Modeling volatility dynamics of the Nepal Stock Exchange Index: Evidence from GARCH-family models

Abstract

This study examined the volatility dynamics of the Nepal Stock Exchange Index using GARCH- based frameworks on 1,103 daily observations from November 2020 to July 2025. The Bai–Perron  structural break test showed a single regime. The GARCH(1,1) model revealed strong volatility persistence and clustering, whereas the EGARCH(1,1) model identified a clear leverage effect, showing that negative shocks amplify volatility more than positive shocks. The PARCH(1,1) model additionally captured nonlinear and asymmetric volatility behaviour, supported by a significant power term. Diagnostic tests validated that all models were correctly specified and free from residual autocorrelation. Among the estimated models, the PARCH(1,1) specification provided the best fit, revealed by a log-likelihood value of 4080.904 and an AIC of –7.3955. The overall findings revealed that NEPSE returns exhibit persistent, clustered, asymmetric, and nonlinear volatility patterns, and the PARCH(1,1) model most effectively captured all these dynamics, offering practical implications for investors and policymakers.