Emerging Stock market Efficiency: Nonlinearity and Episodic Dependences Evidence from Iran stock market

Motivated by the institutional features of emerging stock markets, the present paper examines the weak form efficiency of Iran stock market as an emerging market based on stock returns data generating process. Through a synthesis of the nonlinear structure in stock price series and the stylized facts of stock return series, we adopted a methodology to detect both linear and nonlinear dependencies in the stock return series. The results show that the stock returns data generating process is fitted by the ARMA- GARCH_ M model .Therefore, the future stock returns based on the past pattern of stock price changes would be predictable. Another interesting finding is that the common assumptions of constant variance and Gaussian stock returns are invalid for this emerging stock market. Doubtless the stock market efficiency has been the dominating concept in the theoretical and empirical literature of finance. Due to scarcity of the financial resources, an efficient stock market is critically important in mobilizing national saving and financing of new investment projects. Furthermore, stock market efficiency plays a crucial role in decision making of companies concerning diversification of their sources of investment capital and spread investment risk. In an efficient stock market, stock prices reflect all available information which is relevant for the evaluation of a company's future performance, and therefore the market share price is equal to its inherent value. Any new information, which is expected to change a company's future profitability, is immediately reflected in share prices. In an informationally efficient market, stock is appropriately priced at equilibrium level and there is no distortion in the pricing of capital and risk. Stock market efficiency depending on the type of relevant information appears at three levels. Firstly, the weak form efficiency which claims that current stock prices reflect all relevant and available (historical) information. Secondly, the semi-strong form of efficiency which indicates that present stock prices reflect historical and all relevant public information. Finally, if the stock prices reflect the public and private (insider) information, the stock market would be efficient in strong form. In a stock market which is efficient in weak form, the subject of this paper, stock prices changes only in response to new information which by definition must be unpredictable. Under these circumstances, the prices of stock behave randomly, or without any identifiable pattern. Therefore, in the weak form of efficient stock market the prediction of future pattern of the stock price movements based on past prices is impossible. In view of the well known properties of financial time series such as stylised fact, volatility clustering and no normality, absence of linear dependence does not necessarily mean unpredictability of the stock price movements. In this view, a methodology in testing stock market efficiency should apply which is capable to detect both linear and nonlinear dependences. In this study, using data from the Iran stock market, we seek to examine the concept of weak form efficiency in the light of specific institutional features of the market under investigation. Additionally, the stylised facts of stock return time series are taken into account in the model-building process. Specifically, we adopt a testing methodology in accordance with underlying dynamic in data which enables us to identify the possible existence of non-linear behaviour and episodic dependences in stock prices process. The remainder of the paper is divided into four main areas. The theoretical background is discussed in section 2.Section 3 outlines and explains research methodology which is used. A description of the data employed in the analysis and empirical results is presented in section 3. The paper ends with summarising the main conclusions. Theoretical issues The conventional stock market efficiency tests such as variance ratio test, serial correlation test, runs test and unit root test assume linearity in the stock return (stock price changes) generation process and are not able to capture possible non-linear behaviour in time series. The assumption of linearity implies that the means and