In my last post, I suggested that effective property systems have two important chracteristics: clear boundaries and positive incentives for productive activity. I showed a graph from Bessen and Meurer's Patent Failure suggesting that patents on chemical and pharmaceutical products appear to be behaving as a well-designed property system ought to. Now, the bad news. Here's the same graph for the rest of the patent system:
Again, the dashed line is total profits attributed to patents, while the solid line is the cost of patent litigation to potential infringers. As you can see, the situation is very different in non-chemical industries: in the late 1990s, the costs of litigation from non-chemical patents were several times as large as the profits those patents generated for their owners.
These statistics, if accurate, are quite extraordinary. If real property worked this way, we'd see $4000/month in litigation costs arising out of trespassing allegations for every $1000/month rental property. Needless to say, there wouldn't be much real estate development in such a legal environment.
The obvious response is that we shouldn't be overly concerned with "trespassers" (alleged patent infringers) because they shouldn't have trespassed in the first place. But this is where the point about unclear boundaries come in. What we're seeing here is not that some companies are deliberately infringing on other companies' patents as an alternative to investing in R&D. Rather, the problem is that there are now so many patents on the books, many of them quite broad, that it is effectively impossible to develop almost any kind of technology without infringing numerous patents. Even worse, because the boundaries of patents are so fuzzy, it's generally not even possible to predict which patents will apply to which technologies. Even an innovator who earnestly tried to avoid infringing, by licensing or inventing around all the relevant patents, is likely to run afoul of a patent his lawyers didn't find, or to face litigation over a patent his lawyer thought didn't cover that invention. What this means is that the patent holders and the alleged infringers are largely the same companies. Microsoft, for example, holds close to 9000 patents, yet it faces dozens of patent lawsuits every year from smaller companies.
How seriously should we take Bessen and Meurer's numbers? Their book just came out so I have yet to see serious criticism of their findings. And I don't know this area well enough to have a strong opinion about how seriously we should take their specific methodology. But to my non-statistician's eye, they appear to have done their homework. On the profit side, they survey a lot of different estimates of patent values and tend to accept the highest reasonable estimates, giving the patent system the benefit of the doubt. On the cost side, their results may be more open to challenge. It's important to note that the litigation costs they estimate are not limited to direct expenses like attorney's fees and expert witnesses. Rather, recognizing that litigation imposes significant costs beyond attorney's fees, they attempted to estimate costs by observing changes in stock price in response to the announcement of lawsuits. If the efficient market hypothesis is correct, this should give a reasonable estimate of the total costs of patent lawsuits to defendants. However, the error bars are likely to be large, so the numbers should be taken with a few grains of salt.