMOND shares. The E-mini futures contribute the most to price discovery, followed by the ArcaEx DIAMOND. The DJIA index and regular futures contribute least to price discovery.

Established Markets

Reference

Finance Stoll and

F E B Whaley R U A R Y. 2 0 0 7

(1990)

Nikkei 225 (Japan) Intraday (1 minute) 10 August to 18 September 1998

The Nikkei Stock Average is cross-listed on the OSE and Simex (Japan and Singapore). Costs of trading SIMEX Nikkei 225 futures are significantly lower than trading a similar nominal exposure using OSE Nikkei 225 futures. The cost differences are attributable to lower margin requirements, minimum tick size and bid-ask spreads on SIMEX as well as the existence of negotiated brokerage commissions versus the fixed rate regime operating in Japan. While SIMEX innovations are found to strongly cause Index innovations, no such relationship is documented for OSE innovations. These results are directly attributable to the higher cost of trading on the OSE discouraging informed traders. Consequently it takes longer for the available information set to be impounded in OSE futures prices.

Frino and West (2003)

CAC 40 (France)

The price discovery process is dominated by the futures market. The aggregation of new information into prices is achieved primarily through futures trading and the stock market adjusts quickly to the new equilibrium price.

Alphonse (2000)

Intraday (30 second interval) for January 3rd, 1995 to March 31st, 1995

. . . continued on following page

41

I C R A B U L L E T I N

Index (Country)

Money MIB 30 (Italy)

&

Time Period

Broad Conclusions

Reference

Daily 1994 to 2002

Deviations from equilibrium are corrected by movements in the spot market. Cross market dynamics are more important in the short run.

Pattarin and Ferretti (2004)

Finance

AOI (300) Intraday (5 minute) (Australia) to 21 F E B R U A R Y. 230 January 07 December, 1995

Unlike many international exchanges the Australian futures Turkington market is a completely separate entity and an open outcry and Walsh market. The usual result of futures leading spot is strongly (1999) rejected, with clear bi-directional causality, and with many significant lags. This suggests that an electronic market may enhance price discovery. An index shock appears to induce a very large response in the futures; however, the reverse is not true.

AOI and Share Price Index (SPI) (Australia)

Intraday (1 minute) 1992 to 1996

There is strong equities market enhanced price the SPI futures

Hang Seng (33) (Hong Kong)

Intraday (1 minute) The Hang Seng Index Futures contributes the major share So and Tse November 12, 1999 of the information, followed by the spot market. The futures (2006) to June 28, 2002 market leads the spot in assimilating information into prices.

evidence of the futures market lagging the by 5 minutes. Automation of the ASX has discovery in the equities market relative to market.

West (1997)

Relatively New Markets Zhong et al (2004)

IPC (35) (Mexico)

Daily from April 15, 1999 through July 24, 2002

The futures market in Mexico is a useful price discovery vehicle, as futures returns have both short and long run predictive power for spot returns. Futures trading has also been a source of instability for the spot market.

ASE 20 (Greece)

Daily from August 1999 through June 2002

The results show the presence of a bi-directional causality Kenourgios between stock index spot and futures markets, indicating (2004) that the futures market serves as a focal point of information assimilation and contributes to price discovery.

S&P CNX Nifty (50) (India)

Daily from June 2000 through October 2002

The major findings are that the futures market (and not the spot market) responds to deviations from equilibrium; price discovery occurs in both the futures and the spot market, especially in the latter half of the study period. The results also show that volatility in the spot market has come down after the introduction of stock index futures.

42

Raju and Karande (2003)

Here we analyse Indian stock index and index futures prices/ returns for the past four years in order to examine if the index futures prices provide any information that contributes to the adjustment process of the stock index. This study is important as a negative outcome would imply existence of prohibitive transaction costs or institutional barriers or incorrect trading strategies which could be impeding arbitrage between the two markets and hindering efficient functioning of the derivatives market. The rest of this article is organised as follows: In Section 2 we describe the data set and explain the methodology involved in assessing whether the futures prices effectively provide additional information to the stock market, beyond what is incorporated in the spot market itself. In Section 3 we present the results from the empirical tests and discuss what they indicate. Section 4 concludes with a summary and some observations on the findings.

II. The Data and the Methodology of Analysis The Data For this analysis we use the daily closing prices of the futures contract on the S&P CNX Nifty index and the underlying index values available from the NSE. A brief description of the index and its futures contract is provided in Box 2.3 The Futures on the Nifty index were introduced in the Indian market in June 2000. Index futures trading in the Indian market had lost some lustre in the (financial) year 2001-02 when individual stock futures were introduced in the market; however trading has picked up since 2002-03 and index futures now register considerable turnover and are being more extensively used by FIIs (foreign institutional investors) too [Charts 1 & 2]. For our analysis we concentrate on data from the period March 2002 through September 2006. We choose this period as it would filter out data from the initial stages of index futures trading when the market distortions were not unlikely. This period can be regarded as a more mature phase in the Indian financial derivatives market in terms of higher liquidity and BOX 2 • S&P CNX Nifty is a well diversified 50 stock index accounting for 22 sectors of the economy. • Nifty stocks represent about 59.5% of the total market capitalisation, as on September 29, 2006. • The Nifty Futures (FUTDIX) is traded at the NSE F&O segment. • NSE now stands 4th among worldwide derivatives exchanges with 50,81,055 contracts traded in the latest month. • The futures contracts are available for trading from introduction to the expiry date, which is the last Thursday of each month. • Nifty futures contract has a three-month trading cycle—the near month (one), the next month (two) and the far month (three). A new contract is introduced on the trading day following the expiry of the near month contract. The new contract is introduced for a three month duration. This way, at any point in time, there will be three contracts available for trading in the market, i.e., one near month, one mid month and one far month duration, respectively. • There are no day minimum/maximum price ranges applicable for S&P CNX Nifty futures contracts. However, in order to prevent erroneous order entry by trading members, operating ranges are kept at +/- 10%. • Position/exposure limits are imposed so as to avoid excessive risk taking. • There are prescribed net worth requirements for clearing members to ensure that trading member obligations are commensurate with their net worth. • National Securities Clearing Corporation (NSCCL) undertakes clearing and settlement of all deals executed on NSE F&O segment. NSCCL provides the guarantee for all F&O settlements.

3

2006.

For a detailed account of the Indian financial derivative market see Bose,

I C R A B U L L E T I N

Money

&

Finance

F E B R U A R Y. 2 0 0 7

Index futures trading in the Indian market had lost some lustre in 2001-02 when individual stock futures were introduced in the market; however trading has picked up since 2002-03 and index futures now register considerable turnover and are being more extensively used by FIIs too.

43

CHART 1 Monthly Turnover in the Index Futures Market at NSE (Rs. crore)

I C R A B U L L E T I N

Money

&

Finance

F E B R U A R Y. 2 0 0 7

300000 250000 200000 150000 100000

Sep-06

Jun-06

Sep-05

Dec-05 Mar-06

Jun-05

Mar-05

Dec-04

Jun-04 Sep-04

Mar-04

Sep-03

Dec-03

Jun-03

Sep-02

Mar-02

test relates to the

Dec-02 Mar-03

0

The hypothesis we

Jun-02

50000

predictive power of CHART 2 Daily Trends in FII Derivative Trades (Rs. crore)

prices in the index futures market and 8000

contends that prices

Sell

Open Interest

6000

in the futures market

4000

are a useful

2000

predictor of

0

subsequent spot prices.

Buy

2003 Note: Source:

2004

2005

2006

The data relates to the last trading day of the month of September in each year. SEBI.

lower volatility. Also, from March 2002, FIIs were permitted to participate in all exchange traded derivative products in the Indian market, which allowed them to use the derivatives market more effectively and consequently increased the efficiency of the market as a whole. We use the Nifty futures prices of the near month contracts as these contracts are still the most liquid or most highly traded. Returns are computed by taking the natural logarithm of the ratio of the closing index levels and futures prices on successive trading days.

44

The Analysis We first try to see if the futures market prices possibly contain additional information beyond what is already embedded in the spot prices. The hypothesis we test relates to the predictive power of prices in the index futures market and contends that prices in the futures

market are a useful predictor of subsequent spot prices. The test shows whether prices in one market, say the futures, can explain the variations in the other (i.e. the spot market) with a specified time lag. If it does then we may say that the futures prices contain additional information which helps to predict prices in the spot market.4 In a reverse test, if the spot market does not show similar predictive power over the futures then we can definitely say that futures prices are useful in predicting the spot index. We need to distinguish between short-run prediction and longrun prediction. Short-run prediction implies that a given change in futures prices can only predict temporary (one time) change in spot prices.5 Clearly, long-run predictions require the presence of a long-run relation (referred to as an equilibrium relation in the literature) binding prices in the two markets. Although the index and futures price variables may drift away from equilibrium for a while, economic forces may be expected to act so as to restore equilibrium, i.e., they are expected to move together in the long run irrespective of short run dynamics. Theoretically, arbitrage activities should keep prices in the index and futures markets from diverging beyond what is entailed by differential transaction costs (and other such factors). If a departure from equilibrium occurs, prices in one or both markets should adjust to correct the disparity. Thus the price discovery function implies the presence of an equilibrium relation binding the two prices together. In other words it is said that there is a common factor, or an unobservable efficient price (also called an implicit optimal forecast index) that drives both the spot and futures prices (Hasbrouck, 1995, 2003). The second hypothesis we test for is the existence of this long run relationship.6 If the two price series are found to trend together in time one has to look into the long term equilibrium relation existing between them and test for leads and lags or predictive power under a unified framework. One needs to characterise the two series by joint models, which treat the two securities in a unified, symmetric fashion and clarify the effects of innovations or information shocks. Price discovery involves two effects: the differential reaction of the different markets to new information and the rate at which the new information is incorporated into price. If there is indeed a common factor driving the two sets of prices, any deviation from the efficient price due to new information flow in the market, should cause both prices to adjust to the equilibrium level. Consequent on the existence of 4 We use the Granger causality test, explained in Box 3, to judge this shortterm predictive power of each market over the other. 5 Such predictions though useful may suffer from the drawback that they do not take into account all of the predictable components of the prices as they disregard the long-run dynamics of the series. 6 We test for the existence of a cointegrating relation, as explained in Box 4, between the prices in the two markets to establish a long run relation between them.

I C R A B U L L E T I N

Money

&

Finance

F E B R U A R Y. 2 0 0 7

Although the index and futures price variables may drift away from equilibrium for a while, economic forces may be expected to act so as to restore equilibrium, i.e., they are expected to move together in the long run irrespective of short run dynamics.

45

I C R A B U L L E T I N

Money

&

Finance

F E B R U A R Y. 2 0 0 7

In our case we could say that the futures market leads the price discovery process if the speed

of adjustment parameter for the spot is significantly different from zero. If the model estimates of at least one of these adjustment parameters are not significantly different from zero it

an equilibrium relationship one may try to quantify the dynamics of the price changes in the two markets arising from a movement away from the above mentioned long term relationship.7 The system of equations to be estimated would need to take into account the fact that both the spot and futures prices may be influenced by past and present values of each other. Under this framework we may test for lead-lag relations between the two sets of prices. This model is augmented by the speed of adjustment parameters which would tell us about the relative speed of adjustments in each of the markets in response to a deviation from the equilibrium path. In our case we could say that the futures market leads the price discovery process if the speed of adjustment parameter for the spot is significantly different from zero. If the model estimates of at least one of these adjustment parameters are not significantly different from zero it would contradict the finding of a long run relation and the model would be suspect for misspecification. The estimates would also give us an idea about the nature of adjustments that would take place following any movement away from the long run path due to the flow of exogenous information into the market. Through a method for innovation or policy accounting the model helps to trace the time path of interactions between the two variables and understand the extent of information flow from one to the other in converging to the existing long run relation.8

III. The Findings Trends in the Spot and Futures Prices and Returns Before we present the results of our analyses we take a look at the time series of returns in the two markets. The returns from the two markets are almost similar in nature with the average return over the sample period being equal [Box 3]. The variances in the two markets differ slightly with the futures returns showing marginally higher variance. This is to be expected as the futures market is regarded as a source of price stability to the spot market since it absorbs the brunt of the price adjustments. The returns are not normally distributed, a feature common to returns data from stock markets. As expected, the returns from the two markets are significantly positively correlated, with the index and index futures returns exhibiting a correlation as high as 0.98. The high correlation between the index and the futures market suggests that both markets are related.

would contradict the finding of a long run relation.

46

7 A vector error correction model (VECM) is estimated as explained in Box 5. Under this framework the model for both prices are estimated simultaneously. This incorporates past values of the spot index into the equation for the futures price and vice versa. It includes a term called the error correction term (ECT) whose parameters indicate the speed of adjustment. 8 Impulse response functions and variance decomposition (together called innovation or policy accounting) are used to examine this relationship.

I C R A B U L L E T I N

Box 3

Money

Descriptive statistics for the Returns from Nifty and its Futures: (Period: March 2002 through September 2006) NIFTY

FUTDIX

Mean

0.000965

0.000979

Median

0.002093

0.00178

&

Finance

F E B R U A R Y. 2 0 0 7

Maximum

0.079691

0.095932

Minimum

-0.130539

-0.162581

Std. Dev.

0.014137

0.01517

Skewness

-1.043672

-1.308363

Kurtosis

12.20854

17.77548

Jarque-Bera

4286.832

Probability

0

Observations

10826.55

FUTDIX Note:

trend of the index and its derivatives

0 1154

prices suggests that the two series are

The Cross-correlation matrix:

NIFTY

A look at the time

NIFTY

FUTDIX

very closely linked

1

0.976845

and move together

1

A test for equality of the average returns in the two markets is accepted. A test for the equality of the returns variances in the two markets is rejected, showing that volatility in futures returns is significantly higher than spot returns. The two returns series are characterised by non-normality as evident from the high kurtosis and the high JB statistic.

in time. Formal testing for the existence of a long

We first look for short term lead-lag relations between returns in the two markets. The results [presented in Box 4, Panel A] show that there is strong evidence of the futures market leading the spot market with a day’s lag, while the reverse is not true. This implies that the futures prices react faster to market-wise information and contribute to price formation in the spot market. Hence we may say that the futures market indeed has short term predictive power for subsequent returns/prices in the spot market for the index. Next we move on to the question of existence of a long run binding relationship between the two markets. A look at the time trend of the index and its derivatives prices suggests that the two series are very closely linked and move together in time [Box 4, Panel B]. Formal testing for the existence of a long run relation, as expected, confirms the existence of such a relation between the prices in the spot market for Nifty and its futures market [Box 4, Panel C]. The Price Adjustment Process Since a long run relation is found to be statistically significant we can look into the dynamics of price changes in the two markets in

run relation, as expected, confirms the existence of such a relation between the prices in the spot market for Nifty and its futures market.

47

I C R A B U L L E T I N

BOX 4

Money

&

Finance

F E B R U A R Y. 2 0 0 7

Panel A We look at causality between pairs of returns to discover which market exerts a stronger influence on the other. We test for the Null Hypothesis for Granger NonCausality which posits that “Futures returns do not cause Nifty returns”. Rejecting the Null implies that futures prices lead spot prices and this is consistent with the short-run prediction hypothesis. A similar hypothesis can test for a reverse Granger Non-causality from spot to futures prices. Pairwise Granger Causality Tests Lags: 1

Null Hypothesis:

Obs

F-Statistic

FUTDIX does not Granger Cause NIFTY NIFTY does not Granger Cause FUTDIX

1153 1153

13.0333 1.89403

Probability 0.00032 0.16902

The results show that Futures returns Granger causes Nifty returns, while the reverse is not true indicating clearly that the Futures market leads the Spot market for Nifty, confirming the short run prediction hypothesis for the futures market. Note that many studies presume that a test of Granger causality implies that action in one market causes a reaction in the other. Causality that we see here is interpreted as one market reacting more quickly than the other to an outside influence or shock.

Panel B Time trend of (logarithm of) Nifty and its Futures price. F utdix

Nifty

8.2 8.0 7.8 7.6 7.4 7.2 7.0 6.8

If we plot the two series, we can see that it is unlikely that the data were generated by a stationary process. In econometric Time Series analysis, a stationary series has time independent mean, variance, and autocorrelation that are constant through time. Taking a look at the correlogram of each of the two series also reveals a partial correlation coefficient for both series of 0.99 for the first lag and of approximately

48

. . . continued on following page

I C R A B U L L E T I N

zero for the following lags. This is a strong indication for a non-stationary process. Both the augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) unit-root tests were employed to examine the stationarity property of prices. The hypothesis for stationarity for both series are rejected and we have strong evidence that both series are integrated of order one. If two series are integrated of the same order we can proceed to test for cointegration between them.

Panel C The Johansen and Juselius (1990) procedure of testing for the presence of multiple cointegrating relationships indicating the existence of common trends is applied here. In cointegration analysis one tests for the existence of the number of relationships between the variables. The number of common stochastic trends driving the series, is equivalent to the number of variables in the test minus the number of cointegrating relationships found between them. Test assumption: No deterministic trend in the data Series: LNFTY LFTDX Lags interval: 1 to 1

Eigenvalue

Likelihood Ratio

5 Percent Critical Value

1 Percent Critical Value

Hypothesized No. of CE(s)

0.068467 0.003839

86.20957 4.435146

12.53 3.84

16.31 6.51

None ** At most 1 *

*(**) denotes rejection of the hypothesis at 5%(1%) significance level. L.R. test indicates 2 cointegrating equation(s) at 5% significance level.

Thus the cointegration analysis of Nifty and its Futures prices indicates a single common factor binding the two series.

Money

&

Finance

F E B R U A R Y. 2 0 0 7

The reactions of the spot index and futures prices to the disequilibrium errors captured by the speed of

adjustment show that within one time period, which in our study is a day,

response to any deviations. The long term prediction hypothesis which posits that index futures prices impact spot price changes through the long-run price equilibrium channel is supported by these results, though the support is thin [Box 5].9 The reactions of the spot index and futures prices to the disequilibrium errors captured by the speed of adjustment show that within one time period, which in our study is a day, roughly 10 per cent of the adjustment is achieved in the spot market, while 30 per cent of the disequilibrium error is corrected in the futures market. Thus, as is to be expected, the results show that the futures market responds faster to the previous period’s deviation from the long run equilibrium. This partially reflects the fact that compared with the spot market index the futures market incorporates new information more quickly given its less restrictive nature, for example, with regard to

roughly 10% of the adjustment is achieved in the spot market, while 30% of the disequilibrium error is corrected in the futures market.

9 After accounting for the existence of the long relation the short term predictive power of futures is not supported by the results of this model.

49

I C R A B U L L E T I N

BOX 5

Money

&

Finance

F E B R U A R Y. 2 0 0 7

The Vector Error Correction Model and Speed of Adjustment Since we know that the Futures prices may be dependent on previous periods’ Nifty values and vice versa, we need to model them symmetrically. The model allows the time path of Futures prices to be affected by current and past values of Nifty and also exactly the reverse in case of Nifty. The VEC specification restricts the long-run behaviour of the endogenous variables to converge to their cointegrating relationships while allowing a wide range of shortrun dynamics. Each equation in the VECM model is interpreted as having two parts. The first represents adjustment to long-run equilibrium and measures how the dependent variable adjusts to the previous period’s deviation from long run equilibrium. The cointegration term is known as the error correction term (ECT) since the deviation from long-run equilibrium is corrected gradually through a series of partial short-run adjustments. The parameters of the ECT represent the speed of adjustment of index (futures) returns toward equilibrium. The larger the ECT of Nifty (Futures), Sample(adjusted): 4 1156 Included observations: 1153 after adjusting endpoints Standard errors & t-statistics in parentheses

Cointegrating Eq:

CointEq1

LNFTY(-1) LFTDX(-1)

1.000000 -1.000196 (6.6E-05) (-15222.1)

Error Correction:

D(LNFTY)

D(LFTDX)

CointEq1

0.114827*** (1.28) 0.222453** (1.54) -0.123505 (-0.91)

0.293110* (2.96) 0.401014* (2.61) -0.336542** (-2.32)

R-squared Adj. R-squared Sum sq. resids S.E. equation Log likelihood Akaike AIC Schwarz SC

0.006290 0.004561 0.228983 0.014111 3278.184 3278.189 3278.202

0.015126 0.013413 0.261275 0.015073 3202.128 3202.133 3202.146

Mean dependent S.D. dependent

0.000967 0.014143

0.000973 0.015175

D(LNFTY(-1)) D(LFTDX(-1))

Determinant Residual Covariance Log Likelihood Akaike Information Criteria Schwarz Criteria

1.66E-09 8382.907 8382.921 8382.956

* ,** and *** indicate significance at the 1%, 5% and 10% levels, respectively.

50

. . . continued on following page

I C R A B U L L E T I N

the greater is the Nifty’s (Futures’) response to a deviation from long term equilibrium. On the other hand, insignificantly small values of the ECT imply that the Nifty (Futures) is unresponsive to the previous period’s equilibrium error. (Enders, 1995) The remaining portions of the equations are the lagged first differences, which represent short-run effects of the previous period’s price changes on the current period’s price changes. The choice of optimal lags is given by consideration of minimising the Akaike information criterion (AIC) and Schwarz criterion (SC). The ECTs are significant for both futures and spot but quite low for the spot and accepted as non-zero only at 10 per cent level of significance. The lagged values of Futures is also not significant in the Nifty equation; the lagged Nifty values are significant in the Futures equation.

Money

&

Finance

F E B R U A R Y. 2 0 0 7

Nifty’s response to a one standard

liquidity requirements and transaction costs. Further, unlike index futures contracts the spot index itself is not a traded asset but simply a weighted average of the prices of its component stocks. Thus, changes in the index reflect the weighted sum of the changes in each of the prices of the 50 component stocks, which tend to move much more slowly than the index futures price in response to the arrival of new information. The final estimate of the speed of adjustment term in the spot return is described as the net outcome of the arbitrage and momentum trading effects. A positive speed of adjustment coefficient indicates that arbitrage trading is more prevalent than momentum trading in the spot market. Whenever buying or selling by other traders causes futures prices to move too high or too low relative to underlying stock prices index arbitragers buy the cheaper product, sell the other one, and lock in a gain, in effect transferring the news in futures prices to the stock market.10 As a next step we can make use of innovation accounting [Box 6] to obtain information concerning the interactions among the two variables in the price discovery process. The shape of the impulse response graphs [Panel A] shows the response of futures prices (Nifty) to an innovation or information shock to the Nifty (futures). Nifty’s response to a one standard deviation shock in the innovation of the Index futures price is slightly more intense than the futures response to Nifty innovations. In fact if the difference was sharper, the long term predictive power of the futures prices could have been definitely established. As this difference is too small we may say that the impulse responses indicate that neither the spot market nor the futures dominates the other in the price formation of the Nifty index. Coming to the

10 Momentum traders track the value assessments assigned by other investors and take position in the stock with the hope that its momentum will continue over a certain period and hence tend to keep stock prices sticky.

deviation shock in the innovation of the Index futures price is slightly more intense than the futures response to Nifty innovations.

51

I C R A B U L L E T I N

BOX 6

Money

&

Finance

The shape of the Impulse response functions and results from Variance decompositions together termed innovation or policy accounting, is useful in examining the relationship among economic variables.

F E B R U A R Y. 2 0 0 7

Panel A: Impulse Response A shock to the i-th variable directly affects the i-th variable, and is also transmitted to all of the endogenous variables through the dynamic structure of the VECM. An impulse response function traces the effect of a one standard deviation shock to one of the series on current and future values of the endogenous variables.* FIGURE 1 Nifty Response to Futures Shocks 0.0158 0.0156 0.0154 0.0152 0.0150 0.0148 0.0146 0.0144 0.0142 1

2

3

4

5

6

7

8

9

10

Cholesky ordering is Futures, Nifty

FIGURE 2 Futures Response to Nifty Shocks 0.0155 0.0150 0.0145 0.0140 0.0135 0.0130 1

2

3

4

5

6

7

8

9

10

Cholesky is Nifty, Futures

*The shocks are, however, usually correlated, so that they have a common component which cannot be associated with a specific variable. A method of dealing with this issue is to attribute all of the effect of any common component to the variable that comes first in the VECM system by using a Cholesky ordering (Enders, 1995).

52

. . . continued on following page

I C R A B U L L E T I N

Panel B: Variance Decomposition The forecast error variance decomposition gives us the proportion of the movements in a sequence due to its “own” shocks versus to the other variable. The fact that there exists contemporaneous correlation between innovations in our model means that no unique values may be found for the information shares. Yet, we can determine upper and lower bounds by making use of variance decomposition in combination with a Cholesky ordering. Cholesky Ordering

Nifty, Futdix Futdix, Nifty

Information share of Nifty Nifty

Futdix

98.3

1.67

9.99

90.01

Money

&

Finance

F E B R U A R Y. 2 0 0 7

Information share of Futures Nifty 98.4 9.54

Futdix 1.6 90.46

information-share of each market, the higher the information share, the more the market contributes in the price discovery process. In the Nifty’s price discovery process we see that the cross contributions of both the markets to each other’s price discovery is about the same. Nifty contributes information in the range of 1.6 to 9.5 per cent to the futures process. The futures market on the other hand contributes in the range of 1.7 to 10 per cent to the spot price movement process for the S&P CNX Nifty index [Box 6, Panel B]. Thus while the information shares do not establish the dominance of any one market in the price discovery process, it is clear that both markets incorporate information from each others’ prices.

IV. Concluding Observations We have tried to assess the short and long run predictive power of the futures prices for the underlying S&P CNX Nifty index traded on the National Stock Exchange of India using time series methods of causality tests and an error correction model. Alongside we also examine the long run interactions and information flows between the two sets of prices evolving in the two markets. The Nifty futures market, based on the highly traded 50 share Nifty index, is the most vibrant index futures market of India. Our sample period of over four years from March 2002 to September 2006 relates to a phase of the Indian financial market when a wide variety of market participants had begun to operate. Along with FIIs who are the more experienced traders, there was an increasing trend of retail participation through Mutual Funds or otherwise.11 This period also witnessed a high growth

11 Retail investors consistently accounted for 60 to 62 per cent of turnover in the last six months till September 2006, with their gross traded value ranging between Rs. 6,45,166 crore and Rs.9,15,031 crore.

Nifty contributes information in the range of 1.6 to 9.5% to the futures process. The futures market on the other hand contributes in the range of 1.7 to 10% to the spot price movement process for the S&P CNX Nifty index.

53

I C R A B U L L E T I N

Money

&

Finance

F E B R U A R Y. 2 0 0 7

Importantly, the futures market as expected is seen to respond faster in case of deviation from equilibrium and thus it helps to absorb much of the volatility induced by flow of new information.

of liquidity and investments in the financial market as the robust macro-economic scenario led Indian stock price indices to new highs. Thus at this juncture we considered it important to analyse whether the market outcome is close to what is expected in a well functioning one. Our results related to the price formation process of the index futures market and its underlying Nifty suggest that as expected there is a strong long run relation between the prices evolving in the two markets. Hence, if prices diverge from an efficient level both markets trade in such a way as to correct for imbalances. As for the ability of futures prices to predict subsequent price movements of the underlying asset, on testing for predictability independently for the short run we find that the futures prices/returns lead the spot and not vice versa, indicating significant predictive power of futures prices.12 Therefore, the futures prices may serve as the market’s expectation of a subsequent delivery period spot price. However, it is also necessary to take into account any existing long-run relation between the two sets of prices so as to ensure accurate predictions. If we take into account the long run relation between the two price series the dominance of the futures market cannot be very clearly established. The long run model instead shows both markets to be contributing to price discovery almost equally. Empirical investigations have suggested that if the same asset is traded in different markets, the market with the lowest trading cost usually dominates the price discovery process.13 This indicates that trading costs may not be significantly lower in the Indian index futures market and this does not allow it to gain a clear informational edge over the spot market. It may also reflect the fact that institutional players who are efficient arbitragers are not yet allowed to participate in such activities due to regulatory concerns, and this restricts the price discovery function of the derivatives market.14 However, importantly, the futures market as expected is seen to respond faster in case of deviation from equilibrium and thus it helps to absorb much of the volatility induced by flow of new information.

12

54

The short-run prediction hypothesis contends that lagged futures prices have significant predictive power for spot prices over finite forecasting horizons. The predictions, however, may suffer from reduced accuracy if a long-run relation does exist. 13 In markets with trading costs, price differences for the same asset may be as large as the costs of executing the arbitrage between markets. Moreover, if trading costs differ, trading activity will tend to be concentrated in the lowest-cost market. 14 RBI guidelines on investment by banks in capital market instruments as yet do not authorise banks to use equity derivatives arbitrage (vide RBI master circular DBOD no. DIR. BC. 07/13.03.00/2002-2003 dated 26 July 2002). This disables banks, possessing arbitrage trading skills and institutionalised risk management processes, from trading in the equity derivatives market. Similarly, pension funds and insurers are also non-contributors to this process.

Innovation or policy accounting, which traces the time path of reactions in the two markets to an initial information shock, is useful to market participants as well as regulators. The share of price discovery originating in the futures markets has important implications for hedgers and arbitragers who use these markets. Our results show that the futures market information-share in the price discovery of the underlying Nifty is marginally higher than what Nifty contributes to its futures price discovery. Internationally too, automated futures trading as followed by NSE, is shown to contribute to a much greater extent to information assimilation as compared with the traditional floor-trading mechanisms. The fact that we find bidirectional information flows or feedback between the markets indicates that any regulatory initiative on the futures market will have an immediate and desired impact on the spot market.15

I C R A B U L L E T I N

Money

&

Finance

F E B R U A R Y. 2 0 0 7

References Alphonse, Pascal (2000), “Efficient Price Discovery in Stock Index Cash and Futures Markets”, Annales d’Economie et de Statistique, No. 60. Bose, Suchismita (2006), “The Indian Derivatives Market Revisited”, Money & Finance, Vol. 2, Nos. 24-25. Enders, W. (1995), Applied Econometric Time Series, John Wiley & Sons. Frino, Alex and Andrew West (2003), “The Impact of Transaction Costs on Price Discovery: Evidence from Cross-listed Stock Index Futures Contracts”, Pacific-Basin Finance Journal, No. 11, pp. 139–151. Hasbrouck, Joel (1995), “One Security, Many Markets: Determining the Contributions to Price Discovery”, Journal of Finance 50, pp. 1175-99. Hasbrouck, Joel (2003), “Intraday Price Formation in US Equity Index Markets”, Journal of Finance, Vol. 58, Issue 6, pp. 2375-2400. Kenourgios, Dimitris F. (2004), “Price Discovery in the Athens Derivatives Exchange: Evidence for the Ftse/Ase-20 Futures Market”, Economic and Business Review, Vol. 6, No. 3, pp. 229-243. Pattarin, Francesco and Riccardo Ferretti (2004), “The Mib30 Index and Futures Relationship: Econometric Analysis and Implications for Hedging”, EFMA 2004 Basel Meetings Paper. Raju, M.T. and Kiran Karande (2003), “Price Discovery and Volatility on NSE Futures Market”, Working Paper Series No. 7, Securities and Exchange Board of India. So, Raymond W. (2004), “Price Discovery in the Hang Seng Index Markets: Index, Futures, and the Tracker Fund”, Journal of Futures Markets, Volume 24, Issue 9, pp. 887-907.

15 It is important to remind readers that our results are based on information on daily average of prices in the last half hour of trade in the respective markets. Hence, non-synchronous price information is a major drawback and averaging may have blurred the estimates obtained. Thus conclusive results on the price discovery function would require time stamped data for both markets. Such data would also allow one to meaningfully look into the question of volatility spillovers or the effect that volatility in the futures market has on the spot index. Though our study provides a broad direction for researchers and policy makers, estimates from data sampled at much higher frequencies would be extremely useful for both policy makers and market participants.

55

I C R A B U L L E T I N

Money

&

Finance

F E B R U A R Y. 2 0 0 7

56

Stoll, H.R. and R.E. Whaley (1990), “The Dynamics of Stock Index and Stock Index Futures Returns”, Journal of Financial and Quantitative Analysis, Vol. 25, pp. 441-468. Tse, Yiuman (1999), “Tse, Y. (1999), Price Discovery and Volatility Spillovers in the DJIA Index and Futures Markets”, Journal of Futures Markets, Vol. 19, pp. 911-930. Tse, Y., Paramita Bandyopadhyay and Yang-Pin Shen (2006), “Intraday Price Discovery in the DJIA Index Markets”, Journal of Business Finance & Accounting, Vol. 33, Issue 9-10, pp. 1572-1585. Turkington, Joshua and David Walsh (1999), “Price Discovery and Causality in the Australian Share Price Index Futures Market”, Australian Journal of Management, Vol. 24, No. 2. West, A. (1997), “Do Stock Index Futures Prices Lead the Stock Index?”, Working Paper No.97014f, Securities Industry Research Centre of Asia-Pacific (SIRCA). Zhong, Maosen Ali Darrat and Rafael Otero (2004), “Price Discovery and Volatility Spillovers in Index Futures Markets: Some Evidence from Mexico”, Journal of Banking & Finance, Vol. 28, Issue 12, pp. 3037-3054.