Working Paper: NBER ID: w30976
Authors: Ajay K. Agrawal; Joshua S. Gans; Avi Goldfarb
Abstract: Economic models provide little insight into when the next big idea and its associated productivity dividend will come along. Once a general purpose technology (GPT) is identified, the economist’s toolkit does provide an understanding when firms will adopt a new technology and for what purpose. The focus of the literature has been on commonalities across each type of GPT. This focus is natural, given that the goal of the literature has been to identify generalizable insights across technologies. Broadly, this literature emphasizes heterogeneity in co-invention costs across firms. Each GPT, however, provides a distinct benefit. Steam provided a new power source. The internet facilitated communication. The differences between GPTs are important for understanding adoption patterns. Using the examples of the internet and artificial intelligence, we discuss how both co-invention costs and distinct benefits determine the adoption of technology. For both technologies, we demonstrate that discussions of the impact of a GPT on productivity and growth need to emphasize the benefits as well as the costs. The goal of this paper is therefore to link the literature on co-invention costs with an understanding of the distinct benefits of each GPT.
Keywords: general purpose technologies; co-invention costs; productivity; adoption patterns
JEL Codes: O33; O4
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| co-invention costs (O36) | adoption of GPTs (O33) |
| distinct benefits of GPTs (H43) | adoption of GPTs (O33) |
| distinct benefits of GPTs (H43) | necessity for co-invention (O36) |