Does Firm Size Matter in Herding Behavior? Evidence from Quantile Regression Analysis in the Pakistan Stock Exchange

Abstract

This study examines whether herding behavior varies with firm size in the Pakistan Stock Exchange (PSX) and evaluates whether quantile regression provides additional evidence beyond conventional ordinary least squares (OLS) estimation. Using daily stock-price data for 128 firms from August 1, 2016, to September 30, 2024, firms are classified according to market capitalization, with particular attention given to the smallest and largest firms. Herding is examined using the cross-sectional standard deviation (CSSD) and cross-sectional absolute deviation (CSAD) approaches during extreme market movements defined by the 5% criterion and under nonlinear specifications. In addition to OLS estimation, quantile regression is applied on the lower quantiles (t=10%, 25%) and on the upper quantiles (t=75%, 90%) quantiles to capture heterogeneous herding behavior across the conditional distribution of returns. The OLS results provide limited evidence of herding, primarily among large firms during downward market movements, and fail to detect nonlinear herding across either firm-size group. In contrast, quantile regression identifies more pronounced herding at lower quantiles, particularly during downward market movements, for both the smallest and largest firms using the CSAD specification. Nonlinear quantile estimation also identifies herding among the smallest firms at the 25% quantile. Overall, the findings demonstrate that herding is state-dependent and more evident during periods of market stress, while quantile regression provides greater sensitivity than OLS in detecting heterogeneous herding behavior in the PSX. The study contributes to the literature by highlighting the importance of firm size and distributional heterogeneity when examining investor herding in emerging markets.