Reconstructing and Benchmarking ESG Scores Using a Two-Stage Entropy-Weighted Grey Relational Analysis (ESG-MCDM) Framework: Evidence from the Construction Materials Sector, 2019–2024

This study examines how weighting and aggregation affect environmental, social, and governance (ESG) assessments within a shared provider architecture. A balanced panel of 50 Construction Materials firms over 2019–2024 is reconstructed from ten London Stock Exchange Group (LSEG) category scores. Stage I forms three pillars using annual entropy weights. Stage II re-estimates entropy weights and applies Grey Relational Analysis (GRA), producing a 100-scaled relational index. Composite Pearson correlations with LSEG range from 0.908 to 0.944 and Spearman correlations from 0.938 to 0.966. However, mean absolute errors of 7.48–8.74 points and concordance correlation coefficients of 0.779–0.847 indicate limited numerical agreement. Under the specified common min–max normalization and displayed score scales, the entropy-weighted Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) yields smaller numerical discrepancies in every year. GRA provides stronger rank association under these conditions. Equal Stage-I category weighting also improves benchmark correspondence. Rankings are relatively insensitive to individual firm omission and alternative temporal weights but more sensitive to Environmental Innovation omission. Varying the GRA distinguishing coefficient largely preserves ranks while mechanically shifting score levels. A 2000-replicate firm-trajectory bootstrap assesses uncertainty. The framework provides a transparent diagnostic of reconstruction consistency rather than independent validation of corporate sustainability performance.