Testing for herding behavior in stock market: BIST30 index example

Abstract

Purpose- In the study, using weekly data of 19 companies that are constantly traded in the BIST 30 index, in the period 06.05.2012-27.09.2020, Christie and Huang's (1995) and Chang et al. (2000) approach was aimed to test herd behavior. Methodology- In the study test of herd behavior , Christie and Huang (1995) and Chang et al. (2000) approaches were used. First, as dependent variable cross-sectional absolute deviations for the price series are calculated. Independent variables are assigned as dummy variables for the lower/upper extreme values of 1% and 5% of the market return. Afterwards, estimations were made with Least Squares and Quantile Regression methods. Findings- As a result of the estimation made with Least Squares and Quantile Regression methods, While the coefficient of β_1 for the 1% return slice is statistically significant and the positive, coefficient β_2 as statistically significant and negative, for the 5% return slice, both times the coefficient are statistically significant and positive. The coefficients should be statistically significant and negative in order to pay to show the presence of herd behavior. As a result of the findings, it can be mentioned that 19 firms that are constantly traded in the BIST 30 index have herd behavior for a 1% return slice in the 2012-2020 period. Conclusion- As a result of the analyzes made to determine herd behavior in the period of 06.05.2012-27.09.2020 by using weekly data of 19 companies that are continuously traded in the BIST 30 index, the existence of herd behavior in the relevant period observed. Accordingly, it can be said that investors investing in the BIST 30 index trust the information of other investors when their stock prices tend to fall and invest with this information.