Adaptive market hypothesis and calendar anomalies in selected African stock markets.
Loading...
Date
2019
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
It takes a theory to beat a theory. However, whether the adaptive market hypothesis (AMH) offers better explanations for stock return behaviour than the popular efficient market hypothesis (EMH) still remains a question for serious empirical investigation. This question informed the analyses of efficiency and calendar anomalies in the selected African stock market, namely the Nigerian Stock Exchange (NGSE), the Johannesburg Stock Exchange (JSE), the Stock Exchange of Mauritians (SEM), the Casablancan Stock Exchange (MOSE) and the Tunisian Stock Exchange (TSE) with the sample period spanning from January 1998 to February 2018. The first objective of this study is to investigate whether market efficiency changes in cyclical version over time, according to the AMH. The second objective is to evaluate the effect of market conditions (up, down, bull, bear, normal) on return predictability. The third objective is to analyse whether calendar anomalies disappear and reappear over time. The fourth objective is to determine how the anomalies behave under different bull and bear market conditions.
Various linear testing tools such as the variance ratio test, the autocorrelation test, the unit root tests and the nonlinear of BDS were implemented in rolling window approach to track time-variation in efficiency. A dummy regression model was used to evaluate the market condition effect on return predictability. This study also explored rolling window analyses of several alternative variants of nonlinear models of the GARCH family, to track variation in the behaviour of days-of-the-week (DOW), months-of-the-year (MOY) and intra-month effects. Lastly, the study modelled the switching behaviour of the calendar anomalies under bull and bear conditions by using the Markov switching model (MSM), which is able to generate regime-specific regression results for the calendar anomalies under consideration.
Findings from the various linear and nonlinear tests revealed that there are cycles of significant linear and nonlinear dependence and independence in each of the five markets, suggesting bouts of predictability and unpredictability. The regression analyses of return predictability against series of market condition dummies revealed that highIt takes a theory to beat a theory. However, whether the adaptive market hypothesis (AMH) offers better explanations for stock return behaviour than the popular efficient market hypothesis (EMH) still remains a question for serious empirical investigation. This question informed the analyses of efficiency and calendar anomalies in the selected African stock market, namely the Nigerian Stock Exchange (NGSE), the Johannesburg Stock Exchange (JSE), the Stock Exchange of Mauritians (SEM), the Casablancan Stock Exchange (MOSE) and the Tunisian Stock Exchange (TSE) with the sample period spanning from January 1998 to February 2018. The first objective of this study is to investigate whether market efficiency changes in cyclical version over time, according to the AMH. The second objective is to evaluate the effect of market conditions (up, down, bull, bear, normal) on return predictability. The third objective is to analyse whether calendar anomalies disappear and reappear over time. The fourth objective is to determine how the anomalies behave under different bull and bear market conditions.
Various linear testing tools such as the variance ratio test, the autocorrelation test, the unit root tests and the nonlinear of BDS were implemented in rolling window approach to track time-variation in efficiency. A dummy regression model was used to evaluate the market condition effect on return predictability. This study also explored rolling window analyses of several alternative variants of nonlinear models of the GARCH family, to track variation in the behaviour of days-of-the-week (DOW), months-of-the-year (MOY) and intra-month effects. Lastly, the study modelled the switching behaviour of the calendar anomalies under bull and bear conditions by using the Markov switching model (MSM), which is able to generate regime-specific regression results for the calendar anomalies under consideration.
Findings from the various linear and nonlinear tests revealed that there are cycles of significant linear and nonlinear dependence and independence in each of the five markets, suggesting bouts of predictability and unpredictability. The regression analyses of return predictability against series of market condition dummies revealed that high predictability is associated with the bull, volatility and financial crisis periods, especially in NGSE, SEM and TSE and not in others. It suggests that the effect of market condition cannot be generalised for all markets. Further, rolling GARCH estimations showed that calendar anomalies disappear and reappear over time in line with the AMH. The evaluation of calendar anomaly under AMH provides a clearer picture of the behaviour of African stock markets as adaptive. Finally, the empirical results revealed that regime-switching is an important feature of calendar anomalies and that a calendar anomaly that is found in a bull regime tends to disappear or weaken in a bear regime and vice versa, depending on the market and the calendar anomaly in question.
This study adds to the extant literature on the AMH in Africa and global markets. First, it shows that African stock markets are adaptive. Thus, it is more appropriate to describe African markets as adaptive markets rather than inefficient markets. Secondly, it provides empirical evidence of efficiency cum market condition in African stock markets. Thirdly, the study represents a timely contribution on calendar anomalies under AMH in African stock market. Fourthly, by evaluating DOW, MOY and HOM effects under AMH, this study extends the existing works on Monday and January effects in developed markets. Additionally, this study shows the usefulness of MSM in evaluating calendar anomalies under AMH.
Description
Doctoral Degree. University of KwaZulu-Natal, Durban.