Price of greenness: Classifications and green bond premiums

Financial markets allow a premium to green bond issuers (a.k.a. greenium), which incentivises the transition to green projects. This premium also absorbs costs associated with green bond certification, necessary to prevent greenwashing, and aimed at reducing investors’ uncertainty. Several taxonomies have been created to classify bonds to that end. A question is to what extend are such classifications an effective means as traditional credit ratings serve a similar, more generic purpose? And what is the possible effect on market efficiency of these classifications? An observed variance in green bond premiums across different bond classes would also suggest that charges for the certification of green bonds should vary. Difference in greeniums reveal differences in the way the markets assess distinctive classes of green bonds. Especially, when bond classifications change, or taxonomies are ambiguous this could lead to adverse selection or invite greenwashing. Here we compare 858 pairs of matched green and non-green bonds and use a mixed effects model to estimate how bonds’ greenium differ over credit ratings and ‘Use of Proceeds’ categories. Results show that lower-rated bonds reach higher levels of green premiums, controlling for categorical random effects. Similar effects are found for ‘Use of Proceeds’ classes. However, compared to either of the two other classifications a cross-classification model provides significant improvement, demonstrating the added value of green bond taxonomies for investors. This solves a paradox in the literature that found that high score ESG bonds, but also low credit rated bonds receive a higher premium. The other implication is that market inefficiencies may occur due to segmentation since it is common practice for certification costs to be flat and independent from greenium levels. Counterintuitively, creating a green taxonomy could lead to more uncertainty and adverse selection for “true” green project financing, which would delay the green transition and the desired shift to a low-carbon economy. An implied remedy is to implement differentiated verification charges for green bonds across bond credit ratings. • ‘Use of Proceeds’ categories differentiate greenium curves. By grouping green bonds according to the intended use of proceeds (e.g., renewable energy, energy efficiency, clean transport, green buildings), we find significant variation in both the level and curvature of the greenium across these taxonomic classes, showing that risk, thus the premium differs by project type. • Combining credit-rating and ‘Use of Proceeds’ classifications further improves differentiation. A mixed-effects model that nests greenium curves within both credit-rating groups and Use-of-Proceeds categories delivers the best fit, demonstrating that layered categorization captures heterogeneity in premiums more precisely, suggesting that certification costs should likewise be tailored to these combined classes. • Flat verification fees risk market distortions and adverse selection. Because current accreditation fees (e.g. CBI’s fixed 0.00001 fraction of issuance) do not vary with a bond’s greenium or underlying characteristics, they treat all bonds as one class potentially aggravating distortions in green premiums and inviting adverse selection in an “exogenously costly disclosure” regime. • Certification charges should be tailored to bond class to improve efficiency. To align incentives, verifiers ought to redesign audit and certification fees to match predicted greenium levels based on a bond’s anticipated credit rating and Use-of-Proceeds category. This helps in curbing greenwashing, supporting true green issuers, and enhancing overall market efficiency.