Sharfman, Mark. "Corporate Social Performance and the Cost of Capital: The Reward for Increased Legitimacy." Working Paper, University of Oklahoma. June 23, 2006.
Studies the impact of corporate social performance (CSP) on firms' riskiness and cost of capital. Given past empirical work (see Orlitzky 2004) suggesting a positive relationship between CSP and corporate financial performance (CFP), the study explores how capital markets reacted to this information over the 1999-2002 time period.
The empirical component of the study is a series of regressions of the S&P 500 companies in which KLD social research data (supplemented by Fortune magazine reputational scores) is the independent variable and the firm's cost of capital and a variety of other financial factors (derived mainly from Compustat data) are the dependent variables. Control variables include industry, firm size (market capitalization), the firm's level of risk (beta), and financial leverage (long term debt / market capitalization).
Finds that "higher levels of CSP are associated with lower betas, lower costs of equity, and a lower overall cost of capital. Negative CSP is associated with increases in beta so it will likewise increase cost of equity capital."
In a highly-detailed sub-analysis, the study finds that the cost of equity capital is reduced by strong CSP, but not the cost of debt capital. The study also breaks CSP down into two categories (both derived from an exploratory factor analysis of the KLD data), "Business CSP" and "Social CSP". Companies with high Business CSP scores are highly regarded in the Fortune data, and have positive ratings from KLD in employee relations, environment, and products. Companies with high Social CSP scores are highly rated by KLD on community and human rights factors. In most of the analysis Social CSP was found to have the predicted relationships with the firm's riskiness and cost of capital. The Business CSP measures were mixed however, with many coefficients having the opposite of the predicted sign.
The innovative handling of KLD data and careful management of challenging statistical issues make this study an important point of reference for anyone who plans to do empirical work in the CSR/SRI field using these datasets.