Performance of Semi-parametric Asset Pricing Model in Tehran Stock Exchange

 In asset pricing models, it was traditionally assumed that there is a linear relationship between return and explanatory variables. Therefore, estimating the coefficient in a nonlinear setting would be inconsistent and bias-oriented. In this study, the predictive power of the nonlinear and linear Fama-French Five Factor Model was estimated in the period from March 2010 to March 2020.

The semi-parametric method was used to estimate the nonlinear expected return in FF Five Factor Model. The expected return was also calculated based on the linear FF Five Factor Model.

The estimated return was compared with the realized returns. Then the mean absolute percentage error was used to measure the predictive power of research models. The results show that mean of the mean absolute percentage error in the semi-parametric model is lower than the linear model.

Despite the lower error of the semi-parametric FF Five-Factor model compared to the linear model, no significant difference was observed between the predictive power of these two models. Therefore, the estimating methods will not have a significant impact on the predictive power of the Five Factor Model.