Cost of Capital for ESG and Non-ESG Stocks: Regression- versus Theory-based Approaches

Cost of capital is the most important aspect for investors when deciding to invest in ESG stocks. Against this background there is a puzzle in the ESG literature. When formulating hypotheses or in-terpreting empirical findings, empirical ESG cost of capital papers use economic interpretations de-veloped in theoretical ESG pricing papers. Yet, to derive cost of capital, these empirical ESG papers do not use the valuation formulas of theoretical ESG pricing paper. Instead, empirical ESG papers rely on multi-factor regressions to estimate cost of capital. This might be due to the fact that the theoretical ESG pricing formulas are not presented in an empirically implementable form—they contain preference-dependent parameters.<br><br>Our paper brings theory-based ESG pricing formulas into a form that consists of solely observable components and shows that the cost of capital of ESG stocks is a linear function of the risk premium of the ESG sub-market portfolio whereas the cost of capital of non-ESG stocks is a linear combina-tion of the risk premia of the market portfolio and the ESG sub-market portfolio. This explanatory factor “risk premium of the ESG sub-market portfolio” is a factor derived from an asset pricing model and not an empirical or “guessed” factor and, hence, overcomes the factor zoo problem raised in Fama/French (2018) and Harvey/Liu (2019). Moreover, we demonstrate that the cost of capital differences between regression- and theory-based cost of capital are both statistically and economically significant, where neither the sign nor the size of cost of capital differences can be forecasted with the help of different ESG rating methodologies or stock characteristics.<br><br>Hence, theory-based cost of capital should always be employed even though they are more tedious to implement empirically.