Can international environmental agreements induce innovation on green technologies? It is possible that international negotiations succeed only once technological solutions are available. In this case, agreements would help diffuse such technologies rather than fostering their development. I provide the first quantitative evidence that the Montreal Protocol, and its following amendments to protect the ozone layer, triggered a large increase in research and innovation on alternatives to ozone-depleting molecules. To do this, I use the full text of patents and scientific articles to construct new panel data of the yearly number of published documents about these molecules. I implement a difference-in-differences strategy (DiD) and a synthetic control method (SCM) using hazardous air pollutants as control units. To compare molecules’ chemical and industrial characteristics, I construct descriptive variables by applying machine learning techniques to the documents’ text. The SCM estimates that the post-Montreal regime caused a 144% increase in patents and a 189% increase in articles mentioning substitutes to ozone-depleting substances; the DiD yields comparable estimates. These results challenge the view that agreements foster technological diffusion without affecting much of the dynamics of innovation. Agreements can thus encourage the development of green technologies, which importantly suggests they should be negotiated as early as possible if we hope to solve global environmental problems.

New technologies are important to improve the well-being of poor communities, but many barriers prevent adoption from reaching its socially optimal level. In order to better understand the extent of informational barriers for adoption of a simple household product, we designed a randomized controlled trial on willingness to pay (WTP) for solar lanterns in India. In a non-electrified region of the state of Uttar Pradesh, we give high quality solar lanterns to randomly selected ‘seed’ households. Each seed household provides the names of three friends whom we randomly assign to one of the three following groups: control, passive learning and incentivized communication. We elicit WTP from the control group exactly when the seed receives the lantern. However for the friends in the passive learning and incentivized communication groups, we elicit WTP thirty days after the seed receives the solar lantern. In addition, the seed is given a small reward for arranging a tea meeting with only the friend assigned to the incentivized communication group. During the meeting, the seed showcases the usage of the solar lantern and shares his or her user experience. We show that both passive learning and incentivized communication have large positive effects on WTP. Passive learning increases WTP by 90% and incentivized communication by 145% relative to the control group. Our results have important implications for understanding the magnitudes of learning and improved information flow among peers on WTP for new technologies.

Transitioning to a low-carbon economy requires radical technological change in a range of industries. What institutional mechanisms can promote such change? Economists mostly promote R&D subsidies and carbon taxes as the key instruments. However, the effectiveness of such instruments is premised on a world of atomized firms, capable of independently optimizing their R&D investments. Instead, we argue that radical innovations (such as electric cars or zero-energy buildings) require complementary innovations in interdependent components. And possibly the production of such components is distributed across multiple firms within supply networks. We focus specifically on cases where producers share suppliers, and we show that, when technological change requires investment by both the supplier and the producer, coordination within an industry is needed and can be difficult to obtain. Thus, we investigate how interdependencies in a network of final producers and suppliers can render technological change within the supply chain prone to coordination problems. This result helps unify several findings from case studies in the organization of innovation. The model further allows us to examine potential limits of marginal pricing policies for stimulating ecological innovations and how policies aiming at reducing miscoordination might complement marginal pricing policies.