Multiple Regressions

This chapter provides cross-sectional regressions with multiple explanatory variables. In the previous sections, we focused on simple regressions only with one firm-level predictor as the explanatory variable. Obviously, simple regressions have their limitations due to the fact that there are no controlled variables. Our motivation in running the simple cross-sectional regressions is to show their explanatory power with no controlled variable and let the reader compare them for the included emerging markets when nothing is controlled. The main problem with simple regressions is the difficulty of drawing conclusions about return under the unrealistic assumption that nothing other than the single explanatory variable would affect it. We are aware however that when certain variables are explicitly controlled, explanatory powers of independent variables would change. Expectedly, this would not be a uniform change for different markets and will follow different paths for different countries. This is the other motivation of starting with the simple regression and providing the reader with a comparison benchmark. Once we know the explanatory power of a firm-level predictor in simple regression, we obtain more comparative information when certain control variables are included later.