Forecasting volatility of the Indian stock market

Abstract

The present study on ‘forecasting volatility of the Indian stock market’ was conducted with the main objectives of examining the volatility behaviour of the Indian stock market, to forecast the sector- wise volatility of the Indian stock market and to identify the most efficient volatility forecasting model among the different models used.

For the study the biggest stock market in India in terms of total turnover and volume of transactions, ie, National Stock Exchange was selected. For analyzing the volatility behaviour of the Indian stock market as a whole, S&P CNX Nifty index was taken. Five companies representing five different sectors were selected for forecasting sector – wise volatility. The study used secondary data on daily close prices of individual stocks from November 1994 to October 2008, and for Nifty, daily close values, from November 1995 to October 2008 from the website of National Stock Exchange, www.nseindia.com.

The study revealed presence of strong volatility in the Indian stock market. The histogram drawn for the volatility of all samples showed that the distribution of volatility was not normal. There was positive skewness and all the distribution of volatility was leptokurtic. This proved the presence of high peak values (squared returns) in the sample data, exposing the evidence of volatility clustering and the possibility for prediction of future volatility.

While analysing sector -wise volatility, the diversified sector represented by Reliance Industries Limited showed the highest volatility compared to that of Nifty and the other sectors. In other words, Reliance is the most volatile stock among the samples selected for the study. Reliance and Infosys had good predictability of volatility in the stock market. The best identified model for forecasting the volatility of stock markets is the EWMA. Then comes AR (1) followed by MA (3), RWM and HMM. Random walk model was found suitable for the prediction of volatility of two sectors - IT (Infosys) and engineering heavy (BHEL) only. But the MAPE values of these were high. Historic mean model could not predict the volatility in the stock market with precision, for the index as well as for any of the five companies. Out of three, six, nine and twelve monthly moving averages taken for predicting the volatility three months moving average was found most suitable for all the samples.

Prediction of volatility using the most efficient model of EWMA identified indicated decreasing trend of volatility for the next six months, except for Infosys. The confidence limits for the Nifty and the stocks of five companies based on volatility for the sample period found that for Infosys the distribution of volatilities for the out of sample period are coming within the prefixed UCL and LCL and it ensures that the volatility is under control and predictable with high degree of precision.

The ever increasing market segments, advancement of technology, widening market reach and multi dimensions of stock market provide ample scope for further research in this area to the advantage of the investors and other market participants.