Innovation, investment and patents
One of my long-term themes on this blog is knowledge work - how we measure the productivity or capital value of knowledge, and how it is converted into valuable outputs. And a recurring recent point has been investment, what it is and how we encourage it.
Combining these two is a posting from VoxEU from a few months ago: "Efficiency in the 'market for innovation'" by Alberto Galasso and Mark Schankerman. They claim that
"The 'market for innovation' - the licensing and sale of patents - is, for example, one of the principal incentives for firms to invest in R&D."
My instinct is that this is not true at all. Lots of companies that I work with invest in R&D, and virtually none have any patents. Admittedly I do not work with any large biotech or electronics manufacturers, so my anecdotal evidence is not necessarily representative. But I do believe that a lot of innovation goes on in the marketplace which is not captured in patent statistics.
Economists often deal with information which is hard to measure and even harder to run controlled experiments with. In research on innovation, patents are a useful proxy for both innovation or R&D, because they are hard to measure directly (cf. Nick Rowe's point 2 in my previous posting). But are they a good enough proxy? I think not.
Here are some important factors which argue against patents being a good measure of innovation:
- They are biased towards larger firms. The fixed overheads of registering and defending patents means that a firm needs to be able to amortise their innovation over a minimum level of revenue before it becomes worth patenting. However, lots of innovation does go on in small firms - particularly in the high-tech sector, but also in many others.
- They are biased towards formal, laboratory-based research work. Of course this work is important, but not the only - and probably not the major - source of innovation in the economy. Market-based innovations - a new way of packaging a product, a new way of reaching potential buyers, or a new pricing structure - create economic value just as developing a new type of microchip. Of course most individual market-based innovations have a smaller effect than a newly invented pharmaceutical, but there are many more of them.
- They are biased towards work which is known in advance to be innovative and patentable. Many of the new services that arise in the economy (think of Twitter, Facebook, Walmart, Pret a Manger) arise through trial and error, and the innovative components may only be visible in retrospect. Most services have been through many iterations and discarded 90% of the new ideas they tried, keeping only those that worked in market testing. It's not only impractical, but sometimes legally impossible, to patent a new concept developed in this way.
- They are biased towards long-lasting (and therefore perhaps slow-moving) innovations. Because patents take a couple of years to acquire and last for 17 or 20 years, they are best suited towards big inventions with large investment and high, slow-burn returns. However, a vast number of useful ideas only take a short time to develop and may only bring a return for a few months until they are obsoleted or copied by competitors. Many of these are certainly worth doing, and certainly not worth patenting.
- Patents act as a disincentive to some kinds of investment, as well as an incentive to others. Software patents are the most-cited example: they discourage the development of some new kinds of software because the developers may not be able to afford to licence a known patent, or may fear the risk of infringing an unknown patent. On the other hand they do offer an increased return to the developer who gets the patent. Which of these effects dominates? It is not known, but there are definitely two sides to the question.
- Can we measure, or estimate, the amount of this kind of innovation which takes place, compared to the amount of formal, patentable invention?
- Can we therefore estimate the amount that is invested in these innovations, and the return made on it?
- Does this shed any light on the total amount of investment in the economy?
- Is there an alternative, more economically efficient, system to use instead of patents?