EMPIRICAL TESTING OF ASSET PRICING MODELS: COMPARISON OF TIME-SERIES AND CROSS-SECTION OF APPROACHES

Abstract

The article is devoted to the theoretical aspects of testing pricing models. The main purpose of the article is to compare different methods and techniques for testing the asset pricing models, data collection and analysis, as well as the application of software packages (including the R-studio language). Testing pricing models is based on two main approaches: time-series regression tests and a cross-section regression tests (or rolling cross-section tests). These two approaches have some drawbacks and considered like complementary, so we advice to use simultaneously both approaches in order to get more reliable results. All methods of testing models based on ex-post analysis - historical performance of financial assets. Traditionally, there is used the coefficient of determination and the GRS statistic (that tests whether the estimated intercepts from a multiple regression model are jointly zero) for estimating the quality of the model. If the model is appropriate, the GRS statistics should tend to zero. The effectiveness of portfolio managers, investment funds and strategies are evaluated directly by alpha. If alpha after adjusting for risks (CAPM or 3FF) and transaction costs is not zero, then the pattern could be considered as stock anomaly. Therefore, this work has both theoretical and practical value.