Chasing the ESG Factor

In the time-series (ordinal ESG) or the cross-section (cardinal ESG)? We show analytically that, when proper adjustment to guarantee identical ESG ratings is implemented, the return spread of the factors produced by the two methods is merely noise. We provide a protocol to construct a cross-sectional ESG factor with a targeted ESG rating without screening stocks, hence without harming ex ante diversification (Sharpe ratio). The cross-sectional ESG factor neutralizes the exposure to other firm characteristics. Using ratings from several ESG data vendors, we document strong variations in the ESG factor's alpha in the time series and across data vendors. The alpha filtered from realized returns is negatively related to the level of an ESG sentiment variable based on media attention, while it is positively related to unexpected variations of the sentiment.