حافظه بلند مدت در نوسانات بازار سهام در حال ظهور: در مورد کشور ترکیه

We use methods based on wavelets and aggregate series, which have gained growing acceptance in the finance literature, to test for long memory in the absolute value, squared, and log squared daily returns of the Istanbul Stock Exchange National 100 Index. Our results show that all three volatility series are characterized by long memory, indicating that shocks to the stock index volatility decay slowly and that distant observations of the series are associated with each other. There are several implications of our study for further research. First, models examining the volatility of the Turkish equity returns should include a long memory component in their parameter set. Second, tests should be conducted to assess whether such models result in an improvement in the volatility forecasts of the Turkish equity returns.

The equity markets of emerging economies have enjoyed rapid and substantial growth in the past couple of decades as investors seeking high return prospects and diversification benefits channeled their funds to assets in these markets. These developments have prompted financial economists to study various aspects of stock return behavior in the emerging markets. Wright (2001) examined stock returns in several emerging economies, and found evidence of long memory in stock returns in seven of these markets. Assaf and Cavalcante (2005) found long memory in the absolute, squared, and log squared returns in the Brazilian stock market. Barkoulas et al. (2000) reported significant and robust evidence of positive long-term persistence in the Greek stock market. Using a fractional autoregressive integrated moving average (FARIMA) model, Panas (2001) also demonstrated the existence of long memory in individual stock returns in the Athens Stock Exchange. Turkey, which is considered an emerging market, has seen significant growth in its equity market since the inception of the Istanbul Stock Exchange (ISE) in 1986. As is the case in most emerging economies, investors face high risks in the Turkish equity market. Further development of the equity market in Turkey and its integration with world capital markets are dependent on the ability of the investors to assess the risks they face. Given these facts, it is essential to examine the volatility behavior of equity returns in Turkey. In a recent paper, Kilic (2004) uses both parametric and nonparametric methods to demonstrate long memory in the volatility of the daily returns of the Istanbul Stock Exchange National 100 Index (ISE National-100). In this study, we use wavelets and aggregate series, which are recently becoming popular in the finance literature, to test for long memory in the absolute value, squared, and log squared daily returns of the ISE National-100. The existence of long-memory in the volatility of stock returns implies that the volatility series demonstrate autocorrelation and future volatility can potentially be predicted using past volatility observations. This has risk management implications in the emerging market of Turkey. We provide a review of the volatility measures and describe our data in the next section. Section 3 includes a discussion of long memory in time series. We present the methods and results of our analysis in Section 4, and discuss the future research implications of our findings in Section 5.

In this study, we utilize methods based on wavelets and aggregate series to test for long memory in the absolute value, squared, and log squared daily returns of the ISE National-100. Our results show that the parameter d for all three volatility series is in the interval (0, 0.5) indicating that the series are characterized by long memory. The long memory property of the absolute daily returns, squared daily returns, and log squared daily returns of the ISE National-100 relates to the fact that shocks to the stock index volatility decay slowly and that distant observations of the series are associated with each other. Our finding of long memory in the volatility of the Turkish stock market returns is consistent with the results of previous studies covering the emerging market of Brazil ( Assaf and Cavalcante, 2005), the small market of Finland ( Tolvi, 2003), and the developed market of the United States ( Ding et al., 1993). There are several implications of our study for further research. First, our findings confirm the results of Kilic (2004) that models examining the volatility of the Turkish equity returns will benefit from the inclusion of a long memory component in their parameter set. Second, tests should be conducted to assess whether such models result in an improvement in the volatility forecasts of the Turkish equity returns.