CHAPTER 5 RESULT AND ANALYSIS This chapter presents the results of the study and its analysis in order to meet the objectives. These results confirm the presence and impact of the biases taken into consideration, namely herding, optimism, overconfidence and the disposition effect. The result section is further bifurcated into secondary data analysis followed by primary data analysis. 5.1 SECONDARY DATA ANALYSIS 5.1.1 DETERMINING THE PRESENCE AND ANALYZING THE IMPACT OF HERDING: (Refer tables 5.1.1.1 and 5.1.1.2) Results for the Presence of Herding and its Non Linearity with Market as a Whole The results of unit root tests show that both CSSD and CSAD series are stationary. The coefficients of D L t & D U t in equation (1) are both positive and significant at the 1percent level (Table 5.1.1.1) which shows that CSSD increases with an increase in market return. This refutes the hypothesis of herding behavior. The value of γ2 in equation (4) is also positive and significant at the 1percent level, this means that return dispersion are decreasing (or increasing) at an increasing rate (Table 5.1.1.1). This again highlights the fact that herding does not exist in Indian stock market, but indicates the presence of non linearity in the relationship. The curve estimates reveal that dispersion is nonlinearly related to market returns (Fig.1). Results for Bull and Bear Phase of Market Individual tests for bull and bear phases of market in equation (5) and (6) indicate that, herding prevailed when the market was up (as γ UP 2 was negative and significant at 5percent significance interval). However, no evidence of herding has been found when the market was down (negative insignificant γ DOWN 2 ) (Table 5.1.1.2). 66
5.1.2 INVESTIGATING THE PRESENCE AND ANALYZING THE IMPACT OF OPTIMISM (PESSIMISM) The unit root tests to check the stationarity of all the series are done using the ADF (Augmented Dickey-Fuller) test. It was found that daily Nifty50 index return 10, closing prices of both in-the money and out-of the-money call and put option, were stationary at level. Whereas the derived series, including Optimism (Pessimism), objective risk premium and representative risk premium are found to be stationary at first and second difference level respectively. The output estimates of the study consist of GJR GARCH estimates, pricing kernel estimates (M T, t and M T, t (θ) ), estimates of the nonlinear regression (θ 0, θ 1 ) given by Campbell et al. (1997), sentiment function (Λ t ), the representative investors PDF ( pr) and values of optimism (pessimism). Another set of result includes the time series properties of optimism (pessimism) bias. These are discussed in detail as follows. A. Empirical pricing kernel (M T,t ): It is calculated as per equation (7) which requires the objective density (p) and the risk neutral density (q). The densities p and q are subsequently generated using GJR GARCH estimates. Results of GJR GARCH: (Refer table 5.1.2.1) Table 5.1.2.1 reports the GJR GARCH estimates of Nifty 50 returns, in and out of the money put and call options. It is seen that the leverage effect (measured by the interaction dummy variable I t-1 in equation (8) accounts for all the variables and is verified by a positive γ parameter, except in out-of-the money call option where it is negative but insignificant. The parameter γ is positive and significant for in-the-money put option and in-the-money call options. It is insignificant for the Nifty 50 return series and the out-of-the money put option series although the sign is positive. The GJR GARCH estimates are used to generate probability density functions (PDF). The estimates of Nifty 50 returns are used to generate the objective density (p). On the other hand the risk neutral density (q) is calculated with the estimates of closing option prices such that there are four different sets of risk neutral densities each corresponding to a 10 [] Outliers have been removed from the daily Nifty50 return series to ensure better estimation of all the variables. 67
separate option type. These are used to calculate the empirical pricing kernel (M T, t ). Figure 5.1.2.1 and 5.1.2.2 depict the probability distribution of objective and risk neutral density functions. Here the objective density function is associated with correct beliefs and is later used to estimate rational expectations of investors. 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 0 50 100 150 200 250 300 350 400 450 500 Figure 5.1.2.1: Objective PDF generated using daily values of Nifty 50 returns 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 0 50 100 150 200 250 300 350 400 450 500 Figure 5.1.2.2: Risk Neutral PDF generated using daily values of nifty 50 index options B. Theoretical pricing kernel, sentiment function and representative investors pdf: (refer figure 5.1.2.3) It is determined as per equation (9) where θ 0, θ 1 are estimated with the help of a regression equation given by [28]. 68
Results of regression equation [28]: (Refer Table 5.1.2.2) Table 5.1.2.2 displays the result of the regression equation (10) which reveals that the degree of impatience (θ 0 ) is negatively related to the risk-free rate of return (rf), while the degree of risk aversion (θ 1 ) is positively related to rf. The results are in line with the findings of Campbell et al. (1997). Both the values are significant at the 1 percent level. C. Sentiment function and representative investors pdf: The difference between the theoretical pricing kernel calculated with the help of θ 0 and θ 1 and the empirical pricing kernel is used to estimate the sentiment function. The sentiment function, thus derived is used in order to convert the objective PDF into representative investors PDF according to equation (14). As four sets of empirical pricing kernel have been used each for a separate option type, there are four different series of sentiment function and subsequently the same number of series for representative investors PDF. Figure 5.1.2.3 depicts the PDF of representative investors exhibiting optimism (pessimism) bias for in- the- money put option. 0.012 0.01 0.008 0.006 0.004 0.002 0 0 50 100 150 200 250 300 350 400 450 500 Figure 5.1.2.3: Representative investors pdf generated as per equation 14 Finally, the difference between the mean expected return from the representative investors PDF and the objective PDF is taken as a measure of optimism or pessimism depending on the measure being positive or negative. This means that if a majority of the values of the estimate are 69
positive, then it is taken as optimism and if the values are majorly negative then pessimism prevails. D. Presence of optimism (pessimism) bias: (refer Table 5.1.2.3 and 5.1.2.4) There are four sets of measures for optimism (pessimism) each belonging to a separate option type. The results reveal that the measure is majorly negative in the period 2006-2013. The mean value is negative in all the four cases which implies that pessimism has dominated the investor sentiment. One sample t-test is conducted to verify the significance of the pessimism series and it is seen that the series is significant at the 1 percent level. E. Time series regression for estimating the impact of optimism (pessimism) on risk premium: There are 8 sets of result; four each for call options and put options respectively: a. Put options: (Refer tables 5.1.2.5 to 5.1.2.8). The results reveal that the estimate of optimism (pessimism) is negatively related to objective risk premium. However, it is positively related to the representative investors risk premium. The lags of risk premium are negatively related in each case except for in-the-money put options (table 5.1.2.7). Here a positive, but insignificant relationship prevails between the representative investors risk premium and its first and second lags. All the remaining values are significant at the 1 percent level. b. Call options: (See tables 5.1.2.9 to 5.1.2.12) Results are very similar to put options. The measure of optimism (pessimism) is negatively related to the objective risk premium, but it is positively related to the representative investor s risk premium. All the values are significant at the 1 percent level. The only exception is in-the-money call options where the impact of the optimism (pessimism) estimate on representative investors risk premium is insignificant. The lags of risk premium are positively related in each case except for out-of-the-money call options (table 5.1.2.9), where the relationship is negative. The results highlight the fact that the pessimistic scenario pushes the prices of risky assets towards the downside as investors are in a selling spree to dispose of their loss making assets. People also start considering that investment in the stock market is becoming risky which increases their return expectations thereby increasing the objective risk premium. The opposite scenario happens in the case of optimism, when investors increase their buying behavior 70
subsequently pushing the stock market prices up. This leads to a decline in return expectations due to which the objective risk premium gets reduced. So there is a negative relationship between the estimate of optimism (pessimism) and the objective risk premium. However, this relationship is reversed in the case of the representative investors risk premium. The results reveal that, when investors are irrational, a positive relationship exists between the estimate of optimism (pessimism) and their return expectations and risk premium. This means that as the value of this estimate increases/decreases in the presence of optimism/pessimism; the representative risk premium also increases/decreases. This finding is in alignment with the results of [12] where they point out when investors are biased they tend to show a negative relationship between risk and return. F. Time series regression for identifying the impact of volatility on optimism (pessimism): (See tables 5.1.2.13 to 5.1.2.16). There are four sets of results which reveal that lagged volatility is negatively related to the estimate of optimism (pessimism). The value is significant at the 1 percent level for three out of four option types. It is insignificant for in-the money call option; however the sign is still negative. There is also a positive relationship between the estimate of optimism (pessimism) and its lagged value. The result is significant at the 1 percent level. The only exception is out-of-the money call option (table 5.1.2.15) where this relation is negative. The relationship of optimism (pessimism) estimate with lagged value of market return is found to be insignificant. This shows that when past volatility is higher, investors develop a perception that the stock market is not going through a very stable phase. This leads to a decrease in the estimate of optimism (pessimism) making investors pessimistic. On the other hand, when that past volatility is low, it depicts a stable phase and it increases the value of the mentioned estimate and leads to optimism in investors. The results are again in line with the findings of [12]. 71
5.1.3 INVESTIGATING THE PRESENCE AND ANALYZING THE IMPACT OF OVERCONFIDENCE AND THE DISPOSITION EFFECT A. RESULT FOR MARKET-WIDE VAR: (refer table 5.1.3.1) The results reveal that the log transaction volume of Nifty 50 index is positively related to all the lags of market return with second lag of market return being significant. This relationship prevails even after controlling for volume-volatility relationship. This finding indicates the presence of overconfidence in the Indian equity market. B. RESULTS FOR SECURITY-WIDE VAR: (refer table 5.1.3.2a) The lags with significant positive and/or negative coefficients for security return and market return are reported. The interpretation of same is based on the following parameters. i. Positive and significant values of security return lags (Ri) indicate the disposition effect. ii. Positive and significant values of market return lags (Rm) indicate overconfidence. iii. Positive and significant values of both (Ri) and (Rm) lags indicate both disposition effect and overconfidence. iv. Significant positive value of (Ri) lags but significant negative value of (Rm) lags indicate presence of the disposition effect but inconclusive overconfidence bias. v. Positive and significant value of (Rm) lags, but the significant negative value of (Ri) lags indicate presence of overconfidence bias but inconclusiveness of the disposition effect. vi. Negative and significant values of both (Ri) and (Rm) lags indicate both disposition effect and overconfidence bias are inconclusive. vii. The simultaneous presence of significant positive and negative lag for either (Ri) or (Rm) indicates inconclusive bias. Based on these parameters, the results reveal that out of 45 companies, the disposition effect and overconfidence can be detected in 20 firms with an extent of certainty (refer table 5.1.3.2b). More precisely, overconfidence bias is present with 12 firms, the disposition effect is present 72
with 5 firms and 3 firms are affected by both biases. This makes overconfidence bias to be predominant amongst the two. C. RESULTS FOR IMPULSE RESPONSE FUNCTION Market Impulse Response Function (refer Figure 5.1.3 and Table 5.1.3.3 ): Figure 5.1.3: Market impulse response function with two standard error bands Figure 5.1.3 represents the response of market trading volume (LogT) and market return (Rm) to one standard deviation shock in their respective residuals. The results reveal that there is a positive impact of LogT-shock on LogT and it persists for all the periods. This impact is highest for the first period (31percent), declines in the second and third period and thereafter remains stable ranging between 5-6percent. Further, the graph depicts the positive dependence of LogT to Rm shock which persists for all seven periods. This dependence is clearly illustrated in table 5.1.3.3 which shows that the Rm shock leads to a positive and significant 2percent increase in LogT for third period. This finding verifies that market returns impact the investors confidence and subsequently the trading activity. Security Impulse Response Function It investigates the response of all the three endogenous variables, namely, securities trading volume (LogT), security returns (Ri) and market return (Rm), to the shock in their respective residuals. In the present context, the response of LogT to Ri and Rm shock is being discussed, as it brings more clarity to the overconfidence and the disposition effect theory. The empirical evidence for the overconfidence hypothesis can be verified for 11 firms as the trading activity is positively related to past market returns. This can be observed from the response of LogT for market return shock, which is positive and significant. Whereas, the 73
disposition effect theory can be validated for only three firms wherein, the response of LogT to Ri shock is positive. The periods for which these results are significant are also identified. The results of both market and security impulse response are in alignment with the findings of VAR and suggest that overconfidence is predominant of the two biases. 5.2 PRIMARY DATA ANALYSIS The results are based on the options that investors chose with respect to different situations. These choices subsequently unveil the underlying behavioral biases of respondents. A summary of behavioral biases corresponding to each item is presented in Table 5.2. 5.2.1 RESULTS OF INTERNAL CONSISTENCY: (refer table 5.2.1) The reliability or the internal consistency for each bias is checked separately with the help of Cronbach s Alpha. Results reveal that reliability of items pertaining to optimism (pessimism), overconfidence, and herding is greater than the benchmark value of 0.70 which makes them a preferable scale. The reliability of items corresponding to the disposition effect is lower than the accepted benchmark (0.54). However, the present study considers these items since they are in accordance with the literature on the disposition effect bias. 5.2.2 RESULTS FOR CHI SQUARE TESTS: (refer table 5.2.2) The dependence between the behavioral biases and demographic as well as investor sophistication factors is detected with the help of the chi-square test. In all, seven variables are taken into consideration, where the factors like gender, age, educational qualification, profession and income level constitute demographics of the respondents. The remaining two variables are trading experience and frequency which fall under the category of investor sophistication. The detailed analysis of this association for each variable is discussed below. Gender: There is a significant difference in the responses of male and female for 16 items. These 16 items represent all four biases in the purview of our study. The results are verified by significant chi-square value at the 5 percent and the 1 percent level. This shows that there is an association between gender of the respondents and their behavioral biases. Age: Out of a total 26 items, the responses of 24 items vary as one moves from age group I (i.e. 20-30 years) to age group V (51-60 years). The results reveal that there is a strong association 74
between the age group and the behavioral biases of respondents and is confirmed by significant chi-square value at the 1 percent level. Educational qualification: This variable has 4 categories that include undergraduates, graduates, post graduates and doctorates. The responses of only 6 items differ with each education class. However, these 6 items still capture all the four biases. The results are significant at the 5 percent and the 1 percent level. Profession: The profession of respondents has been divided into 5 subgroups that are: employees of PSUs and government sectors (excluding banks), private sectors (excluding banks), public and private sector banks, financial experts (like market advisors and chartered accountants) and the rest who are employed otherwise fall into others category. It is seen that as one moves from one class of profession to another the decision choices differ for 22 items. This difference is significant at the 5 percent and the 1 percent level. Income: There is a difference in choices of 9 items across different income classes. These classes include income level less than 2-4 lakhs, 4-6 lakhs, 6-8lakhs, 8-11 lakhs and income greater than 11 lakhs. The association is confirmed for the four biases by significant chi square value at the 5 percent and the 1 percent level. Trading Experience: The responses of 10 items vary with respect to change in respondents trading experience. This variable ranges from low (less than a year) to high (more than 7 years). This result is substantiated by a significant chi square value at the 5 percent and the 1 percent level and it captures all the biases of the study. Trading Frequency: the investors are categorized based on their trading frequency as intraday traders, 0-3months, 3-12 months, 12-36 months and greater than 36 months. The chi square results show that investor responses to 20 items out of 26 items alter with a change in their trading frequency. The results are significant at the 1 percent level. The chi square results illustrate that there is dependence between investors demographics and sophistication factors and their behavioral biases. This association is confirmed for all 7 variables and all 4 behavioral biases so that the null hypothesis H0 1 and H0 2 can be rejected. The results also reveal that age of an investor creates the highest difference in the behavioral biases followed by trading frequency and profession of the investors. 75
5.2.3 RESULTS FOR INDEPENDENT SAMPLE T-TEST This test gives a more lucid view as to the investor specific characteristics corresponding to each bias. It helps in determining the change in the values of biases with respect to changes in demographic factors and variables measuring investor sophistication. The influence of each variable is described below. Gender (refer table 5.2.3.1) 11 : The test results show that gender does not have much impact on investors decision making process. The responses significantly differ in only two instances. It is seen that men are more overconfident than women with respect to their knowledge of the Indian stock market (B1). On the other hand, women are more pessimistic than men as they expect the gold prices to improve in the next quarter (A4). Thus, it can be said that the values of only optimism and overconfidence change with respect to gender. Age: (refer table 5.2.3.2) The test reveals that the respondents belonging to age group IV (51-60 years) are highly vulnerable as they are prone to all the four biases. It is seen that this category is optimistic in their outlook to the Indian equity market (A1) and feel that NSE can recover a loss of 3percent within a few days (B7). However, the respondents of other age groups (I, II, and III) either disagree or take a neutral stand in this situation. In addition, age group I and III believe that the gold prices will improve in future (A4). They also think mostly about potential losses before considering any investment option (A6). Both these responses indicate the pessimistic tendency in age group I and III. Taking into consideration the overconfidence, it is seen that all the age groups are affected by this bias. However, the situations where they show overconfidence are different. Age group IV is overconfident about their knowledge of the Indian stock market (B1), while age group III believes that they have the ability to pick better stock than others (B2) and age group II agrees that their past investment successes are attributed only to their own skills and understanding (B3). Further, the disposition effect influences the trading volume of only age group II where they agree that past investment success make them invest more in stocks (B6). As far as herding 11 Refer Table2 of descriptive statistics for the codes of categories in each variable. 76
is concerned all age groups give due importance to the opinion of their peer group, except group II. Group IV shows additional herding tendency by agreeing that discussing investment decisions with colleagues reduces their pressure of being successful. Profession: (refer table 5.2.3.3) Profession influences optimism (pessimism), overconfidence and the disposition effect but, does not create any difference in herding bias. Further, it has an impact on the three biases under specific situations. Group IV is optimistic (A1 and B7), overconfident (B1 and B2) and also shows the disposition effect (B6). However, groups I, II, and III are pessimistic (A4) or slightly optimistic (A1) with respect to the Indian equity market. Group II is neutral about investing in the stock market in next quarter while group V is optimistic about it (B7). Group I is overconfident as they attribute the success only to their own skills and understanding (B5) while the other groups either disagree (group V) or have a neutral opinion on this statement (Group II, III and IV). Finally, the impact of the disposition effect (B6) is seen in group I, II and IV but not in group III and V. Income: (refer table 5.2.3.4) It is seen that income level influences three out of four biases which includes overconfidence, optimism (pessimism), disposition effect but not herding. It is observed that income classes I, IV and V are less pessimistic and more optimistic than income classes II and III as they think that gold prices will remain stable. They think a little about potential loss before investing and they feel that NSE can recover within few days. This in contrast with classes II and III who take into account mostly the potential loss from an investment (A6) and think that gold prices will improve in future (A4). Income class II is neutral about the recovery of NSE. The results also show that income class IV and V are susceptible to overconfidence while classes III and IV are prone to the disposition effect. Here, overconfidence is seen with respect to knowledge of the Indian equity market (B1) and the ability to pick better stocks than others (B2) while, the disposition effect is seen with respect to item B6. Investment type: (refer table 5.2.3.5) A major difference can be seen between the investors of old companies (group II) and new companies (group I). This variation is observed in 4 items corresponding to overconfidence, 77
optimism and herding. The results reveal that the investors of new companies (group I) are slightly more optimistic than the investors of old companies, derivatives and commodities and high grade corporate bonds (group II, III and IV respectively). The pessimism of group II, III, and IV is suggested by the fact that they think mostly about potential loss before investing (A6). In addition to this, group II and IV feel that the gold prices will improve in next six months (A4) which further suggests their pessimism towards the Indian equity market. The difference between group I and other categories also lies for other biases like overconfidence and herding. Group I is overconfident that their knowledge of the Indian equity market is sufficient (B1) while other groups take a neutral stance. This group is also prone to herding as they feel that discussing their investment decisions reduces their pressure of being successful (B4). Trading Experience: (refer table 5.2.3.6) It is observed that with increase in experience, the investors become more prone to overconfidence. For instance, group III-V agree on situations representing overconfidence. They agree that they have sufficient knowledge of stock market (B1), they have the ability to pick better stocks (B2), and they alone are fully responsible for their investment performance (B3). All the groups are prone to the disposition effect as tend to sell their winners early to lock in their gains (A8), except group II. The groups IV and V are optimistic of the Indian stock market as they feel that gold prices will remain stable (A4), they plan to increase their investment in next quarter (B7), and also feel that NSE can recover from a fall within few days (B16). Herd mentality only affects the highest (V) and lowest experience class (I). However, the situations in which they try to herd may vary. For class I the opinion of peer group is considered to be important. For class V, discussion with colleagues reduces the pressure of being successful as the respondents seek confirmation from them on their decisions. Conversely, when they take contrarian position from the group and fail, they become highly disappointed. In an ideal situation the decision of crowd should not matter if they are not herding. Trading Frequency: (Refer table 5.2.3.7) Intraday traders are found to be prone to all the 4 biases. They are optimistic, overconfident, herd and have the disposition effect on the selling side. Investors who trade less frequently are found 78
to be more cautious in comparison to intraday traders. Elaborating further, it is seen that intraday traders are optimistic about trading in stock market in next quarter (B7). They also think that NSE can recover within few days after falling 2-3percent (B16). These traders are prone to the overconfidence bias, as they agree that they have sufficient knowledge of stock market (B1). Intraday traders also herd as they consider the review of their peers to be important (A9) and discussing these decisions with colleagues reduces this pressure (B4). The only bias which affects all the categories equally is the disposition effect as all of them would sell their stocks to lock in their gains. However, this bias has an additional influence on intraday traders as they tend to increase their trading activity after experiencing past success in their stocks (B6). The results of independent sample t-test illustrate that investors with certain characteristics are prone to a specific bias. These characteristics can be summarized to develop an investor profile related to each bias (refer table). The same is being discussed here (refer table 5.2.3.8). Overconfidence affects those investors who are male (group I), lying under the age group of 31-60 years (group II, III and IV), with annual income either less than 2 lakhs (class I) or 8-11 lakhs (class IV). These investors mostly invest in new companies with high growth (group I), with a trading experience of 3 years or more (group III, IV, and V) trade on intraday basis (group I). They are mostly employed with PSU s and Government sector (group I) and/or they can be financial experts (group IV). Optimism is observed in men (group I), with the age group of 51-60 years (group IV), annual income less than 2 lakhs (class I) and greater than 8 lakhs (class IV and V), who invest in new companies with high growth (group I), with an experience of more than 5 years (group IV and V) intraday traders and financial experts. On the other hand, pessimism is observed in women, with age group II1-30 years (group I) and 41-50 years (group III), having annual income between 2 to 8 lakhs. It is also seen in respondents who invest in stocks of old companies (group II), derivatives and commodities market, and high grade corporate bonds. Herd behavior is seen in relatively old investors of age 51-60 years (group IV), those who invest in new companies with high growth (group I), with very low experience (less than a year) or very high experience (greater than 7 years), and are intraday investors (group I). 79
Finally, the disposition effect influences men and women equally. However, it is seen clearly in investors coming within the age group of 31-40 years (group II), with annual income 5-11 lakhs (group III and IV), and an experience of less than one year or more than 3 years (group III, IV and V). The investors prone to this bias are either public and private sector employees (excluding banks) or financial experts (group I, II and IV) and they trade on intraday basis (group I) or 0-3 months. All the results are significant at the 5 percent and 1 percent level. 5.2.4 Results for one sample t-test (Part-A): (Refer Table 5.2.4) It can be seen that 44.6percent of respondents are slightly optimistic towards the outlook of the Indian equity market (A1). When asked about the average return of the Indian equity market in the past 15 years, 54.29percent respondents gave a realistic estimate (A2) and 43.9percent are sure about it (A3). Their perspective was cross checked with their future estimate on gold prices (A4). As discussed in earlier sections, a highly optimistic view for gold prices indicates that investors are uncertain about the performance of the stock market. This makes them channelize their demand towards gold, which is considered to be a safe asset. In the present study, 33.7 percent, investors feel that gold prices will improve in next six months and an equal percentage of the sample feels that the prices will remain stable. 69.3percent respondents are sure about their gold price estimates (A5). This result provides a mixed viewpoint of investors towards the Indian equity market. However, greater clarity on respondents outlook is developed from responses in subsequent items in part B. Taking into consideration, what investors look for in an investment (A6), it is found that 32.4percent look for security of investment i.e. risk versus return and 31.4percent think mostly about potential gain from an investment. The remaining sample respondents either think about losses or both gains and losses. Items A7 and A8 correspond to the disposition effect in investors. These situations testify the two legs of this bias i.e. keeping the losing shares and disposing winners early. The investor responses show that 44.6percent individuals would remain invested in stock whose price falls by a certain percent, as they look for long term growth. On the other hand, 50.4percent individuals 80
would sell their wining stocks to lock in their gains. It could be inferred from the results that these individuals are prone more to the selling side of the disposition effect. Investors were also found to be prone to herding bias as 47.1percent individuals consider their peers to be an important source of information and 51.4percent consider the opinions of market experts to be important. 5.2.5 Results for Rank-test (Part B): The responses for Part B are captured by 5 point Likert-scale that shows the agreeability of respondents with respect to their perception about their own trading behavior and hypothetical stock market situations. The items in Part-B cover four behavioral biases i.e. optimism (pessimism), overconfidence, the disposition effect and herding. As explained in the methodology section, ranking method is used to investigate prevalence of these biases. For this purpose the initial step is to calculate the mean level of importance and test its significance using one sample t-test. It is seen that two statements (B5 and B8) have insignificant means. These two statements along with control items (B10 and B11) are excluded from the analysis. Thus, out of 16 Likert scale items, 12 statements are applicable for further research. Results of Overall Ranking: (Refer table 5.2.5.1) Table (5.2.5.1) shows that the highest importance is given to the statement B3 as 57.1percent respondents agree that they take full control and responsibility of their portfolio performance (mean importance equals 3.84). This is followed by item B2, which states that respondents are confident of their ability to pick better stock than others. 52.61percent respondents agree with this statement and the mean level of importance is 3.67. The items B2 and B3 correspond to overconfidence bias. Third rank of importance is given to B7 where investors agree that they plan to increase their investment in the next quarter, which shows their optimism about the Indian equity market. Rank 4 and 5 are given to items B15 and B1, which pertain to herding and overconfidence. The item B16 corresponding to the optimism of investors on the recovery of NSE takes rank 6. The remaining of the items capturing herd behavior (B4, B14 and B13) gets ranks7, 9 and 10 respectively. The items on disposition effect get relatively lower ranks than all other biases wherein statement B6 is given rank 8 and B12 and B9 are given lowest ranks i.e. 11 and 12. The results are significant at the 5 percent and the 1 percent level. 81
The overall ranking provides a broad idea on which statement is given the highest importance. However, it does not illustrate the same result for the four biases since items capturing the same bias have different ranks. For instance items B3, B2, and B1 relate to overconfidence and they take ranks 1, 2 and 5 respectively. This makes it inconclusive for the researcher to ascertain a particular rank to overconfidence. Similar case prevails in all the other biases. Therefore, to get a clear picture of the order of prevalence, bias wise ranking of items is done, which is followed by the consolidation of means. Bias- wise ranking: (Refer tables 5.2.5.2) It is seen that item B3 gets the highest level of importance (3.83) under overconfidence bias. Taking into consideration the optimism bias, the results reveal that statement B7 is given the highest importance (3.67). The respondents show highest herding tendency with the item B15 (mean level of importance = 3.45). They agree to statement B15 where they feel extremely disappointed if they go opposite to the general trend and lose while their friends make money. Finally, the disposition effect is observed highest in the situation where people increase their trading activity in stocks after experiencing past success (B6, the mean level of importance = 3.37). The results are significant at the 1 percent level. Order of the prevalence of biases: (refer table 5.2.5.3) After consolidation of mean values of importance each bias gets one value and based on this value the ranks are given starting from highest to lowest. The results reveal that overconfidence is most prevalent bias with the highest mean (3.64) closely followed by optimism (3.54) and herding (3.10). The disposition effect is found to be least important bias with a mean of (2.83). 82