Markus Leippold, and Tingyu Yu
Review of Finance, Volume 30, Issue 2, March 2026, Pages 423–458, https://doi.org/10.1093/rof/rfaf058
The need for green innovation has never been more urgent due to its central role in the net-zero transition. While patents are the most widely used measure of innovation in academic research, they are limited in that they do not fully capture the breadth of corporate innovation. Many innovations are not patented or patentable. Moreover, patents reflect the first stage of innovation—the invention—whereas innovation extends to the commercialization and implementation of related technologies. These subsequent processes involve costly and skill-intensive activities, and economic value emerges only when innovations are adopted. In the context of green innovation, this broader definition is particularly relevant as inventors and adopters often diverge and adopters are typically carbon emitters with large responsibilities to cut emissions.
In this paper, we introduce a novel text-based measure of green innovation that captures both the invention and adoption dimensions. By applying ClimateBERT and GPT language models to earnings call transcripts, the measure reflects not only the creation of new green technologies but also their integration into business operations. Compared to patent-based metrics, it identifies commercially impactful green technologies (given that several patents never reach the market) and innovative activities of non-patenting firms (through non-patentable inventions and adoption). Geographically, green invention is concentrated in a few states with leading universities and research hubs, whereas adoption is widely distributed across the country. By sector, invention is led by Electrical Equipment and Automobiles firms, while carbon-intensive industries such as Utilities and Metal Mining commonly emphasize adoption.
We empirically examine the impact of green innovation on asset pricing. A portfolio that is long in firms with high green innovation engagement and short in those with low engagement generates an average annual return of -5.4%. This negative effect likely stem from green innovating firms’ ability to hedge against transition risk as they demonstrate improved environmental performance and superior resilience to climate regulatory shocks. Such within-industry effects, which hold in carbon-intensive sectors, suggest that investors offer lower costs of capital to “brown” firms actively pursuing green innovation. Notably, both invention and adoption are priced in the stock market, with adoption effects extending to non-patenting firms at a comparable magnitude. This underscores the complementarity of our measure and highlights the need to incorporate non-patentable invention and adoption to fully understand corporate technological engagement in addressing the climate crisis.