Decisions by institutional investors are increasingly shaped by environmental, social, and governance (ESG) ratings. But recently, investors are paying closer attention to the divergence of ratings issued by different ESG rating providers for the same firms. The authors analyze the level of this disagreement to gain a better understanding of the magnitude and whether such disagreement is reflected in a firm’s accounts and financial metrics. They also investigate the real-world impact of these rating differences on firms and investors by examining the relationship between ESG rating disagreement and stock returns.

The authors collect data on S&P 500 companies for the period 2010 to 2017. These data include such financial performance metrics as stock returns, stock volatility, beta, market cap, momentum, and book value. (Stocks that had negative book value were excluded.)

In addition, the authors gather ESG ratings on S&P 500 stocks from seven different ESG raters: Refinitiv, Sustainalytics, Inrate, Bloomberg, FTSE, MSCI KLD, and MSCI IVA. Each of the raters provides an overall ESG score as well as a rating for each of the ESG pillars: environmental, social, and governance. And each rating company provides ratings for an average of more than 400 equities in the S&P 500 Index, indicating that the sample is representative of the index.

The authors examine the correlations between ESG ratings from different providers. They use panel regressions to find the relationship between ESG rating disagreement and various variables, including industry/sector, as well as those related to the balance sheet, industry, investor transparency, valuation, and price. Panel regression is also used to determine the relationship between stock returns and differences in ratings across each of the ESG pillars.

The authors conduct portfolio sorts based on ESG rating disagreement to determine whether rating disagreement provides a profitable investment signal. In this analysis, the authors control for different industries to rule out industry effects in the results and established return factors.

The pairwise correlation between rating providers for overall ESG ratings is 0.45, far lower than the 0.99+ correlation among credit raters. The governance pillar has the lowest pairwise correlation between ESG rating providers (0.16), while the environmental pillar has the highest (0.46). That could be explained by the greater transparency and quantification of environmental matters than found in governance or social issues.

The authors find that three variables are key to explaining rating disagreement. More profitable firms tend to have lower ESG rating disagreement, and firms without credit ratings and those that are larger/more complex have higher ESG rating disagreement. In addition, those firms with more tangible assets tend to have lower rating disagreement, and equities with higher institutional ownership and those with high book-to-market value ratios have higher disagreement.

The authors find that rating disagreement is positively correlated with an equity’s returns. In other words, with high (low) rating disagreement, there will be better (worse) returns than with low (high) disagreement. They also find that these returns are primarily driven by disagreement in the environmental pillar.

A long–short portfolio based on the overall ESG rating generates raw monthly returns of 21 bps. The long-only portfolio of high-disagreement stocks produces significant raw monthly returns of 134 bps.

Responsible investment strategies should incorporate the lowest level of ESG rating disagreement within an industry.

Separately, CFOs should consider rating disagreement when making capital budgeting decisions because disagreement raises the cost of capital and hence the threshold for making investment decisions.