Apologies for the slow rate of posting throughout the fall. Though I have a big backlog of posts coming, the reason for the delay has been that I’ve spent the last few months preparing for the job market. As the academic job market is essentially a matching problem, below I describe my basic research program; if you happen to be a reader of the blog on the “demand side” from an economics department, business school or policy school, and my research looks like it might be a good fit, I would be eternally grateful for a quick email (k-bryan AT kellogg.northwestern.edu).
Most broadly, I study the process of innovation. Innovation (inclusive of the diffusion of technology) is fundamental to growth, and growth is fundamental to improving human welfare. What, then, could be a more exciting topic to study? Methodologically, I tend to find theory the most useful way to answer the questions I want to answer. Because the main benefit of theory is generalizability (to counterfactuals, welfare estimates, and the like), I try to ensure that my theory is well grounded both by using detailed historical examples within the papers, and by drawing heavily on existing empirical work by economists, historians and sociologists. Beyond innovation, I have a side interest in pure theory and in the history of thought; both these areas provide the “tools” that an applied theorist uses.
Recently, I’ve worked primarily on two questions: why do firms work on the type of research they do, and how does government policy affect the diffusion of invention? On the first question, I have three papers.
My coauthor Jorge Lemus and I have developed an analytically tractable model of direction choice in invention, where there are many inventions available at any time, and successful invention by some firm affects which research targets become available next. We shut down all sources of inefficient firm behavior in the existing literature, and still find three sources of inefficiency generated by direction choice alone. We fully characterize how this inefficiency operates on a number of “invention graphs”. This is actually a pretty cool model which is really easy to use if you are familiar with the patent race literature.
In my job market paper, I use the invention graph model to study how government R&D works when firms may have distorted directional incentives. The principal result is a bit sobering: many policies like patents and R&D tax credits that are effective at increasing the rate of invention on socially valuable projects will, in general, exacerbate distortions on the direction of invention. Essentially, firms may distort to projects that are easy even though those projects are not terribly profitable. With this type of distortion, any policy that increases the payoff to R&D generally will increase the payoff of the inefficient research target by a larger percentage than the payoff of the efficient research target. I show how these policy distortions may have operated in the early nuclear reactor industry, where historians have long worried that the eventual dominant technology, light water reactors, was ex-ante inefficient.
My third paper on directional inefficiency is more purely historical. How can a country invent a pioneer technology but wind up having no important firms in the commercial industry building on that technology? I suggest that commercial products are made up of a set of inventions. A nation may have within its borders everything necessary for a technological breakthrough, but lack comparative advantage in the remaining steps between that breakthrough and a commercial product; roughly, Teece’s complementary assets operates at a national and not only a firm level. Worse, patent policy can give pioneer inventors incentives to limit the diffusion of knowledge necessary for the eventual commercial product. I show how these ideas operate in the airplane industry, where ten years after the Wright Brothers’ first flight, there was essentially no domestic production of US-designed planes.
On diffusion more purely, I have three papers in progress. The first, which we are preparing for a medical journal and hence cannot put online, suggests that open access to medical research makes that research much more likely to be used in an eventual invention. My coauthor Yasin Ozcan and I generate this result from a dataset that merges every research article in 46 top medical journals since 2005 with every US patent application since that date; if you’ve ever worked with the raw patent application files, for which there is no standard citation practice, you will understand the data challenge here! We have a second paper in the works taking this merged dataset back to the 1970s, with the goal of providing a better descriptive understanding of the trends in lab-to-market diffusion. My third paper on diffusion is more theoretical. I consider processes that diffuse simultaneously across multiple networks (or a “multigraph”): inventions may diffuse via trade routes or pure geography, recessions may spread across geography or the input-output chain, diseases may spread via sexual or non-sexual contact, etc. I provide an axiomatic measure of the “importance” of each network even when only a single instance of the diffusion can be seen, and show how this measure can answer counterfactual questions like “what if the diffusion began in a different place?”
My CV, a teaching statement, and an extended research statement can be found on my academic website. I am quite excited about the research program above; I would love to chat if you are at a convivial department filled with bright students and curious academics that is hiring this year.