TESTING THE LONG RANGE DEPENDENCE FOR THE CENTRAL EASTERN…测试中央东部的长程依赖性.pdfVIP

TESTING THE LONG RANGE-DEPENDENCE FOR THE CENTRAL EASTERN EUROPEAN AND THE BALKANS STOCK MARKETS Pece Andreea Maria, Ludusan (Corovei) Emilia Anuta, Mutu Simona Ph.D. student, Faculty of Economics and Business Administration/Department of Finance, Babeş-Bolyai University Cluj-Napoca, Romania andreeapece@ Ph.D. student, Faculty of Economics and Business Administration/Department of Finance, Babeş-Bolyai University Cluj-Napoca, Romania emilia_lud@ Ph.D., Teaching Assistant, Faculty of Economics and Business Administration/Department of Finance, Babeş-Bolyai University Cluj-Napoca, Romania simona.mutu@ubbcluj.ro Abstract: In this study we tested the existence of long memory in the return series for major Central Eastern European and Balkans stock markets, using the following statistical methods: Hurst Exponent, GPH method, Andrews and Guggenberger method, Reisen method, Willinger, Taqqu and Teverovsky method and ARFIMA model. The results obtained are mixed. The Hurst Exponent showed the existence of long memory in all indices, except PX. After applying the GPH method, the results showed that BET, ATHEX, SOFIX and CROBEX have a predictable behavior. The ARFIMA model results support the existence of long memory for BUX, SAX and BELEX. The predictable behavior of index returns may suggest that the CEE and Balkans stock markets are not weak form efficient. Keywords: emerging markets, long memory, market efficiency, ARFIMA model JEL Classification: C14, C58, G14 1. Introduction The efficient market hypothesis has major implications for the capital markets, investors behavior and trading strategies. According to Fama (1965) the random walk hypothesis implies that price changes do not have memory, which means that the past cannot be used to predict the future. For this reason, is essential to identify the presence of long memory in index ret