Same Firms, Different Verdicts: ESG Rating Choice and the Measurement of Greenwashing

Battle Count: 70

The paper is moderately relevant to quantitative trading. It directly informs ESG-integrated portfolio construction by showing that index membership predicts wider disclosure-performance gaps under CDP lens, implying adverse selection in index-linked ESG screens. The finding that operational ratios (renewable share, capex intensity) are more reliable screening criteria than disclosure-based scores has direct implications for factor construction. However, the paper does not propose trading strategies, does not use time-series or ML models, and its primary contribution is to academic understanding of ESG measurement rather than to alpha generation. The proxy-sensitivity finding is relevant for any quant strategy using ESG scores as signals.

The core estimation is standard OLS with HC3 robust standard errors, which is straightforward. However, the six-stage model selection process (36 candidates, 421 specifications, five variable pools, stepwise forward search under AICc) adds moderate complexity. The main implementation challenge is data collection: triangulating data from CDP, LSEG, company filings, SBTi, TCFD archives, and index constituent lists for 200 firms across four sectors. The within-sector Z-standardisation and DPG construction are simple but require careful handling of ordinal-to-cardinal conversion. No ML or deep learning components are involved.

The paper provides detailed methodology including the six-stage model selection process, variable definitions, data source hierarchy, and full regression diagnostics. However, no code or data repository is explicitly linked. The sample construction (STOXX Europe 600, four sectors, top 50 by market cap) is described but the exact firm list and data collection process would require significant effort to replicate. The CDP scores and LSEG scores are proprietary databases.