NAME: Peter Bossaerts
POSITION: Professor of economics and management and professor of finance, California Institute of Technology; professor, Swiss Finance Institute, Ecole Polytechnique Federale, Lausanne
BACKGROUND: Assistant professor, Carnegie Mellon University Graduate School of Industrial Administration; PhD, management (finance), University of California, Los Angeles; doctorandus, applied economics, University of Antwerp
The US patent system is in the spotlight again. Patent reform legislation is being mulled in Washington and many pundits are urging further development of a so-called "innovation economy" to weather the current economic downturn.
One question continually raised is, 'Is patenting the best way to promote intellectual discovery?' It might not be, according to a group of international economists, which last week published research suggesting that a system based on co-operative markets might be better.
In this system, described in a paper published in the March 6 edition of Science, participants own shares in the components of potential discoveries and can trade those shares in an anonymous market. The incentives to invent in such a market are indirect, through changes in share prices, according to the researchers.
To test their theory, the researchers used a classical applied mathematical model known as the knapsack problem, in which participants must determine the most valuable subset of objects that can fit in a knapsack of fixed volume.
The researchers found that their market-based system performed better than a patent system in the knapsack problem, primarily because compensation for inventions is shared, and because discovery remains in the public domain, thereby avoiding "both distortion in the provision of newly invented products and stifling of future discovery."
CalTech's Peter Bossaerts, who led the study, discussed some of its implications for current patent-based economies in an interview with BTW. Following is an edited transcript.
What was your motivation for conducting this study?
The motivation for patenting comes out of classical economic theory, which assumes that economic agents are very rational, have the potential to solve any problem that is out there, and are homogeneous. If you look at it from that point of view, then of course a patenting system is optimal. And if you then analyze what such agents would do if there were no patents, but only markets, in fact markets would be a disaster: nobody would spend any effort trying to solve problems, because from the moment someone offers to trade, he or she will not find a counterparty – the mere fact that [he] wants to trade signals that [he] knows something.
That's what economic theory predicts. But this is a theory that works only with abstract people – economic agents – that really aren't human beings, [but] are hyper-rational, fully capable of solving anything, and homogeneous. But we know from modern psychology that humans aren’t like that; they don’t think that they are all equally good [or] that they can potentially solve all things. To the contrary, studies have shown that when you award a prize to a person that does the best, it has a negative effect on their initiative and on their effort. They simply don't believe that they will ever be the best. They think that they may be second or third best. That’s bad for [the] patent system [as] we know it, because it only awards the first best.
[ pagebreak ]
We wanted to start thinking about mechanisms for intellectual discovery that weren’t designed for the agents that populate theoretical economic models, but for human beings as they actually are. We decided to think about something we have known for a long time about human beings: they are overconfident. Specifically, more than 50 percent of people usually think that they are better than the median. For instance, you may ask about driving skills, and 70 percent of people will tell you that they're better than the median, which is not possible. This is a very widespread phenomenon.
If that's the case, then since market prices actually reflect the median, we conjectured that markets could do the job after all. That's what we tested.
There is a contradiction in the ideas that very few people believe they are most likely to come up with a unique solution, but that more than 50 percent of people believe they are better than the median.
They're not [contradictory]. One is about the tail-end of the distribution of skills. It is about what statisticians call an order statistic: the maximum; the best. If you ask a person, 'Are you the best driver in town?' you will find very few people who will answer affirmatively. However, you will find that the majority of the people think they are better than the median. The median is the middle of the distribution, not the tail-end. There is a subtle but very crucial difference.
For a patent system to work, you want 10 percent, 20 percent, or even more of the population to think they are the absolute best. In fact, in economic theory, 100 percent of the people know they are the absolute best, and are all homogeneous, and they all expect to win the prize. When one person wins over another, it's just an accident.
What we found amounts to an absolute rejection of classical economic theory, which predicted that our market experiments would not work at all. We do show that a patent system works, and a classic economist would say, 'Yes, we told you that.' But [classic economists] also would say that our market system couldn't possibly work, and that no innovation will ever happen in such a system. But we showed that it works as well as a patent system, if not better.
You’re talking about the knapsack experiment, right?
Yes. The issue [was]: How do you set up an experiment about intellectual discovery? The classical way of thinking about intellectual discovery in the economics literature is information aggregation. That is, to solve the problem – inventing and discovering something – is a matter of just finding the right information, and putting all the information together.
We take issue with that, because if you think about discovery, if you have all the pieces of information, you don't necessarily know what works right. It's not an additive problem. In fact, it's the uniqueness of your recombination that fuels discovery – you have to know how to combine pieces of the puzzle.
Knapsack problems are an idealized version of that type of problem. But they are very difficult. It is what the mathematicians call an NP hard problem, which means that the only way for you to know that you have the absolute best possible solution is to try all solutions. That’s not true for information aggregation: as you collect more and more information, you get closer to the solution. So, that’s relatively easy.
This idea of several individuals combining information or effort to come to the best possible solution – is this necessarily at odds with a patent system? In many disciplines, each patent represents an incremental advance that, when building upon other advances, eventually leads to an ideal solution.
That is correct, but one accusation that has been leveled against the patent system is that to implement the latest link in the chain, you have to go talk with all the other patent holders who hold the other links, and coordinate with them to bring the discovery to an end. Any link in the chain can now hold you up. If you have the last link, and you know it, but you also know that you need previous links in the chain, then any of these links can just say, 'Look, I want a major piece of the pie.' That's not true in the market system because the ideas are in the public domain. Nobody can hold up an invention.
[ pagebreak ]
Do you believe that this experiment and your patent market-economy theory applies across all industry segments? Biotech patenting is very different from IT patenting, or engineering patenting.
Just as the [current] patent system, which favors intellectual discovery in certain fields because it is more amenable to describe what intellectual property is in those fields, our market system favors intellectual discovery in those fields, as well.
In fact, one thing you can tell is that it will hinge very much on the existence of economic rents. Basically, you need the components that go into your discovery to be in 'short supply.' There shouldn't be too much of it, so that the owner of that component – for instance, using the examples of platinum in fuel cells – is able to make extra money because he is actually holding something that is in limited supply. It doesn't have to be goods, it can also be services.
Let's say you want to solve the energy problem, and you can find a solution that depends on resources that are plentiful – let's take water as an example. If you could transform water into energy without much cost … then our system wouldn't work – except that the prices of other energy sources that compete with your invention, such as oil, coal, and gas, will all crash. And then you can profit from those price drops.
But when you know that the introduction of your discovery is not going to affect any prices whatsoever, then it's not going to work. It's not going to incentivize people.
This study might be construed by some as an indictment on a patent-based innovation system, especially in the US, which relies heavily on patents.
This is very timely in that people are talking about this. But I don't think this is an indictment on the patent system. In many respects, what we actually tested the patent system against already exists. The fuel cell example [in the paper] is a very nice case in point. But the energy example I gave you is also a good case. If you really are convinced that you've found a way to produce cheap energy, then you should take a position in competing energy sources.
It's not an obvious thing to be doing, though. For instance, many of your readers are experts in only one part of the job. Indeed, in the market system, they not only have to solve a problem, but they have to understand how this discovery will affect the economy as a whole, and prices in particular. They have to think about how introduction of an invention will change prices of goods and services out there. And that is of course something that very few inventors are used to.
But that’s in fact not unlike what happens in a patent system, [in which] you also need skill beyond discovering things if you really want to monetize your patents. Most inventors don’t have this skill and solve this by talking to venture capitalists or financiers in general.
So I don't think it is an indictment of the patent system, but it is pointing out that there is something else out there that, to a certain extent, is already enhancing the patent system.
Another point on this is that I had someone suggest to me that the US is the most inventive nation in the world, and that this is because of patent laws. But you cannot prove this unequivocally. During the period that patents were introduced and the US became arguably the most inventive nation on Earth, it also promoted free markets, and established many well-organized markets such as the Chicago futures market. The latter is a confounding factor if one wants to attribute historical inventiveness to patents. This is why we need not only look to history to understand what works, but to do controlled experiments to try and explain what the incentives are that patents give and that markets give. This is what we did.