آزمون پیش بینی در بازارهای سهام در حال ظهور

In this paper we test whether returns for emerging stock markets are predictable. We analyze predictability by means of multivariate variance ratios using heteroscedastic robust bootstrap procedures. Empirical results suggest that emerging equity indices do not resemble a random walk while for developed country indices (US and Japan) we are not able to reject this hypothesis. Furthermore, by employing variable moving average (VMA) and trading range break (TRB) technical trading rules we show that there is some evidence of forecasting power. However, when we take into account trading costs and a buy and hold strategy, only a few rules generate positive excess returns. We check for robustness by analyzing returns from 1559 different trading rules, testing different sub-samples, analyzing returns in bear and bull markets, and also comparing results found for emerging markets to the US and Japan. Furthermore, for the US the Variable Moving Average trading rules suggested in Brock et al. [J. Finance 47 (1992) 1731] do not seem to have forecasting power for the recent sample used, which could be due to the fact that these rules have been widely employed by market participants having the potential abnormal gains from them disappear.

Emerging markets have received massive inflows of capital in the past and have become interesting alternatives for investors seeking diversification. Indeed, Harvey (1995) shows that emerging markets provide investment opportunities for world investors. In general, emerging markets offer high expected returns with an associated high risk. The aim of this paper is to assess whether emerging stock markets in Latin America and Asia exhibit predictability by testing the random walk hypothesis (RWH) using a multivariate version of the variance ratio (VR) test with a heteroscedastic robust bootstrap procedure and by testing technical trading rules such as the variable moving average (VMA) and trading range break (TRB) levels. We study Argentina, Brazil, Chile and Mexico in Latin America, and India, Indonesia, Korea, Malaysia, the Philippines, Thailand, and Taiwan in Asia. We also present results for Japan and the US for comparison purposes.1 Most papers use variable and fixed moving average trading rules to assess whether there are significant profits to be made by technical analysis, following the seminal paper of Brock et al. (1992). The current paper analyzes whether technical analysis possesses forecasting power for price changes by studying the forecast power of 1559 different trading rules using a bootstrap procedure. The VMA technical trading rules proposed by Brock et al. (1992) are used as a benchmark.2 We also test trading rules using TRB levels. Furthermore, the analysis is performed on a very recent sample—the period from January 1991 through January 2004—in which most countries have liberalized their current accounts and received massive foreign portfolio inflows. We compare sub-samples (before and after the Asian crisis) and check whether the results are robust to the period used in the analysis. We also study the results of trading rules in bear and bull sub-samples.3 The remainder of the paper is organized as follows. Section 2 presents a brief literature review. The empirical methodology is discussed in Section 3. In Section 4 we describe the data and the sample used in this paper. Section 5 shows empirical results. Finally, Section 6 concludes the paper.

From multivariate VR statistics evidence suggests that emerging equity market indices do not resemble a random walk. On the other hand, for the US and Japan we are not able to reject the RWH using the same methodology, which is in line with the findings of Chow and Denning (1993) for the US using another multivariate VR statistic. These findings suggest that there is predictability in emerging equity returns. Our empirical results for trading rules show some evidence of predictability, which is in line with those found with the use of VR. However, the empirical results suggest that, in general, trading rules do not generate statistically significant profits after taking into account both transaction costs and a buy and hold strategy. Furthermore, on average TRB rules tend to perform worse than moving average trading rules. The technical trading rules employed in this study do not seem to have forecasting power for the US. This could be due to the immense literature that has focused on this market for the past decade. If traders and analysts read financial papers, since the literature has pointed out that these trading rules have significant forecasting power, the widespread use of these rules could have wiped out the potential for gains from them. In Chile's case, Parisi and Vasquez (2000) find evidence providing strong support for technical strategies. Our paper suggests that these opportunities seem to have been decreased substantially. Nonetheless, in this paper we have used a more recent data set and compared the returns for trading rules with a buy and hold strategy, which could explain the difference in the results. With regard to Ratner and Leal's (1999) results, which point out that Taiwan, Thailand, and Mexico emerged as markets with trading opportunities, this paper has found that evidence for Mexico has disappeared while there is renewed evidence for other Asian emerging markets. In general, trading rules seem to perform better for both Malaysia and the Philippines. When we investigate a combination of 1559 rules we find that there is some forecasting power, but not much is significant. These results suggest that the usual trading rules that have been used in most of the literature stemming from Brock et al. (1992) do not possess forecasting power for these markets but other rules could have to some extent. This could be due to the fact that these rules are widely known and have been broadly employed, which would minimize the potential gains that one could obtain from them. The general conclusion from the studies of Brock et al. (1992) and Hudson et al. (1996) is that technical trading rules have predictive ability if sufficiently long series of data are considered. In contrast to their findings, our empirical results suggest that for emerging markets, if we take into account both transaction costs and a buy and hold strategy we find some evidence of predictability using technical trading rules but not much is statistically significant. The results in this paper indicate that it is worthwhile to investigate more elaborate rules since simple technical trading rules can yield positive excess returns even when accounting for transaction costs and a buy and hold strategy. An interesting alternative would be to test for the forecasting power of nonlinear rules.