Return Predictability and Market Conditions : Evidence from Nigerian, South African and Mauritian Stock Markets

Recently, debate on stock market efficiency has shifted to the investigation of changes in return predictability under different market conditions, notably in few developed markets. This paper investigated adaptive market hypothesis (AMH) in Nigerian, South African and Mauritian stock markets with a focus on varying return predictability-cum-market conditions. The sample period spanned from January 1998 to September 2017 and comprises daily returns of all share index. P-values of joint variance ratio test and BDS test, generated by implementing the tests in two-year rolling windows, rolled forward by one month, were adopted as monthly measures of linear and non-linear return predictability. It was discovered that up, bull, bear and normal market conditions are associated with low linear return predictability in the three markets and low non-linear return predictability in Nigerian and Mauritian markets. There are also low and high linear and non-linear return predictability respectively in Nigeria and high linear and non-linear predictability in Mauritius during the crisis era. It is noteworthy that the return predictability behaviour of Nigeria and Mauritius is similar under different market conditions compared to South Africa. Thus, while the exact effect of market conditions on return predictability may not be generalized for all markets, it can be seen that the effect is more similar in some markets than in others. In essence, hypothesis of changing return predictability due to changing market conditions is found but not strong enough in the sampled African markets.________________________________________