Eric von Hippel, an economist at MIT Sloan, is not particularly well known to people not working on innovation, but he should be. His background is quite interesting: before doing a PhD back in the 1970s, von Hippel was a practicing research scientist in an R&D Lab, and he has the patents to prove it. I can only imagine that this background led him to see the standard wisdom on invention – firms spend money on R&D which probabalistically leads to inventions which are then patented – as ridiculous. We know from many other studies that, outside of a handful of industries, patents are not particularly important to R&D, and that purposeful R&D is not often the source of new products, but through his interviews and detailed studies of specific industries, von Hippel throws out even more of the old wisdom.
First, in many industries, manufacturers don’t “invent” what they sell. For instance, in the scientific instruments industry, about 80% of inventions were first developed as one-offs by users. Essentially, scientists were mocking up minor ad-hoc changes to, say, improve the resolution of their electron microscope, and the manufacturer of the microscope incorporated those improvements into the next commercial model. This may seem like a special case, since academic scientists are willing to do free R&D for manufacturers. It turns out this pattern isn’t unusual, though. A study of PC circuit boards and semiconductors found a similar pattern: users of semiconductors made “homemade” one-off versions of some invention (for example, the process of building semiconductors on silicon rather than germanium). They then either became manufacturers themselves, or transferred their knowledge back to the manufacturer via an order for “formal” versions of the new invention, or knowledge of the new process or product became known in the industry (“everyone knew x has a design for y”).
This pattern wasn’t true everywhere, however. In tractor shovels, for instance, nearly all important inventions come from manufacturers. In wire termination equipment (basically, the caps that terminate a cable), suppliers of the caps and connectors often developed inventions that made it easier to use their products, then gave those designs to manufacturers. Why do different industries see totally different sources of innovation? The answer is nothing surprising for an economist: incentives. In general, it is difficult for a firm to “switch industries” – e.g., go from being a user to being a manufacturer – and in general, patents and trade secrets do not offer very much protection (patents can be invented around and trade secrets leak out during plant visits by suppliers, among others). So the question of who will innovate depends on how much profit can be gained by innovating. If users have the appropriate engineering talent, costs of improving existing products in-house will be low. Scientific instrument users have such skills, while tractor shovel users (i.e., construction workers) do not. If users can keep a new process or product secret, it is more likely to be developed by them than by manufacturers, who are limited in how secret they can keep a product they wish to sell to other people. The basic point is that it is not always (rarely?) the case that manufacturers are best placed to develop new processes and products at low cost and then extract rents from the innovation.
Two other interesting results: First, informal trade of technical know-how is widespread, even among rivals. Engineers often call workers at rival firms when they are stuck on a problem. Even more interesting, and related to some of my own current research, firms often publicly give away details of new and unpatented innovations at scientific conferences and the like. What is going on here? Von Hippel hypothesizes that firms are playing a type of repeated prisoners’ dilemma, and if transaction costs for formal licensing are high, and the value of information is relatively low (because, for instance, other firms could figure out the information with some effort), this sharing relationship is stable. Nothing here that would surprise an anthropologist! Second, firms can affect the sources of innovation. Von Hippel and a coauthor examine a medical testing device sold by three manufacturers. The device made by Dupont was, for technical reasons, much harder for users to modify, but otherwise was very similar to the other two. Looking at medical publications, they find that users are much more likely to develop new products and processes using the other two devices. The moral: firms at loci that are poorly places for extracting rents from innovations might be well served by product design choices which are more conducive to user-led innovation.
(Update: I meant to jot down a quick follow-up. One of my favorite eras in history gives us a fourth locus of invention beyond manufacturer, supplier and user: the rent extractor. There is a lot of historical evidence that various Scythians took actions in the period BC to encourage trade so that they could, once trade routes were established, then steal some of that trade. I like to imagine these nomads giving knowledge of inventions requiring minerals from both China and the Middle East to both ends, then “taxing” back the R&D costs!)
http://web.mit.edu/evhippel/www/sources.htm (Von Hippel is great. He somehow convinced his publisher to let him put this entire book online for free. Given that the book is from 1988, this must be an ex-post negotiation. Good for him! Paper copies are available at Amazon or your local library, of course.)
A Fine Theorem 