The Information Content of Green Bond Prospectus

Abstract

Since its inception in 2013, green bonds, as a nascent financial instrument, have channeled trillions of financial capital towards environmentally-friendly projects. Yet, the pricing dynamics of this instrument remain a puzzle: green bonds often enjoy a lower cost of debt even though they share similar fundamentals with comparable non-green bonds. To better understand this empirical puzzle, this study delves into bond prospectuses, capturing both environmental disclosures using a machine-learning-based (\textit{word2vec}) green dictionary, along with other linguistic features widely examined in the finance literature. My findings show that more extensive environmental disclosure in prospectuses can significantly reduce the bond yield at issuance. However, these findings appear to be driven by non-green bonds. For green bonds, the green label, but not environmental disclosure in bond prospectuses, is priced in the bond at-issue yield. To explore whether the insignificant pricing effect of green disclosure in prospectuses is driven by a lack of information content or the ignorance of bond investors, I link the linguistic metrics constructed based on bond prospectuses with the environmental performance of bond issuers following the green issuance. The results show that both the environmental disclosure and negative tone contained in the prospectuses can offer valuable insights into the issuers' future environmental performance. Collectively, my findings suggest that bond investors need to pay more attention to the qualitative information revealed in bond prospectuses to identify genuinely eco-friendly corporate investments.