Asymmetric Multiply Error Model And Its Applications

Since Multiply Error Model (MEM) has been proposed, there are a number of studies about MEM all over the world. In china, many academics do major empirical researches about Chinese financial markets with the MEM model, but less on asymmetrical MEM Model. In fact, the non-negative time series in the financial markets mostly are asymmetrical, Chinese financial market is no exception, such as fluctuating asymmetry, the asymmetry of financial duration and so on. Based on the basic theory of MEM model, we explore asymmetric MEM model with drawing on ideas and structure of asymmetric ACD model, which pulls asymmetric structure in general MEM and builds asymmetric model, then we use Monte Carlo simulation to compare the Characterization capability of these two models-classical MEM and asymmetric MEM. Finally, we do empirical research with asymmetric MEM about Chinese financial market.This article does analysis as follows:(1) Based on the classical MEM model, drawing from the thoughts of constructing asymmetric GARCH model and asymmetric ACD model, this paper explores the asymmetric MEM model structure and builds an asymmetric MEM model, then discusses the selection of distribution of the error term and the estimation of asymmetric MEM model.(2) To compare the classical model with asymmetric MEM model, which depicts the financial time series with asymmetrical, this paper used Monte Carlo simulation method to randomly generate the time series with asymmetrical as required, then used MLE to estimate the classical MEM and the asymmetric MEM model, and compared the two model by the predictive power index. It has showed that the asymmetric MEM model is better than the classical MEM model on characterizing the financial time series with asymmetrical.(3) The high-frequency financial data including price duration, the transaction duration, volume, high, low and so on, have some typical features, such as peak and thick tail, calendar effects, autocorrelation and long memory, asymmetric and other features, the paper selected volume, high, low of CMB stock as the research objects, did the empirical analysis whether the time series of CMB stock have these typical characteristics or not. It has showed that the volume, high, low of CMB stock have these typical characteristics.(4) This paper has analysed the five minutes high-frequency data of CMB stock from January22014to February242014, and selected the volume, high, low as the research objects, first, eliminated days effects of volume, and got the price indicator variable and trade intensity, and then build an asymmetric MEM model, which can describe the asymmetric impacts of price changes on trade intensity. And the result told us that both positive and negative changes will make transaction increase, which means that price volatility will strength transaction. But the direction of price changes has different degree of influence on trading intensity, which indicates that the transaction has significant asymmetrical.Through the study of asymmetric MEM model, one the one hand, it is the expansion of asymmetric GARCH model and asymmetric ACD model, and provides us a more powerful tool for the study of non-negative time series, and enriches the MEM model. On the other hand, it does empirical research on trade intensity with asymmetrical in Chinese financial market, in china, they do major study that market factors has the long-term or short-term impacts on trade intensity, therefore, this paper does empirical study, which make investors and us understand the impact of trading systems and market structure, and has significant practical value on improving Chinese securities market regulation.This paper is supported by the2011National Natural Foundation of China (71101118).