Economics of Science

Paula E. STEPHAN
This talk focuses on the economic perspective of science, why scientists do research, what their rewards and motives are, how science is produced, and suggestions regarding under-researched areas. The focus will be on research done in non-profit research settings, but some of the discussion also applies to scientists working in industry.

There are at least four reasons why scientists engage in research. First, scientists are very interested in establishing and building reputations. Second, there is a clear interest in solving puzzles. Third, money is another factor. Finally, scientists are also interested in "doing good."

One area that economists have not spent much effort studying is the distinction between motives for people to conduct scientific research, such as the utility they derive from research, and incentives for doing research. It is important to pay attention to this distinction. It is also important to think about how motives vary across fields and sectors. This is partly due to self-selection, as well as to the fact that values get reinforced in an organization. For example, when we measure the monetary motive, we find that people in industry have a much stronger monetary motive than people in academia. How much of this is learned as opposed to self-selection is difficult to determine.

Reputation in science is established by being the first to communicate a finding or, as Robert Merton says, "To establish the priority of discovery." A necessary condition to establish priority is to report one's results. This has lead to a huge industry of counting citations to measure the importance of a scientist's work.

We see all kinds of evidence of the importance of priority in the behavior of scientists. The order in which one's work appears on a program, for example, is extremely important. It is not unknown for an article to be written and submitted for publication on the same day or for scientists to enter into lengthy discussions with editors regarding their submission. If one knows someone else is submitting a related paper to a journal for publication, discussions can occur on whether the two papers can be published in the same issue. The publication of the resulting "tie" is a somewhat common occurrence that is an under-researched topic. Arguments concerning who will be the first author on a publication can also occur.

The recognition awarded to priority takes several forms. The first is eponymy, the practice of naming the discovery for the scientist, for example, Hodgkin's disease and Moore's law. Prizes are also very important in science and these continue to grow with some having significant amounts of cash associated with them. It is interesting to see the tremendous growth in the number of prizes being awarded, particularly between 1995 and 2005 when the world economy was doing relatively well. This is another area that is under-researched.

Another form of recognition is election to learned societies. Most countries have societies that the most luminary scientists are elected to, for example the National Academies of Science in the U.S., the Royal Society in England, the Academie des Sciences in France, and the Japan Academy as well as the National Academy of Engineering in Japan.

Publication is a lesser, but nonetheless important form of recognition, as it is the way a scientist makes the discovery his own. Darwin only finally agreed to publish when he found out that Wallace might do so first, thereby establishing priority.

Scientists are also truly interested in puzzle solving. The Nobel physicist Richard Feynman stated, "The prize is the pleasure of finding the thing out, the kick of discovery." There are many other examples of scientists describing the joy of solving of puzzles.

When asked to score motives for doing research, scientists rank an interest in challenge and independence higher than an interest in salary, advancement, or contribution to society. This truly complicates methods of modeling research behavior because economists are used to thinking that one gets satisfaction or utility out of goods that go into one's utility function and that the price of these goods is independent of the amount consumed. However, according to Gary Becker's theory of time intensive goods, if satisfaction is derived from time spent solving puzzles, then one is doing the research activity not solely for the end result, but because of the enjoyment of the process and price becomes endogenous. This makes finding a unique solution almost intractable.

With regards to money, people are interested in and motivated by money. Salaries in science are greater than in the past. The gap between highest and lowest salaries has increased. Royalties from patents play a huge role. University scientists start up firms and they also consult. However, it is very hard to get good consulting data, at least in the United States.

Compensation in science and engineering varies considerably across countries. This is related to flows of human capital across countries as well as the funding for science. In the United States, compensation varies by rank, field, and productivity. There is a distinct difference in the salaries of full professors and associate professors as well as between the average salary and the highest salaries. There is a relationship between rank and supposed productivity. A number of studies have looked at the relationship between salaries and publications, citations, etc.

Another financial reward in research is receiving income through patent royalties, and this has become increasingly important in recent years. For some countries it is difficult to get good data on who is patenting. Additionally, data on the percent of people who are patenting depends tremendously on the kind of instrument that is used to collect data and who is included in the sample. Across all disciplines and university rankings, about 10 years ago 13.7% of faculty were patenting in the U.S. A different study surveying only top researchers in the biomedical field more recently found that 43% were applying for patents.

In the U.S., patents belong to the university, but the royalties are shared with the faculty. There is variation in the sharing formula. On average, the faculty receives 41%. There is considerable skew in the amount of royalties universities receive. It is estimated that 375 faculty shared approximately US$560 million in royalties resulting from blockbuster patents in 2004, but very few receive more than US$1000-2000. At more than half of research campuses, only an estimated 3-5 earn more than their salary each year in royalties.

Some questions that emerge are whether faculty patent because there is a very small possibility that they can get a very large amount of money, whether faculty know what the royalty sharing formula at their university is, whether faculty would be motivated to choose a job at a university with a generous formula if they have a high probability of patenting, and whether faculty are more likely to disclose if they are at a high royalty institution. This is an area which is under-researched.

Another way for scientists to earn money is to start a company. Most of the rewards are realized at the time a company makes an initial public offering. On average, a founder will receive US$5 million in the form of equities. Additionally, advisors for these companies are mostly drawn from the same faculty; they are paid for their advisory role and they are given stock as well.

In summary, scientists are motivated by an interest in reputation, puzzle solving, and money. However, further research reveals that other factors matter as well. One factor is "doing good," that is, the motivation to engage in scientific research in order to make an impact on society. Our research shows a positive relationship between how highly scientists value contributing to society and the amount of patent activity. However, we find this only in the biomedical sciences, which does seems logical since it is one of the fields where patent activity is extremely important in the commercialization of drugs. Scientists in this field may recognize that they cannot help people unless they have patents.

Science has many characteristics of what economists refer to as a "public good." In economics, the lighthouse is often given as an example of a public good: It is not "used up" by use and once built individuals cannot be readily excluded from its use. Science has similar properties. When one learns of something someone else has discovered, that person gains knowledge, but the knowledge of the one who discovered it is not depleted. Furthermore, once a discovery is made public in science, others cannot be excluded from its use. In economics, it is well known that markets do not do a good job in creating public goods because it is not possible to capture the rewards. That is a cornerstone of economic theory.

Reputation helps solve the public good problem. The only way a scientist can establish reputation is to "give" his research away, generally through publication. Thus reputation is a very functional reward system. It also solves the monitoring problem. In labor economics, there is huge concern about how to determine compensation in jobs where it is difficult to monitor how people are doing. In science, reward is based on achievement, rather than effort. This, additionally, resolves the problem of shirking as it is clear who one's competitors are.

Science has aspects of a public good. The reward structure has evolved based on priority and solves a number of issues. However, an interest in reputation does not necessarily mean full disclosure. It is possible for scientists to keep back data or material for themselves.

Some economists describe science as a winner-take-all contest. However, this is an extreme view since replication has a social value and a single problem may have multiple solutions. It is more realistic to describe science as a tournament model which does not only reward the person in first place, but also rewards those in lower places, keeping individuals in the game, and keeping their motivation up for the next round.

Economists have a long tradition of thinking about how things are produced. They examine the production function, the price of inputs, and the relation between these two. A good example is the case of cars. Cars are produced by labor, equipment, and raw materials. As labor becomes more expensive, companies are incentivized to substitute labor with cost effective equipment and the composition between labor and capital changes. As raw materials become more expensive, the production process shifts to different kinds of inputs in manufacturing.

Science is similar in many ways. The production function depends upon effort, materials, equipment, skills, and knowledge. The product is an output such as the publication of an article or an idea, but the scientist does not supply all the inputs directly.

There are big discipline differences. While every scientific discovery requires effort and skill, in certain disciplines other inputs are required as well. Equipment intensity varies considerably across disciplines. The way research is organized, the importance of materials and collaboration patterns also vary.

Costs affect the way research is conducted. In the U.S., for example, there is a trend of staffing laboratories with cheap labor, such as graduate students and post-doctorate students. Scientists share material and build equipment to save costs. The sensitivity of research activity to costs is another under-researched area in economics.

The headline in the New York Times on March 10 read "Decoded Genome gives New Hope in Confronting Disease." The genes were decoded at a company rather than a university. This was significant in that the lead investigator was Leroy Hood, the inventor of the first sequencing machine. He is now sending his sequencing project to a company for sequencing.

One of the most important inputs into the production process is the scientist's time and effort, or persistence. In a study where 25 physicists were given 25 adjectives to describe the reason for their success, over 50% chose the word persistence.

Cognitive resources are extremely important. On average, scientists are highly intelligent. Creativity is important. In terms of knowledge, tacit knowledge is particularly important because knowledge does not come only from reading scientific publications, but also from the people scientists work with.

Another aspect is collaboration. Research is rarely done in isolation and a variety of skills and knowledge is important. On the other hand, there is specialization within a lab. The way labs are staffed varies considerably across countries. In the U.S., it is common to staff labs with graduate students and post-doctorates because they are cheap, temporary labor. In other countries, for example, in Europe it is common to have permanent staff and research is often conducted at a public institute affiliated with a university.

It is very easy to see how the importance of collaboration has grown over time. Over the span of 45 years team sizes have increased in most fields, even in the field of math. A quarter of U.S. scientists have coauthors who reside overseas. Cross-university collaboration is on the rise, while sole authorship is going down. The reason for this is because interdisciplinary research is becoming increasingly important. Specialization is becoming a trend, leading people to acquire narrower fields of expertise. Collaboration also minimizes risk.

Equipment also plays an important role in research. A recent profile of Lila Gierasch states that she was "Wooed by an NMR machine" to the University of Texas Southwestern Medical Center. This highlights how important access to equipment is to researchers. Most equipment is expensive and the most expensive equipment is shared. To gain access to equipment, job offers in the U.S. come with start-up packages which provide resources for equipment and staff for the first three years. Subsequently, equipment is acquired through various grant proposals and institutional proposals and access to extremely large equipment is gained through the proposal process.

It is important to realize that scientists are often instrumental in creating much of this equipment. For example, the automatic sequencer was developed by biologist Leroy Hood. This was later replaced by new technology and will soon be replaced by even newer technology. Costs of sequencing have fallen. There is a prize now with the goal to sequence the genome for US$10,000 and scientists are closing in on this goal. How costs for scientific equipment are determined is an under-researched question. These are very specialized markets. In addition, many instruments are one of a kind, giving scientists a monopoly, in a sense, of discovery. Currently, there is no one who specializes in studying markets for research equipment.

Materials such as plants and animals are extremely important in scientific research. While inexpensive material such as yeast, zebra fish, and fruit flies are important in research, 90% of all animal models are mice. They are expensive. Purchasing mutant strains and recovering strains from cryopreservation are especially expensive. The cost of keeping mice and equipment for mice is also growing.

Science is big business. While the matter of how science is funded is a topic for another lecture, suffice it to say, in most countries funding comes from the government. The government funds scientific research because it is a public good. Despite the incentives to do scientific research, research is expensive and scientists are unable to gather the resources on their own. There is also strong evidence that science contributes to economic growth. That, too, is a subject for another lecture.

AOKI Reiko
There are many things we should learn from Prof. Stephan's papers. We should also learn how economics contributes to learning about science. Understanding how science works from an economics perspective is very important to understand how science will link to the next step of innovation and technology.

The economics of science is a branch in applied economics and it is about using economic methods to study the production of science. Just as applied economics can be theoretical and empirical, the economics of science has both theoretical and empirical aspects. Because it is an academic discipline, it has self-generating paradigms which, as Prof. Stephan pointed out, are areas for future research. It is also an input for other sectors of society such as business and public policy.

What should we learn from an economic perspective? This would include the incentives for scientists to do research; the production function of science that produces public good, human labor, human capital, and collaboration; and the cost of inputs.

From a policy perspective, as Prof. Stephan pointed out, the output of science is a public good, and therefore knowledge is an important aspect. Since the priority system, is not a for-profit market oriented concept, the academic community could play an important role in maintaining the integrity of this system. Prof. Stephan also elaborated at length on patents. The funding system was saved for another lecture as Prof. Stephan talks about institutions versus research funding in another paper, however, this is currently a very important topic in Japan.

Universities can be interpreted as having a production function of science for the purpose of public policy analysis. If this is the case, how we can go to the production frontier and how universities can act efficiently as a production function become the research questions, as are the implications of science as a production function and polices regarding networks and consortiums. Another aspect is the relationship between science and the labor market, such as designing career paths for scientists and how to help the science labor market function better. In terms of human capital investment, the optimal polices for graduate students and young scholars should be considered. Another big question in Japan at the moment is how to fund "Big Science" as a domestic project as well as in terms of international collaboration.

In regard to the relevance of the economics of science, from the public policy perspective in Japan, we are currently at the end of the Third Science and Technology Basic Plan and the economics of science has much to contribute to the design of the Fourth Basic Plan.

As for academic relevance, there are well known tools in economics for approaching issues in the production of science, namely human capital, organization and governance, and industrial organization. Prof. Stephan pointed out many under-researched questions of academic relevance.

As for public policy relevance, science can be input for public policy. The rationale for public funding for basic science is that scientists produce a public good and science contributes to economic growth. Another important aspect is that science can be linked to innovation. Currently there is a needs-pull approach to innovation and linking basic science to green innovation and health innovation is important. We are still at the stage of trying to shift from a needs-pull approach to the needs pull approach. We also need to learn how to design an efficient innovation platform from an economics of science viewpoint.

In a similar view, there is a need to prescribe a road map using methods to implement green and health innovations. In economic terms, there will be incentives and institutional design. Funding will relate to the intermediate microeconomic cost minimizing of science's production function. Human capital investment strategies are also part of the roadmap that the economics of science can provide.

It is also important for those engaged in the economics of science to make this science relevant. We should also always keep in mind that we can use science as an input. It is important for policy makers and voters as consumers of science, who use science as an input, to understand what is on offer.

Q: Are you defining science as something with monetary value such as patents, creating some kind of quantifiable concept such as welfare or utility for science, or handling it as something which cannot be quantified but creates something similar to a production function?

Paula E. STEPHAN
From a general point of view, I think of science as the creation of new knowledge. Of course, it is difficult to have a quantifiable measure. Therefore we always look for proxies, such as publications or patents. In the abstract, my definition of science is the generation of new knowledge that can push the frontier of what we know about the laws of nature. However, I have no concept to which you can put a monetary value.