The Impact of ESG on Cost of Capital for Large European Corporations

Abstract

This study seeks to understand the relationships between ESG and the cost of capital of large European firms. By analysing ESG scores from LSEG Data & Analytics (formerly Refinitiv) together with financial data from Compustat we intend to answer whether ESG impacts cost of capital for the firms included in the STOXX Europe 600 Index for the timespan from January 2018 to December 2023. From an initial sample of 600 companies and 3600 observations on the three pillars incorporated in ESG, we run OLS regressions on individual companies under control variables such as Return on Assets, Book-to-Market ratio, Debt-to-Assets, Debt Delta, Capital Expenditure and Operational Expenditure.

Secondly, we divide our sample into quartiles based on ESG scores and investigate whether there are differences between the full sample and each quartile. We find that both cost of debt and cost of equity are vaguely affected by the change in ESG scores. High ESG performers provide results that indicates a marginal decrease in the cost of capital on an overall basis. This is only an indication as our results do not hold any statistical significance.

To ensure robustness, we apply Total Assets as a proxy for SIZE and excludes this variable in order from the lowest to the highest quartile and investigate if the interactions between ESG quartiles and SIZE varies across the different levels. This methodology yields leverage to the assumption that larger companies get results that indicate lower cost of debt, whilst an increase in cost of equity. As for the smaller companies, our results indicate no significant changes in cost of debt, and a significant decrease in cost of equity. I.e. would smaller companies have greater benefit of increasing their ESG score.

In conclusiveness, our main assumption that higher ESG performance yields lower cost of capital is backed by our findings, but with significance to varying degrees we are not able to conclude with an overall statistical significance.