Time-varying return predictability and adaptive markets hypothesis: Evidence on MIST countries from a novel wild bootstrap likelihood ratio approach

Abstract

This paper investigates the evolution of the return predictability (or market efficiency) degree for Mexico, Indonesia, South Korea, and Turkey (MIST) countries and examines whether the findings are consistent with the implications of adaptive markets hypothesis (AMH). For this purpose, the novel wild bootstrap likelihood ratio approach of Kim and Shamsuddin (2020) is applied on the monthly data from January 1993 to July 2020 with the rolling sub-sample windows method to determine whether the ability of inflation and trading volume to predict stock market returns varies over time. The empirical findings verify that the return predictability (or market efficiency) is time-varying consistent with the implications of the AMH for all MIST countries. This verification is also strengthened by using other predictor variables, namely, exchange rate and realized volatility. This paper reveals that the AMH is more successful in explaining real stock market behavior than efficient markets hypothesis