Patents have many problems, deadweight loss from temporary monopoly being only one of them. There are also patent thickets where so many patents bind on any downstream investment, particularly in high tech, that patenting is only a defensive move – apparently these were less politically correctly called “Mexican standoffs” in the literature in the 1970s. There is the problem of inducing invention toward favored fields since patents treat everyone equally, and Moser’s lovely 2005 AER shows that in the era when patents were limited to fewer industries, they definitely affected the direction of innovation. These problems have led to proposals for prize systems to partially replace patents, such as the proposals of Michael Kremer, or the X Prize contests. But is there any evidence that prizes work?
Brunt, Lerner and Nicholas look at a system of prizes given by a Royal Society in the UK for agricultural improvements between 1839 and 1939. A year before each annual agricultural fair, a list of targeted invention areas was published, along with monetary prizes and “medals” which could be awarded; the McCormick Reaper won a Gold Medal, for example. Inventors were still allowed to patent. The monetary prizes were generally less than the sale price of even a single unit of a new invention, but the medals and monetary prize lists were publicized fairly widely and were thought at the time to be a valuable marketing tool. For a couple decades in the mid-1800s, the general targeted area rotated in a three year cycle and there were no major changes to the relevant patent laws. Patents in the UK were quite expensive, and required an expensive renewal after three years, so many studies have used “renewed patents” as a cutoff for signal for economically important invention; note that even this is not perfect, though, due to many inventions like the spinning jenny not being patented at all.
The monetary prizes were small, so unsurprisingly in the most convincing econometric specification, they induce little invention in the patent record of the product area targeted in any given year’s contest. But medals were valuable, and an announced possible gold medal for improvements to, say, harvesting technology caused a roughly 30 percent increase in the number of renewed patents that year (as well as, of course, a substantial increase in the number of annual fair entrants in that area). A test also checks against switching – that is, are the increases just due to displacement from the years when two previous years that no prize was offered in the targeted area? Checking patents by non-entrants only finds the positive effect of the prize announcement remains, and in an economically meaningful way. My takeaway from this paper is that it gives little evidence about the elasticity of monetary prizes to directed invention, but that it does show that high profile contests are capable many potential inventors to shift effort in a desired direction. Certainly this could be meaningful for, say, NIH or NSF driven contests and the like.
One final quibble, though. I have a pet peeve about famous papers being cited inappropriately. In the present paper, Lazear and Rosen’s 1981 tournaments result is mentioned as being relevant because it says prizes like those offered in these agricultural fairs induce higher effort among potential inventors. I don’t buy this. For one, Lazear and Rosen’s result relies on firms in a competitive labor market hiring those workers, and the existence of other firms who can make offers to the potential inventors is critical for the result. Second, these contests aren’t even tournaments in the sense of Lazear and Rosen: the number of prizes (or even whether a medal was awarded at all) varied depending on the quality of the entrants. Lazear and Rosen specifically deal with the case where the best, in the ordinal sense, worker always wins the prize.
http://www.prizecapital.net/Prize_Capital/Prizes_files/11-118_1.pdf (2011 HBS Working Paper)
“The effect of a link on the Economist’s website on the quantity and quality of comments: Evidence for a natural experiment with n=1”