A multivariate test for stock market eYciency: the case of ASE MANOLIS G. KAVUSSANOS* and EVERTON DOCKERY { City University Business School, Barbican Centre, London EC2Y 8HB, UK and Department of Accounting and Finance, Athens University of Economics and Business, 76 Patission Street, Athens 104 34, Greece. { Department of Economics, StaVordshire University, StaVordshire, ST4 2DF, UK Market eciency tests in developing markets display mixed evidence, in contrast to evidence on developed markets where the null hypothesis seems to be supported. Speci®cally, previous tests for market eciency on the index and on samples of stocks traded in the Athens Stock Exchange (ASE) are broadly not supportive of the ecient market hypothesis. This paper introduces multivariate generalizations of the univariate Dickey±Fuller likelihood ratio tests to the class of Seemingly Unre- lated Regressions, to investigate empirically the stock price eciency of ASE. The method takes into account the contemporaneous correlation between stocks in the ASE, and avoids the sample biases which may result by considering only subsets of stocks listed in the exchange. Conclusively, the results con®rm that the ASE is informationally inecient, implying that past stock prices contain some information as to future price movements which investors may act on. I. INTRODUCTION The notion that a stock market is ecient, in the sense that stock prices quickly and accurately re¯ect all available information, has long received considerable support in the ®nance literature. There is cumulative evidence in sup- port of market eciency of major industrialized countries’ stock markets such as the USA, the UK and elsewhere (see e.g. Fama, 1965, 1991; Dryden, 1969). The general consen- sus in these studies is that these markets are ecient in either weak-form or semistrong form. 1;2 The issue of e- ciency of smaller markets has also been undertaken by among others Hong (1978), Cooper (1982), Barnes (1986), D’Ambrosio (1980), Arrif and Finn (1989), Panas (1990), and Butler and Malaikah (1992). Thus, Barnes (1986) ®nds that the ecient market hypothesis could not be supported for the Kuala Lumpur stock market, Butler and Malaikah (1992) unearth evidence of ineciency in the Saudi Arabian stock market, though not in the case of the stock market of Kuwait, and Panas (1990) concludes that market eciency could not be rejected for the Athens Stock Exchange (ASE). In weakly ecient markets investors are unlikely to make excessive pro®ts by exploiting information inherent in previous price changes. This is because, if information on prices is available to rational pro®t seeking market `agents’, arbitrage operations would ensure that security prices are at their equilibrium. A sucient condition for market eciency is that the random walk hypothesis holds, which also suggests that stock price changes are unpredict- able. Evidence of this, however, is questionable in view of the ®ndings on the predictability of stock market returns such as those of Fama and French (1988), Lo and MacKinlay (1988), and Jegadeesh (1990). Among the Applied Financial Economics ISSN 0960±3107 print/ISSN 1466±4305 online # 2001 Taylor & Francis Ltd http://www.tandf.co.uk/journals DOI: 10.1080 /0960310001001300 6 Applied Financial Economics, 2001, 11, 573±579 573 * Corresponding author: E-mail: M.Kavussanos@city.ac.uk 1 Notwithstanding, DeBondt and Thaler (1985, 1987), Lo and MacKinlay (1988), and Conrad and Kaul (1988) ®nd that some of the most developed stock markets in the world are in fact characterized by ineciency. 2 Fama (1970) de®nes three forms of market eciency: (a) Weak-form eciency, under which historical market data are incorporated in current market prices, and are of no use in predicting future prices; (b) Semistrong form, under which all publicly available information, such as earnings and pro®ts/losses announcements, is quickly incorporated in prices; as a result no excess returns may be earned by investors; (c) Strong-form eciency, under which stock prices fully re¯ect all public and private information.