Comparative analyses of mean-variance and mean-semivariance approaches on global and local single factor market model for developed and emerging markets

Purpose - The beta coefficient used for the cost of equity calculation is at the heart of the valuation process. This study conducts comparative analyses of the classical capital asset pricing model (CAPM) and downside CAPM risk parameters to gain further insight into which risk parameter leads to better performing risk measures at explaining stock returns. Design/methodology/approach - The study conducts a comparative analysis of 16 risk measures at explaining the stock returns of 4531 companies of 20 developed and 25 emerging market index for 2000–2018. The analyses are conducted using both the global and local indices and both USD and local currency returns. Calculated risk measures are analyzed in a panel data setup using a univariate model. Results are investigated in country-specific and model-specific subsets. Findings - The results show that (1) downside betas are better than CAPM betas at explaining the stock returns, (2) both risk measure groups perform better for emerging markets, (3) global downside beta model performs better than global beta model, implying the existence of the contagion effect, (4) high significance levels of total risk and unsystematic risk measures further support the shortfall of CAPM betas and (5) higher correlation of markets after negative shocks such as pandemics puts global CAPM based downside beta to a more reliable position. Research limitations/implications - The data are limited to the index securities as beta could be time varying. Practical implications - Results overall provide insight into the cost of equity calculation and emerging market assets valuation. Originality/value - The framework and methodology enable us to compare and contrast CAPM and downside-CAPM risk measures at the firm level, at the global/local level and in terms of the level of market development.