This paper investigates the random walk behavior of currency returns for 15 African countries, namely, Ethiopia, Gambia, Ghana, Kenya, Madagascar, Mali, Mauritius, Morocco, Nigeria, Rwanda, Sierra Leone, South Africa, Tanzania, Uganda, and Zambia. In particular, the paper applies the nonparametric variance-ratio tests proposed by Wright (2000). The two important findings that emerge from this study are: first, contrary to most of the earlier studies on this issue, this study finds evidence against random walk behavior for currency returns of all sample countries; second, the rejection of the null hypothesis of random walk suggests that shocks to the currency return series are temporary. This finding implies that foreign exchange markets of the sample countries are inefficient and present opportunities to aggressive investors who seek high rates of return irrespective of the level of risk involved.

Shocks are considered to be permanent if currency returns are characterized by random walk behavior. This would imply that future movements in currency returns cannot be predicted using past information or historical data. However, if currency returns are not random walk processes, then their future movements can be predicted using past information or historical data. The issue as to whether or not currency returns exhibit a random walk behavior is important to investors who seek to exploit the opportunities created by inefficiencies in foreign exchange markets. Chen and Jeon (1998) point out that the dollar price of a foreign currency can be represented by the nominal exchange rate, especially when it is expressed as the quantity of the US dollar per unit of foreign currency. For instance, the nominal exchange rate given by dollar/currency can be interpreted as the dollar price of a foreign currency. Similar to the prices of common stocks and bonds, the dollar price/va lue of a foreign currency changes over time. Based on this notion, Chen and Jeon (1998) suggest that foreign currency can be viewed as a form of financial asset, which they referred to as currency asset.

A number of studies have examined the random walk behavior of currency returns and have produced mixed results. For instance, Wright (2000), using ranks and sign tests, rejects the notion that currency returns are random walk processes for the US. In contrast, Belaire- Franch and Opong (2005), applying the nonparametric variance-ratio tests, find evidence in support of the hypothesis that Euro currency returns exhibit random walk behavior. Chang (2004), utilizing the variance-ratio tests, reexamines the random walk hypothesis for Canadian dollar, French franc, Deutsche mark, Japanese yen, and the British pound for the period August 7, 1974-December 30, 1998. The results for the full sample reject the null hypothesis that currency returns for the sample countries follow random walk. However, the results for the sub-period 1989-1998 fail to reject the null hypothesis that currency returns for the Canadian dollar, French franc, Deutsche mark, and the British pound are random walk processes. Liu and He (1991), applying the standard variance-ratio test and data spanning over August 7, 1974 to March 29, 1989, find that Canadian dollar, French franc, Deutsche mark, and the British pound are not random walk processes. Other studies, including Giddy and Dufey (1975), Cornell and Dietrich (1978), Logue et al. (1978), and Hsieh (1988), suggest that exchange rates exhibit random walk behavior and based on this finding, they concluded that movements in exchange rates cannot be predicted using their past information. Chen and Jeon (1998), applying the variance-ratio and the regression test, investigate the random walk behavior of currency returns and show that currency returns exhibit random walk behavior.

Applied Economics Journal, Testing Random, Walk Behavior, Currency Returns, Ethiopia, Gambia, Ghana, Kenya, Madagascar, Mali, Mauritius, Morocco, Nigeria, Rwanda, Sierra Leone, South Africa, Tanzania, Uganda, African Countries.