Test of the Chen-Roll-Ross macroeconomic factor model: Evidence from Croatian stock market

This paper empirically examines the well-known Chen-Roll-Ross model on the Croatian stock market. Modifications of definitions of the Chen-Roll-Ross model variables showed as necessary because of doubtful availability and quality of input data needed. Namely, some macroeconomic and market variables are not available in the originally defined form or do not exist. In that sense this paper gives some alternative definitions for some model variables. Also, in order to improve statistical analysis, in this paper we have modified Fama-MacBeth technique in the way that second-pass regression was substituted with panel regression analysis. Based on the two-pass regression analysis of returns of 34 Croatian stocks on 4 macroeconomic variables during the seven-and-half-year observation period the following conclusion is made. In contrast to the results of Chen, Roll and Ross (1986) for the U.S. stock market, their model is not successful when describing a risk-return relation of Croatian stocks. Nevertheless, one observed version of the Chen-Roll-Ross model showed certain statistical significance. Namely, two risk factors in that version of the model were statistically significant: default premium, measured as risk premium for the corporate short-term bank loan financing, and term structure premium, measured on short-run basis.