Derivative Trading and Structural Breaks in Volatility in India: An ICSS Approach

Researchers argue that ignoring the structural breaks in the time-series variance can cause significant upward biases in the degree of persistence in estimated GARCH models. Against this backdrop, the present study empirically examines the effect of stock futures on the underlying stock’s volatility in India by incorporating the structural breaks with the help of ICSS test and AR (1)-GARCH (1, 1) model for 30 most liquid and actively traded underlying stocks and their associated futures contracts. The study period ranges from the 1st January 2000 or the listing date of the particular stock (whichever is prior) till 31st March 2019. The study contributes to the on-going debate regarding the effect of derivatives on the underlying stock market’s volatility in two ways. Firstly, by taking into consideration the breaks in the volatility and, secondly, studying the effect of single stock futures will allow us to evaluate company-specific response to futures trading directly. The study offers a mixed outcome for the stocks under consideration. However, there is evidence of a decline in unconditional volatility for the majority of the stocks. The overall findings indicate that trading in stock futures may not have any detrimental effect on the underlying stock’s volatility.<br>

underlying market's volatility and, thus, its job in increasing or decreasing the underlying markets' volatility has remained an intense subject of empirical and analytical interest.
Questions about the effect of derivatives trading on underlying market volatility have been empirically addressed in two ways. Firstly, by analyzing variation in volatility over the pre-and postderivatives trading phases and, secondly, by measuring the effect of derivatives trading on the behavior of the underlying markets by comparing the performance with proxies. Moreover, most studies examining the effect of derivatives on the underlying market volatility used some type of GARCH model with dummy variable regressors 1 . However, this approach is based on the underlying presumption that any changes detected during the post-derivatives phase are caused by derivatives trading alone.
An increase in volatility could be the outcome of various other events, such as the initiation of a rolling settlement system, circuit breakers, and changes in regulations, and so on. If the structural breaks in variances of the examined time-series are ignored, the degree of persistence of the GARCH model estimate may be significantly biased. Several studies, such as Diebold (1986), Granger and Hyung (1999), Mikosch and Starica (2000), Diebold and Inoue (2001), have stated that neglecting the structural breaks can lead to spurious GARCH model estimation. The primary reasons for such structural breaks could be the changes in the mechanism of exchange rate systems, global financial markets crisis, or the evolution of the stock markets. The shocks produced by these significant economic or political events may cause a deviation in the financial time-series (Andreou & Ghysels, 2002; Wang & Moore, 2009).

Literature review
The derivatives market and its effect on the underlying market volatility are debated again and again with supporting and countering theories.

Increased volatility due to futures trading
Wats (2017) examined the effect of the derivatives contracts' expiration on the underlying market's volatility using the GARCH family models. He concluded that spot market volatility has increased during the expiry days and week after the listing of the derivatives. Other studies that find a significant increase in the Index return volatility following the listing of futures include Harris (1989), Brorsen (1991), Lee and Ohk (1992), Antoniou and Holmes (1995), Yao (2016).

Decreased volatility due to futures trading
Others argue that futures' listing potentially reduces the spot market's volatility, thus stabilizing the market. One of the clarifications for the destabilizing theory is that trades in the derivatives market destabilize the underlying market by providing an alternative route for the transmission and reflection of data in the cash market (Cox & Ross, 1976;Ross, 1989). Gulen and Mayhew (2000) studied the effect of index futures on the volatility of the international equity markets by taking the sample of 21 European nations by applying the BEKK model and GJR-GARCH. They found that the volatility of the underlying market has declined for most of the countries under study.

Mixed evidence/no impact of futures trading
Using the GARCH (1, 1) model, Rahman (2001)  Indian studies based on stock futures focus on conceptual clarity or cover only a short period. Research focusing on the index analysis does not consider the stock-specific characteristics, which could also play a significant role in the formation of the volatility. considering the breaks in the volatility. It aims at identifying the structural breaks, if any, in the stock prices by applying the ICSS test of Inclan and Tiao (1994). Secondly, studying the effect of single stock futures will allow us to directly evaluate company-specific responses to futures trading, in contrast to the market-wide effect gained from research with index futures.

Method
The Individual Stock Futures (ISF) has proved to be a hugely successful financial instrument on Indian bourses, and NSE has continued to account for the majority of total volumes traded in the ISF segment all over the world. The resulting sample for this study comprises 30 most liquid and actively traded underlying stocks on which futures contracts are available. These 30 stock futures contribute to around 70-80% of the total trading volume of the F&O segment of NSE, excluding the index futures.
The majority of them are also part of the S&P Nifty Index, the Benchmark Index of NSE. The data extracted for 30 stocks have been procured from the Bloomberg database. The study period will range from the 1st January 2000 or the listing date of the particular stock (whichever is prior) till 31st March 2019.
If no sudden changes occur during the entire sampling duration in the variance of the sequence, k D oscillates about zero. If there are one or more sudden shifts in variance, then the k D statistics will drift either above or below the zero. The ICSS algorithm helps in identifying breaks in variance of the time-series at different points in time.

Linking the structural breaks in volatility with trading in stock futures
First, the dates for the structural breaks in the stocks will be estimated. Later, these structural breaks were matched with the dates of the listing of stock futures on the individual stocks. If a structural break is found within six months of the listing of stock futures, it has been attributed as likely to derivative trading.
AR (1)-GARCH (1, 1) is a GARCH family model, in which the mean is determined by a firstorder auto-regressive AR (1), with a GARCH (1, 1) error: Once all the structural breakpoints are identified, dummy variables are created for each detected break. Each dummy variable is denoted with value one onwards from the identified location until the end of the data series and 0 elsewhere.      Total persistence has risen for ten stocks while declined for fourteen stocks. On the other hand, α has increased for eleven stocks, while it has decreased for thirteen stocks.

Discussion
Through this study, an attempt has been made to model the underlying stock's volatility with stock futures by taking into consideration the breaks in the volatility. Several studies, such as Diebold

Conclusion
Through this analysis, any consistent patterns were not found in terms of changes in total persistence, unconditional volatility, and α for the underlying stocks for the period after the relevant breaks. The mixed outcome could be due to stock-specific characteristics, which could also play a