Is Volatility Uniform Across the Stock Market? Evidences from Select Indices of NSE 23

Transcription

1 Sankersan Sarkar Prashant Verma Abstract Extreme volatility in stock markets is a matter of concern for both regulators and investors because it can cause widespread losses. This paper attempts to find out whether the phenomenon of volatility is the same across the entire stock market or differs across the underlying sectors. It studies volatility in select indices of NSE which include one broad market index and five sectoral indices over a period of fifteen quarters. It is found that volatility significantly differs across sectors within the stock market as well as between sectors and the market as a whole represented by the broad market index. Keywords: Volatility, stock market, NSE, sectoral index, Nifty 23

2 1. INTRODUCTION Volatility is bad; and it is good. It is bad because it has the potential for massive and widespread losses. It is good because it also results in gains and sometimes massive gains. It is only due to the phenomenon of volatility and the consequent gains and losses that it entails that investors are motivated to trade in securities on a regular basis, thus keeping the markets alive and kicking. It is like a necessary evil in the financial markets because in the absence of volatility there would be little scope for gains (and losses) from the day-to-day transactions. Volatility actually turns out to be bad when it suddenly increases, rather violently, causing massive losses across the market. Equity investors tend to become risk averse in the presence of volatility and compensate themselves by including a risk premium in their expected returns. Thus the higher the volatility in the market the higher would be the expected returns of equity investors resulting in a higher cost of equity capital. This, in turn, would make it difficult for companies to raise equity from the markets at lower costs thus affecting their plans for investment. Often during times of extreme volatility, regulators intervene in the markets in order to bring it within control. Interventions happen when volatility in the broad market indices exceeds certain pre-defined limits and trading is halted across all equity markets in India. While such intervention is necessary in order to curb acute volatility in the equity markets, it also affects the process of price discovery, which is one of the important roles of financial markets. It is thus important for the regulators to understand the nature of volatility in order to control it effectively. Since there are equity stocks belonging to various industrial sectors, we propose in this paper that a market-wide halt of trading should be done only when volatility exceeds certain limits in all the sectors of the market. If volatility persists in some of the sectors only, then an effective intervention measure would be to halt trading in those sectors only, while allowing trading to continue in the remaining sectors. We make an attempt in this paper to find out whether the phenomenon of volatility observed in the broad market index is the same or different, at least with reference to some of the sectoral indices. A difference in volatility even in one sectoral index from the rest of the market might indicate that regulatory intervention on stock market trading based only on a broad market index may not be appropriate. This, in turn, might pave the way for more intensive studies in this context. While it might be easy to form a view that different sectors of business are likely to exhibit different volatility in the stock market, because they have different risk characteristics, the evidences found in this study do not seem to support this view. 2. LITERATURE REVIEW Some of the studies have attempted to find out the implications of FII investments for stock market volatility in India. Samal (1997) has explained that the main feature of the Indian equity markets which is emerging gradually is its integration with the world markets and the resulting problems are arising out of the inflows and outflows of capital by FIIs. Pal (1998) has attempted to verify the belief that portfolio investments by foreign investors can improve the stock market and economy, and found that evidences in India do not support the same; his study in the context of the Indian economy does not support that portfolio investments lead to economic development. Kaur and Dhillon (2015) studied the contribution of investments by foreign institutional investors (FIIs) to the volatility 24

3 of stock market returns in India; their research confirms that FII investments significantly contribute to volatility of stock returns. Further, the impact of gross purchases by FIIs on volatility is more than that of gross sales. There have been studies on volatility in the Indian and foreign stock markets around the period of the global financial meltdown. Goudarzi and Ramanarayanan (2011) have studied the impact of good and bad news on volatility in the Indian stock markets during the global financial meltdown of They found that bad news has greater impact on stock market volatility in India than good news. In other words, bad news increases stock market volatility more than good news. Seďa (2012) studied the potential risk in Central and European stock markets during the period of the global financial crisis and analysed its characteristics. The study found evidence of increase in volatility in these markets before, during and after the crisis. Sakthivel et al (2014) studied the effect of the global financial crisis on the major Indian stock markets and found that there was an increase in volatility of stock returns in the post-crisis period over that in the precrisis period. Olowe (2009) studied the relationship between stock returns and volatility in Nigeria under the mixed conditions of financial sector reforms and stock market crash in Nigeria, and the global financial crisis. Little evidence was found on the relationship between stock returns and volatility. However, it was found that the stock market crash of 2008 had contributed to the persistent high volatility in the Nigerian stock market. There are some studies which have attempted to examine the spillover phenomenon across markets for different financial assets or across different economies. Saha and Chakrabarty (2011) studied the volatility spillover effect around the financial crisis of on the foreign exchange markets and stock markets in India, USA, UK and Japan. They found evidence of volatility spillover but no asymmetric impact between stock to exchange rates and vice versa. Further, their study shows the changing nature of volatility contagion between financial markets. Mukherjeea and Mishra (2010) studied stock market integration and volatility spillover between India and its major neighbours. They found significantly positive bidirectional spillover of intraday returns between India and the major Asian countries. Further, there is significant information flow from four Asian markets viz. Hong Kong, Korea, Singapore and Thailand. Some of the studies have attempted to examine the concept of implied volatility in the stock markets in various economies. Dania and Malhotra (2014) have examined the predictive power of implied volatility in the US stock markets and the returns in BRIC markets. They found that the returns in the BRIC markets were significantly negatively influenced by USVIX. Also they found evidence of volatility spillover phenomenon from USVIX to BRIC market returns. De and Chakrabarty (2016) have attempted to explain the association between FII flows and the implied volatility index (VIX). They have found that VIX has no influence on the FII flows to equity markets in India. However, they found significant evidence of reverse causality. Chandra and Thenmozhi (2015) have studied the asymmetric relationship between India Volatility Index (India VIX) and the stock market returns. They found that stock market returns are in general negatively related with the changes in India VIX. Also the India VIX provides a better measure of volatility than the traditional measures including the ARCH/ GARCH family-based approaches. 25

4 There have been some studies which have examined the concept of conditional volatility. Glosten et al (1993) have found a negative relationship between expected monthly returns and conditional variance of monthly returns. They have also shown that monthly conditional volatility is not likely to be persistent as understood generally. Bley and Saad (2015) have studied the forecasting accuracy of two types of models historical volatility based models and conditional volatility based models - on estimates of idiosyncratic risk of individual Saudi Arabian stocks. They found that in general, the historical volatility based models are better. Srinivasan (2015) has attempted to forecast conditional volatility of the S&P CNX Nifty index returns using various advanced models belonging to the GARCH family. He found that the asymmetric GARCH models provide better forecasts of conditional volatility than the symmetric GARCH models. Demir, Fung and Lu (2015) have tested the validity of the factors: conditional consumption and market return volatilities, in explaining stock return differences in the context of India, an emerging market. They found that conditional volatility has limited effect on stock returns of large capitalisation companies, whereas it has a significant effect on stock returns of small value and small growth companies. Some of the studies have examined the impact of introduction of index-based derivatives on stock market volatility in India. Shenbagaraman (2003) has attempted to assess the impact of introducing index futures and options on volatility of the stock market index in India. It was found that the introduction of index derivatives contracts did not cause any change in the volatility of the stock market. Further, the introduction of index derivatives had improved liquidity and reduced informational asymmetries. Nandy (Pal) and Chattopadhyay (2014) have studied the impact of introduction of derivatives trading on stock market volatility in India. They found that introduction of some types of derivative instruments (index futures, index options and interest rate futures) have reduced stock market volatility whereas introduction of currency futures had a destabilising impact on stock market volatility. There are some studies which have examined the relationship between volatility in various asset markets. Mishra, Das, and Mishra (2010) have tested the causality between domestic gold prices and stock market returns in India. They found that there exists Granger causality between gold prices and stock market returns i.e. both variables are found to Granger cause each other. Their finding indicates a linkage between gold price volatility and stock market volatility in India. Agrawal, Srivastav and Srivastava (2010) have analysed the relationship between stock market volatility and Rupee-Dollar exchange rate movements. They found a unidirectional causal relationship running from stock market returns to exchange rates, which indicates that stock market volatility caused volatility in Rupee-Dollar exchange rates during the study period. Some of the studies on stock market volatility do not fall within the commonly observed categories of studies. Roy (2001) studied the implications of financial liberalisation on the growth and efficiency of the Indian stock markets and found that it has resulted in volatility in the Indian stock markets. Raju and Ghosh (2004) have found that volatility in the Indian market is higher than in the other markets. Das, Kulkarni, and Kamaiah (2014) have examined volatility patterns in the Indian stock markets. They did not find evidence of volatility persistence and leverage effect in the Indian stock markets during the period of their 26

5 study. Gaye Gencer and Demiralay (2016) have attempted to estimate one-day and one-week-ahead VaR in eight emerging stock markets. They found that the skewed Student-t distribution resulted in the best forecasts of one-day-ahead VaR in all the stock markets. They also found that the predictive power of their models decreased over longer forecasting horizons. Wachter (2013) has attempted to find the reasons why equity premiums are very high and why stocks are very volatile. The author has attempted to explain the same on the basis of a time-varying probability of a consumption disaster. It has been explained that the time-variation in the probability results in high stock market volatility and excess return predictability. Sarkar, Chakrabarti, and Sen (2009) have studied the volatility transmission channel from the sectoral indices in India and developments in the global markets into the Indian stock markets. The sectoral indices in India include capital goods, consumer durables, FMCG, healthcare and IT; the sectoral indices and the stock market index are based on the Bombay Stock Exchange. They found that volatility in the developed market indices around the world Granger causes volatility in the Indian stock markets, indicating evidence of global contagion. Further, among the domestic sectors, capital goods and consumer durables are the most significant contributors to volatility of the stock market as a whole. It is understood from the survey of literature that most of the studies can be classified along the following broad themes: Studies on the implications of FII investments on stock market volatility in India. Studies on stock market volatility around the period of the global financial crisis. Studies on volatility spillover effects across several countries. Studies on the concepts of implied volatility and conditional volatility. Studies on the impact of introduction of index-based derivatives on stock market volatility in India. Studies on the relationship between volatility in various asset markets. Studies based on miscellaneous and other themes. There appears to be a gap in the extant literature on volatility in the Indian stock markets, at least, as it is silent on whether the phenomenon of volatility is uniform across the market or otherwise. It also appears from prior studies reviewed in this paper that the research pertaining to stock market volatility has been mainly focused on the broad market indices and has not taken into consideration the volatility in the sectoral indices. We found only one study which has taken the sectoral indices into consideration (Sarkar, Chakrabarti and Sen, 2009). The present study differs from majority of the prior studies on stock market volatility in at least two ways. First, it is based on volatility of the index values of the broad market index and the sectoral indices, and not on the volatility of the stock market returns, as in the prior studies. Second, it attempts to compare volatility in the index values for the broad market and those of the various sectors within the market. So far in the survey of literature on this subject, we came across only one study in the Indian context which has attempted to examine volatility in the sectoral indices of the stock market (Sarkar, Chakrabarti and Sen, 2009). However, this study examines the influence of developments in the sectoral indices on the volatility of the stock market index, instead of comparing volatility across the broad market and sectoral indices. 27

6 3. OBJECTIVES The objectives of this study are: To find out whether volatility as a phenomenon differs between the broad market index and the sectoral indices. To identify the source(s) of difference among the sectoral indices, if volatility is found to differ between the broad market index and the sectoral indices. For each trading day, the stock exchange as well as the PROWESS database report only four values of each index - opening, high, low and closing. So the estimation of daily volatility of each index covered in this study is based on four reported values only. We could not use the intra-day values of the indices for the entire study period because historical data on the intra-day index values is not freely available in the public domain. 4. DATA AND METHODOLOGY 4.1 Data The study requires daily index values for a broad market index and sectoral indices. Owing to the growing significance of National Stock Exchange (NSE), it was decided to select the indices for this study from the NSE. Hence, we have chosen one broad market index and five sectoral indices, all of them from the NSE. The broad market index is CNX Nifty 50 and the five sectoral indices are: Nifty Auto Index, Nifty FMCG Index, Nifty Infrastructure Index, Nifty Metal Index and Nifty Pharma Index. The rationale for choosing the aforementioned sectoral indices is that all of them belong to the nonservices sector i.e. all of them represent sectors that involve some type of industrial or manufacturing activity. The period of study covers fifteen quarters starting from April 1, 2012 to December 31, Data prior to April 1, 2012 was not consistently available for all the five sectoral indices chosen. Hence, the period of this study could not be made any longer. The data for the above indices have been collected from the CMIE PROWESS database. The study thus uses secondary data. Prowess is a database for financial and market data on Indian companies and it contains data in the form of time series from the year and is updated regularly. 4.2 Methodology This section explains the analytical steps in terms of the following: Dividing the study period into sub-periods and reclassifying the index data Estimation of volatility of index values, and Analytical techniques used Dividing the Study Period and Reclassifying the Index Data For each of the six indices, the data during any financial year was divided into 4 quarters viz, April 1 to June 30, July 1 to September 30, October 1 to December 31, and January 1 to March 31. So for the financial years to , there are 12 quarters and for the rd current financial year, , the data up to the 3 quarter (ending December 31, 2015) have been considered. In all, data collected for the six indices were segregated into 15 quarters. The rationale for segregating the index values into quarters is that after the end of every quarter, companies release unaudited quarterly financial results. The flow of financial information pertaining to the previous quarter might affect the volatility in the current quarter. Hence, segregating the index data according to quarters would keep values for the same index characterised by different extent of volatility 28

7 over different time periods, separate. Thus index values for the same index over different quarterly periods would be characterised by different extent of volatility and would be analysed separately Estimation of Volatility of Index Values The purpose of the present study is to find out whether the phenomenon of volatility in a stock market is uniform across all its sectors, or not. For the purposes of this study, volatility was computed on a daily basis. If, instead, volatility is calculated over periods longer than a trading day, then such a measure of volatility would be affected by unusual or abnormal volatility occurring on one or more trading days during such period of computation. Calculating volatility on a daily basis will enable us to keep days with unusually higher or lower volatility separate from the other days. Hence, the measure of volatility on any trading day would be independent of the influence of volatility occurring on other trading days. Further, the measure of volatility for this study is daily relative volatility of index values. We define daily relative volatility as coefficient of variation of daily stock index values. By computing coefficient of variation of stock index values on any trading day, we can eliminate the influence of shifts in mean and standard deviation in index values from one day to another, and focus our attention specifically on volatility as a phenomenon rather than as an absolute measure only. Further, computing coefficient of variation on a daily basis will enable us to avoid the influence of long term trends in stock prices (up / down) on the estimate of volatility. So, by using a relative measure of volatility, we can focus our attention on volatility as a phenomenon independent of the influence of any secular trend in asset prices. This apart, the most important reason for using the relative measure of volatility is to make it comparable across the sectoral indices the values of the various sectoral indices have different orders of magnitude. More specifically, the following ranges of values are observed for the indices during the study period: CNX Nifty 50: The opening index values of CNX Nifty 50 varied from on April 2, 2012 (starting date of the study period) to on December 31, 2015 (ending date of the study period), with the highest value of on March 4, 2015 and lowest value of on May 18, Nifty Auto Index: The opening index values of Nifty Auto Index varied from on the starting date to on the ending date of the study period, with the highest value of on January 3, 2015 and the lowest value of on June 4, Nifty FMCG Index: The opening index values of Nifty FMCG Index varied from on the starting date of the study period to on the ending date, with the highest value of on February 25, 2015 and the lowest value of on April 10, Nifty Infrastructure Index: The opening index values of Nifty Infrastructure Index varied from on the starting date to on the ending date of the study period, with the highest value of on June 10, 2014 and the lowest value of on August 28, Nifty Metal Index: The opening index values of Nifty Metal Index varied from on the starting date of the study period to on the ending date, with the highest value of on April 4, 29

8 2012 and the lowest value of on September 29, Nifty Pharma Index: The opening index values of Nifty Pharma Index varied from on the starting date to on the ending date of the study period, with the highest value of on April 9, 2015 and the lowest value on on June 15, It is obvious from the above discussion that the values of the six indices studied in this paper have different orders of magnitude and tend to vary between different ranges of values. Hence, the most appropriate way to compare their volatility would be to use relative measures of volatility as explained above. However, our measure of daily relative volatility will suffer from a data limitation problem. This is because on any trading day, the stock exchange reports only four values of any index viz, opening, high, low and closing. Hence, the calculations of mean, standard deviation and coefficient of variation for any day will only be rough estimates of the true variability on that day. This will be the case for all trading days Analytical Techniques Majority of the studies that we have come across have used some technique which belongs to the GARCH family. Glosten et al (1993) have employed a modified GARCH-M model to draw their conclusions. Olowe (2009) applied the E-GARCH-in-mean model to draw out the findings. Saha and Chakrabarty (2011) used a multivariate GARCH model to conduct their study. Seďa (2012) applied Jump-Diffusion GARCH model in carrying out his study. Sakthivel et al (2014) have employed GJR GARCH model in order to draw their conclusions. Goudarzi and Ramanarayanan (2011) have used EGARCH and TGARCH based models to find out the impact of good and bad news on volatility. Kaur and Dhillon (2015) have used GARCH family based approaches in their study of the returns on BSE Sensex during the period 1999 to The other analytical technique that has been used by some of the studies is the technique of Granger causality. Tests of Granger causality have been used by Sarkar, Chakrabarti, and Sen (2009); Mishra, Das, and Mishra (2010). However, since the objectives of this study are different from the studies covered in the review of literature, the analytical techniques that were applied are different from the prior studies. The first step in the analysis was to calculate daily relative volatility by computing the coefficient of variation of the index values on each trading day during the quarter. Since the sampling distribution of coefficient of variation is not known, it was decided to follow the nonparametric procedures of testing which do not make any assumption about the underlying statistical distribution. In order to find out whether volatility as a phenomenon differs between the broad market index and the sectoral indices, the Kruskal-Wallis ANOVA by ranks test was used. The ANOVA is a parametric approach that requires the population to be normally distributed, while Kruskal-Wallis ANOVA is a nonparametric approach and does not need the assumption of normal distribution. In the second stage of the analysis, post-hoc analysis was conducted by doing pair-wise comparisons between the indices to identify the indices where differences existed. The sample size varies for each sample but is sufficiently large, none being less than 300 data points. The above mentioned statistical tests were conducted for each quarter on the values of daily relative volatility across 30

9 the six indices. Further, all the tests were conducted at a significance level of 5%. The tests were carried out by SPSS software. 5. RESULTS AND DISCUSSION Tables 1 to 15 show the results of Kruskal-Wallis ANOVA by ranks for independent samples (KW ANOVA) generated by SPSS software for the financial years (FY) to For FY to FY , the tests are carried out for four quarters, viz. Quarter 1 to Quarter 4. However, for FY , the tests have been carried out for Quarter 1 to Quarter 3. This is because Quarter 4 of FY is currently under progress; hence, data for the complete Quarter 4 for FY are not available that will be available only upon its completion. Each table shows the results for a specific quarter of a particular FY. Further, the result table for any quarter is divided into two parts, A and B. Part A shows the value of the test statistic and Part B shows the p-value and the inference from the test i.e. the decision to accept or reject the null hypothesis in the test. The null hypothesis for each of these tests is: The distribution of CV is the same across categories of index. CV represents the daily coefficient of variation of the index values for the indices covered in this study. The notations used to represent the categories of index are described below: 1. denotes: CNX Nifty 50 index 2. denotes: Nifty Auto Index 3. denotes: Nifty FMCG Index 4. denotes: Nifty Infrastructure Index 5. denotes: Nifty Metal Index 6. denotes: Nifty Pharma Index Though these notations for the indices do not appear in the Tables 1 to 15, they do appear in the subsequent tables. All tables from 1 to 15 clearly show that the p-value for the test statistic is close to zero. Hence, the differences across the six groups are statistically significant. So the null hypothesis that the distribution of CV is the same across categories of index is clearly rejected in all these 15 tests of KW ANOVA for 15 quarters. This implies that the distribution of relative volatility (CV) in all the quarters differs across the indices under study. The indices that were subject to this study include one broad market index and five sectoral indices. This means that there is a significant difference in volatility among the sectors and the market-wide indices. We extend the analysis further to find out the indices among the entire group of six indices where significant differences exist. For this purpose, post-hoc analysis based on pair-wise comparison among the indices was carried out using SPSS software. The results of these tests are shown in tables 16 to 30. Each table shows the results of pair-wise comparison among the indices for a specific quarter in a FY. The right-most column of each table shows the adjusted p-value for the pairwise comparisons carried out. Based on the results in tables 16 to 30, the following observations are made with reference to index 1 (CNX Nifty), the broad market index, and the sectoral indices 2 to 6: Table 16 shows that in quarter 1 of FY , between the broad market index and indices 2, 4 and 5 respectively (Nifty Auto Index, Nifty Infrastructure Index and Nifty Metal Index Table 17 shows that in quarter 2 of FY , significant differences in volatility with reference to the broad market index are seen among indices 2, 3, 4 and 5 (Nifty Auto Index, Nifty FMCG Index, 31

10 Nifty Infrastructure Index and Nifty Metal Index Table 18 shows that in quarter 3 of FY , between the broad market index and the indices 2, 3, 4 and 5 (Nifty Auto Index, Nifty FMCG Index, Nifty Infrastructure Index and Nifty Metal Index Table 19 shows that in quarter 4 of FY , significant differences in volatility exist between the broad market index and the indices 2, 3, 4 and 5 (Nifty Auto Index, Nifty FMCG Index, Nifty Infrastructure Index and Nifty Metal Index Table 20 shows that in quarter 1 of FY , there exists significant differences in volatility between the broad market index and the indices 2, 4 and 5 (Nifty Auto Index, Nifty Infrastructure Index and Nifty Metal Index Table 21 shows that in quarter 2 of FY , significant differences in volatility with reference to the broad market index exist in the indices 2, 4 and 5 (Nifty Auto Index, Nifty Infrastructure Index and Nifty Metal Index Table 22 shows that in quarter 3 of FY , significant differences in volatility exist between the broad market index and the indices 2, 4 and 5 (Nifty Auto Index, Nifty Infrastructure Index and Nifty Metal Index Table 23 shows that in quarter 4 of FY , between the broad market index and the indices 4, 5 and 6 (Nifty Infrastructure Index, Nifty Metal Index and Nifty Pharma Index Table 24 shows that in quarter 1 of FY , significant differences in volatility with reference to the broad market index exist in the indices 4 and 5 (Nifty Infrastructure and Nifty Metal Table 25 shows that in quarter 2 of FY , significant differences in volatility exist between the broad market index and the indices 2, 4, 5 and 6 (Nifty Auto Index, Nifty Infrastructure Index, Nifty Metal Index and Nifty Pharma Index Table 26 shows that in quarter 3 of FY , between the broad market index and the indices 2, 3, 4 and 6 (Nifty Auto Index, Nifty FMCG Index, Nifty Infrastructure Index and Nifty Pharma Index Table 27 shows that in quarter 4 of FY , significant differences in volatility with reference to the broad market index exist in the indices 3, 4 and 5 (Nifty FMCG Index, Nifty Infrastructure Index and Nifty Metal Table 28 shows that in quarter 1 of FY , significant differences in volatility exist between the broad market index and the indices 5 and 6 (Nifty Metal Index and Nifty Pharma Index Table 29 shows that in quarter 2 of FY , between the broad market index and the index 6 (Nifty Pharma Index). Table 30 shows that in quarter 3 of FY , significant differences in volatility with reference to the broad market index exist in the indices 5 and 6 (Nifty Metal Index and Nifty Pharma Index The following additional observations can be made from tables 16 to 30 with reference to the sectoral indices: Table 16 shows that in quarter 1 of FY , 32

11 between the sectoral index 6 (Nifty Pharma Index) and sectoral indices 2, 4 and 5 (Nifty Auto Index, Nifty Infrastructure Index and Nifty Metal Index Table 17 shows that in quarter 2 of FY , between the sectoral index 6 (Nifty Pharma Index) and sectoral indices 2, 4 and 5 (Nifty Auto Index, Nifty Infrastructure Index and Nifty Metal Index Further, there is a significant difference in volatility between sectoral indices 3 and 5 (Nifty FMCG index and Nifty Metal Index Table 18 shows that in quarter 3 of FY , between the sectoral indices 6 and 5 (Nifty Pharma Index and Nifty Metal Index Table 19 shows that in quarter 4 of FY , between the sectoral index 6 (Nifty Pharma Index) and sectoral indices 2, 4 and 5 (Nifty Auto Index, Nifty Infrastructure Index and Nifty Metal Index Further, there are significant differences in volatility between sectoral index 3 (Nifty FMCG Index) and sectoral indices 4 and 5 (Nifty Infrastructure Index and Nifty Metal Index Also, there is a significant difference in volatility between sectoral indices 2 and 5 (Nifty Auto Index and Nifty Metal Index Table 20 shows that in quarter 1 of FY , between sectoral index 6 (Nifty Pharma Index) and sectoral indices 3, 4 and 5 (Nifty FMCG Index, Nifty Infrastructure Index and Nifty Metal Index Further, there is a significant difference in volatility between sectoral indices 3 and 5 (Nifty FMCG Index and Nifty Metal Index Table 21 shows that in quarter 2 of FY , between sectoral index 6 (Nifty Pharma Index) and sectoral indices 3, 4 and 5 (Nifty FMCG Index, Nifty Infrastructure Index and Nifty Metal Index Further, there is a significant difference in volatility between sectoral indices 2 and 5 (Nifty Auto Index and Nifty Metal Index Table 22 shows that in quarter 3 of FY , between sectoral index 6 (Nifty Pharma Index) and sectoral indices 2, 4 and 5 (Nifty Auto Index, Nifty Infrastructure Index and Nifty Metal Index Further, there is a significant difference in volatility between sectoral indices 3 and 5 (Nifty FMCG Index and Nifty Metal Index Table 23 shows that in quarter 4 of FY , between sectoral index 3 (Nifty FMCG Index) and sectoral indices 4, 5 and 6 (Nifty Infrastructure Index, Nifty Metal Index and Nifty Pharma Index Table 24 shows that in quarter 1 of FY , between sectoral index 3 (Nifty FMCG Index) and sectoral indices 4 and 5 (Nifty Infrastructure Index and Nifty Metal Index Further, there are significant differences in volatility between sectoral index 6 (Nifty Pharma Index) and sectoral indices 4 and 5 (Nifty Infrastructure Index and Nifty Metal Index Also, there is a significant difference in volatility between sectoral index 2 (Nifty Auto Index) and sectoral index 5 (Nifty Metal Index). Table 25 shows that in quarter 2 of FY , 33

12 between sectoral index 3 (Nifty FMCG Index) and sectoral indices 4 and 5 (Nifty Infrastructure Index and Nifty Metal Index Also, a significant difference in volatility is found between sectoral indices 2 and 5 (Nifty Auto Index and Nifty Metal Index respectively), and between sectoral indices 6 and 5 (Nifty Pharma Index and Nifty Metal Index Table 26 shows that in quarter 3 of FY , between sectoral indices 3 and 4 (Nifty FMCG Index and Nifty Infrastructure Index respectively), and between sectoral indices 6 and 4 (Nifty Pharma Index and Nifty Infrastructure Index Further, there are significant differences in volatility between sectoral indices 2 and 5 (Nifty Auto Index and Nifty Metal Index) and sectoral indices 4 and 5 (Nifty Infrastructure Index and Nifty Metal Index Table 27 shows that in quarter 4 of FY , between sectoral indices 2 and 5 (Nifty Auto Index and Nifty Metal Index respectively) and sectoral indices 6 and 5 (Nifty Pharma Index and Nifty Metal Index Table 28 shows that in quarter 1 of FY , there are no significant differences in volatility among the sectoral indices, though significant differences in volatility have been found between the broad market index and two sectoral indices (discussed earlier). This is the only exceptional quarter, in this sense. Table 29 shows that in quarter 2 of FY , between sectoral indices 2 and 6 (Nifty Auto Index and Nifty Pharma Index respectively) and sectoral indices 3 and 6 (Nifty FMCG Index and Nifty Pharma Index Table 30 shows that in quarter 3 of FY , between sectoral indices 2 and 5 (Nifty Auto Index and Nifty Metal Index respectively), sectoral indices 4 and 5 (Nifty Infrastructure Index and Nifty Metal Index respectively), sectoral indices 3 and 5 (Nifty FMCG Index and Nifty Metal Index respectively) and sectoral indices 6 and 5 (Nifty Pharma Index and Nifty Metal Index Based on the above observations, the following can be said: There are significant differences in volatility between the broad market index and one or more sectoral indices in every quarter. Moreover, volatility also differs among the sectoral indices. Almost every sectoral index that is covered in this study has been found to differ significantly in volatility from the broad market index in one or more quarters. But all sectoral indices have not differed significantly in volatility from the broad market index at the same time. The number of sectoral indices which were found to differ significantly from the broad market index differs from one quarter to another. The same pairs of sectoral indices are not found to differ significantly in volatility in all the quarters. The number of pairs of sectoral indices that differ significantly between each other differs across the quarters. In summary, it can be said that volatility significantly differs across some parts of the stock market while it remains the same for the other parts. Those parts of the market in which volatility differs and those parts in which it does not, change from one quarter to another. 34

13 6. LIMITATIONS This study suffers from three major limitations. First, the findings are limited by the amount of data available across all years. The data available across the indices covered in this study span from FY quarter 1 to FY quarter 3 (15 quarters). Data were not available for all the indices before FY ; hence, the study could not cover a period earlier than Second, the findings are limited by the amount of data available on any single trading day. The CMIE PROWESS database from where the data were collected reports only 4 index values for any trading day. The index values during the intra-day period on any trading day are neither reported by the stock exchange nor captured by the aforementioned secondary database. Third, the study covers only 6 indices of NSE it does not cover all the indices of NSE for the broad market or the other sectoral indices. 7. CONCLUSIONS The study infers that the phenomenon of volatility is not uniform across the entire stock market. Further, the differences in volatility occur across different sectors of the stock market over different periods of time, while it remains the same across the remaining sectors. This finding does not support the reasoning that stocks belonging to different sectors of business should exhibit different volatility in the stock market b e c a u s e s u c h s e c t o rs h ave d i f fe r e n t r i s k characteristics. If volatility in the stock market were to be influenced or determined by the business risks of specific sectors, then the same sector(s) would have exhibited different volatility from the market as a whole (the broad market index) as well as from the other sectoral indices, in every quarter. The finding that the sectors which show different volatility and those which show same volatility with reference to the rest of the market are found to change from one quarter to another, might indicate that volatility in different sectors of the stock market do not have a clear association with the business risks of such sectors. There is enough scope to make an attempt to generalise the findings of this study, because this study is not based on all the sectors of the stock market. This apart, depending on the amount of data available for each trading day, the findings can be improved further. The present study only serves as a precursor to more intensive and extensive studies. The important implications of the findings of this study are as follows: 1. Attempts to model volatility over time based on a broad market index over some time period might explain volatility of markets to some extent but such market based models may not be used as a basis to explain volatility at individual stock level. This can have implications for investment decisions or strategies which might be based on the volatility in market indices. 2. Any regulatory intervention on stock market trading based only on the movements of a broad market index, that attempts to curb increasing volatility might also affect trading in some of the underlying sectors which might not have experienced the same extent of volatility as the rest of the market. Such intervention on trading across the entire market could affect the price discovery process in those sectors which might not have experienced the same extent of volatility. Thus, regulatory intervention on trading should be initiated only after considering volatility in the sectoral indices. Situations in which it is found that extreme volatility persists only in some sectors, interventions on trading should be initiated for such sectors only. 35

14 This study contributes to the body of knowledge by evidences of differing volatility across different sectors of the stock market i.e. lack of uniformity of the phenomenon of volatility across the stock market. This might require attention of the regulators and investors alike, because volatility is a matter of concern to both. References Agrawal, G., Srivastav, A. K., & Srivastava, A. (2010). A Study of Exchange Rates Movement and Stock Market Volatility. International Journal of Business and Management, 5 (12), Bley, J., & Saad, M. (2015). Idiosyncratic Volatility Forecasting in the Stock Market of Saudi Arabia. Emerging Markets Finance & Trade, 51 (6), Chandra, A., & Thenmozhi, M. (2015). On Asymmetric Relationship of India Volatility Index (India VIX) with Stock Market Return and Risk Management. Decision, 42 (1), Dania, A., & Malhotra, D. (2014). Transmission of U.S. Stock Market Implied Volatility to Equity Markets of Emerging Countries. Journal of Wealth Management, 17 (2), Das, S., Kulkarni, A., & Kamaiah, B. (2014). Estimating Volatility Pattern in Stock Markets: The Indian Case. IUP Journal of Applied Economics, 13 (4), De, S., & Chakrabarty, T. (2016). FII Flow and Volatility Expectations in Indian Equity Markets. The IUP Journal of Applied Finance, 22 (1), Demir, E., Fung, K. W., & Lu, Z. (2015). Capital Asset Pricing Model and Stochastic Volatility: A Case Study of India. Emerging Markets Finance & Trade. 2016, Vol. 52 Issue 1, p52-6, 52 (1), Gaye Gencer, H., & Demiralay, S. (2016). Volatility Modeling and Value-at-Risk (VaR) Forecasting of Emerging Stock Markets in the Presence of Long Memory, Asymmetry, and Skewed Heavy Tails. Emerging Markets Finance & Trade, 52 (3), Goudarzi, H., & Ramanarayanan, C. (2011 ). Modeling Asymmetric Volatility in the Indian Stock Market. International Journal of Business and Management, 6 (3), Lawrence R. Glosten, R. J. (1993). On the Relation between the Expected Value and the Volatility of the Nominal Excess Return on Stocks. Journal of Finance, 48 (5), M. T. Raju, A. G. (2004, April 2004). Stock Market Volatility An International Comparison. SEBI working paper series No.8. Manjinder Kaur, S. S. (2015). IMPACT OF FOREIGN INSTITUTIONAL INVESTORS INVESTMENT ON INDIAN STOCK MARKET VOLATILITY: A STUDY OF BSE SENSEX. International Journal in Commerce, IT & Social Sciences, 2 (7). Mishra, P. K., Das, J. R., & Mishra, S. K. (2010). Gold Price Volatility and Stock Market Returns in India. American Journal of Scientific Research (9), Mukherjeea, K. N., & Mishra, R. K. (2010). Stock market integration and volatility spillover: India and its major Asian counterparts. Research in International Business and Finance, 24 (2010), Nandy (Pal), S., & Chattopadhyay, A. K. (2014). Impact of Introducing Different Financial Derivative Instruments 36

15 in India on Its Stock Market Volatility. Paradigm, 18 (2), Olowe, R. A. (2009). Stock Return, Volatility And The Global Financial Crisis In An Emerging Market: The Nigerian Case. International Review of Business Research Papers, 5 (4), P. Sakthivel1, K. V. (2014). Impact of Global Financial Crisis on Stock Market Volatility: Evidence from India. Asian Social Science, 10, (10;), Pal, P. (1998). Foreign Portfolio Investment in Indian Equity Markets: Has the Economy Benefited? Economic and Political Weekly, 33 (11), Roy, M. K. (2001). Stock Market in a Liberalised Economy: Indian Experiences. Economic and Political Weekly, 36 (4). Saha S, C. G. (2011). Financial Crisis and Financial Market Volatility Spill-Over. The International Journal of Applied Economics and Finance, 5, Samal, K. C. (1997). Emerging Equity Market in India: Role of Foreign Institutional Investors. Economic and Political Weekly, 32 (42), Sarkar, A., Chakrabarti, G., & Sen, C. (2009). Indian stock market volatility in recent years: Transmission from global market, regional market and traditional domestic sectors. Journal of Asset Management, 10 (1), Seďa, P. (2012). Impact of the global financial crisis on stock market volatility: Evidence from Central European stock market. Proceedings of 30th International Conference Mathematical Methods in Economics, (pp ). Shenbagaraman, P. (2003). Do futures and options trading increase stock market volatility? NSE News Letter, NSE Research Initiative(20). Srinivasan, P. (2015). Modeling and Forecasting of Time-Varying Conditional Volatility of the Indian Stock Market. The IUP Journal of Financial Risk Management, XII (1), Wachter, J. A. (2013). Can Time-Varying Risk of Rare Disasters Explain Aggregate Stock Market Volatility? Journal of Finance, 68 (3), APPENDIX : Quarter 1: Kw Anova Result (Independent Samples) Table 1A Total N 378 Test Statistic Asymptotic Sig. (2-sided Test)

16 Table 1B Null Hypothesis Test Sig. Decision 1. The distribution of CV is the same across categories of index.000 Reject the null Asymptotic significances are displayed. The significance level is : Quarter 2: KW Anova Result (Independent Samples) Table 2A Total N 378 Test Statistic Asymptotic Sig. (2-sided Test).000 Table 2B Null Hypothesis Test Sig. Decision 1. The distribution of CV is the same across categories of index.000 Reject the null Asymptotic significances are displayed. The significance level is : Quarter 3: KW Anova Result (Independent Samples) Table 3A Total N 366 Test Statistic Asymptotic Sig. (2-sided Test).000 Table 3B Null Hypothesis Test Sig. Decision 1. The distribution of CV is the same across categories of index.000 Reject the null Asymptotic significances are displayed. The significance level is

17 : Quarter 4: KW Anova Result (Independent Samples) Table 4A Total N 372 Test Statistic Asymptotic Sig. (2-sided Test).000 Table 4B Null Hypothesis Test Sig. Decision 1. The distribution of CV is the same across categories of index.000 Reject the null Asymptotic significances are displayed. The significance level is : Quarter 1: KW Anova Result (Independent Samples) Table 5A Total N 378 Test Statistic Asymptotic Sig. (2-sided Test).000 Table 5B Null Hypothesis Test Sig. Decision 1. The distribution of CV is the same across categories of index.000 Reject the null Asymptotic significances are displayed. The significance leve l is : Quarter 2: KW Anova Result (Independent Samples) Table 6A Total N 378 Test Statistic Asymptotic Sig. (2-sided Test)

18 Table 6B Null Hypothesis Test Sig. Decision 1. The distribution of CV is the same across categories of index.000 Reject the null Asymptotic significances are displayed. The significance level is : Quarter 3: KW Anova Result (Independent Samples) Table 7A Total N 372 Test Statistic Asymptotic Sig. (2-sided Test).000 Table 7B Null Hypothesis Test Sig. Decision 1. The distribution of CV is the same across categories of index Asymptotic significances are displayed. The significance level is Reject the null : Quarter 4: KW Anova Result (Independent Samples) Table 8A Total N 378 Test Statistic Asymptotic Sig. (2-sided Test).000 Table 8B Null Hypothesis Test Sig. Decision 1. The distribution of CV is the same across categories of index.000 Reject the null Asymptotic significances are displayed. The significance level is

19 : Quarter 1: KW Anova Result (Independent Samples) Table 9A Total N 360 Test Statistic Asymptotic Sig. (2-sided Test).000 Table 9B Null Hypothesis Test Sig. Decision 1. The distribution of CV is the same across categories of index.000 Reject the null Asymptotic significances are displayed. The significance level is : Quarter 2: KW Anova Result (Independent Samples) Table 10A Total N 378 Test Statistic Asymptotic Sig. (2-sided Test).000 Table 10B Null Hypothesis Test Sig. Decision 1. The distribution of CV is the same across categories of index.000 Reject the null Asymptotic significances are displayed. The significance level is : Quarter 3: KW Anova Result (Independent Samples) Table 11A Total N 348 Test Statistic Asymptotic Sig. (2-sided Test)

20 Table 11B Null Hypothesis Test Sig. Decision 1. The distribution of CV is the same across categories of index.000 Reject the null Asymptotic significances are displayed. The significance level is : Quarter 4: KW Anova Result (Independent Samples) Table 12A Total N 372 Test Statistic Asymptotic Sig. (2-sided Test).000 Table 12B Null Hypothesis Test Sig. Decision 1. The distribution of CV is the same across categories of index.000 Reject the null Asymptotic significances are displayed. The significance level is : Quarter 1: KW Anova Result (Independent Samples) Table 13A Total N 366 Test Statistic Asymptotic Sig. (2-sided Test).003 Table 13B Null Hypothesis Test Sig. Decision 1. The distribution of CV is the same across categories of index.003 Reject the null Asymptotic significances are displayed. The significance level is