In a world of finite resources, what would become of humankind if everything were someday depleted? Such a stark scenario is softened by human ingenuity and its resulting technological change, which over the ages have consistently shifted our focus from expensive, limited resources to cheaper, more plentiful ones and, in the process, have transformed seemingly free resources into valuable ones. We have turned sand into silicon chips, harnessed the energy in sunlight, and mined our very bodies for valuable protein-based drugs. Most resources will never actually be depleted: as their supplies dwindle, prices will rise and trigger a search for replacements. This process is as sustainable as our ingenuity.
Our ingenuity also has its downside: clean air and water, for example, are among the once-free resources that we have transformed into valuable ones by making them scarce. Still, by creating value and efficiencies in resource use, technological change is our best engine of growth. Since technological change comes at a cost, there is at all times some optimal level of investment in it. To invest more, or less, than this amount is to use our resources at less than full efficiency. This article is about how we find that optimal level and how, in an economy where investment decisions are decentralized, we attempt to induce that level of investment.
Patents figure heavily in both of these considerations. To firms, patents give monopoly rights to their intellectual property; as a reward for innovating, patents therefore provide incentives to perform research and development (R&D). To economists, patents not only correct imperfections in markets for research, but they are also a rich source of data: there is an abundance of them (more than 5.3 million U.S. patents have been issued since 1790), the data are easily collected, and they contain technically precise categorical information about each innovation. Patents are perhaps the most widely used indicator of the innovative output of a market economy.
Patents are more than a reward or a source of data, however. They have real effects on firms, their rivals, and consumers. The primary purpose of my research on patents has been to estimate these effects. My work has been motivated by having observed the extremely intense competition for patents in certain fields (with large consequences for the winners and losers). What is a patent worth? How lasting are its effects on rival firms? My secondary purpose has been to use these estimates as a way to improve the performance of patents as an indicator of research output. Useful as they are, patents have several shortcomings in that role, which I will discuss below.
Why patents are valuable
A monopoly is the most profitable of all possible market structures. Since patents create monopoly power for inventors, patents can be a powerful inducement to innovate. (Where an invention is one of several similar products on the market, however, the inventor's monopoly power might be quite limited.) Where a patent would lead to large profits, it can attract entrants to the field, raising the chances of inefficiencies from multiple firms, performing duplicate research. Where patents are weak, firms may rely instead on trade secrecy, and the social benefit of the information disclosed in a patent is lost. Although monopolies themselves come at a social cost—they raise prices and lower output relative to any other market structure—there tend to be positive net social benefits, in both the short and the long run, from innovation.
Biotechnology research, a high-stakes gamble among hundreds of very young firms, is where these ideas are illustrated most forcefully. Building on the initial discovery of recombinant DNA techniques in the 1970s, firms suddenly could genetically engineer any of dozens of drugs that until then could only be distilled painstakingly and in small quantities from natural sources. At times, eight firms and more have raced to discover and patent a protein's genetic sequence, to perfect a process for synthesizing the protein, or sometimes—with luck—to do both. Often, however, product and process patents have been issued to separate firms, and this has necessitated many expensive, time-consuming lawsuits to sort out who owns the rights to the market.
Since the 1970s, biotechnology research has become a high-stakes gamble among hundreds of young firms, each racing to discover and patent a protein's genetic sequence or the means of producing it.
Perhaps the most revealing example of the stakes involved was the race to develop synthetic erythropoietin (EPO), a kidney hormone, the absence of which causes anemia. EPO-replacement therapy eliminates the need for blood transfusions in dialysis patients, whose failed kidneys no longer manufacture EPO. Since the patients then must dose three times per week while awaiting a kidney transplant, the EPO market was extremely attractive to potential suppliers. In this race, however, there was no immediate lucky winner: in October 1987 a biotechnology firm named Amgen got the patent on the genetic sequence coding for EPO. The trouble was that a rival firm, Genetics Institute (G.I.), had won a patent on the EPO molecule itself three months earlier—to Amgen's complete shock, because G.I. had filed its application a year after Amgen filed its own.
Because the scope of these two patents seemed to overlap, each firm sued the other for patent infringement. The litigation played out over four years before the case was settled in Amgen's favor: the firm currently has a complete monopoly over synthetic EPO in the United States. Largely as a result of this monopoly, Amgen's corporate worth has increased more than tenfold in recent years, to the point where it is now the world's largest biotechnology company. Meanwhile, G.I. was forced to stop manufacturing EPO and has since had to seek an outside buyer.
Although Amgen's patent ultimately forced G.I. out of the EPO market, without the prospect of one firm's winning an exclusive patent, it is possible that neither firm would have pursued its EPO research at all. Without exclusive rights, not only would the rivals have had to share the market, but by competing they would have undercut the monopoly profit outcome—leaving each with half of a smaller pie. In a world without patents, the two firms also would have been vulnerable to market entry by any other firm. If the absence of a patent system would have made expected returns from EPO investment appear negative, the two firms might not even have begun the research, and there would be no EPO treatment for dialysis patients.
Thus, in a market economy—where most R&D investment takes place in the private sector—patents are necessary to allow firms to protect their investments in R&D. While firms pursue patents for profit, governments award patents because of the so-called public goods problem: once new knowledge such as an invention has been produced, it often can be duplicated at low cost and disseminated without financial benefit to the innovating firm. Without a patent system, firms would have to rely strictly on secrecy. This would slow the pace of technological change, since patents not only encourage R&D, but also codify and disseminate new knowledge in a way meant to help future innovators improve on the inventions. In the extreme, if secrecy were ineffective at allowing firms to appropriate at least some of the benefits created by their inventions, firms would have no incentive to innovate, since they would recover none of their investment.
Building on the discovery of recombinant DNA techniques in the 1970s, biotechnology firms became able to genetically engineer dozens of drugs that until then could only be distilled painstakingly and in small quantities from natural sources. The ability to obtain a patent—exclusive use of certain knowledge—means that companies can make money on the results of their research, thus encouraging them to invest in R&D. The patented results of research then become widely available, providing benefits to society as a whole
Using patents to monitor R&D investment output
The importance of technological change to growth is widely acknowledged, but it is a challenge to know how much private (and public) R&D investment there should be. Among all possible R&D projects, the law of diminishing returns dictates that beyond a point, each additional dollar of investment in R&D will yield less than a dollar in total benefits. Firms will not choose to invest to this point unless they can appropriate all of the benefits from their investments—a practical impossibility, since firms would have to be able to charge individual consumers their maximum willingness to pay for the firms' products. Patents cannot fully overcome the appropriability problem, but they can raise investment levels closer to the optimum than they would otherwise be.
A patent system that is too weak will not stimulate enough research; one that is too powerful could induce too much—by attracting too many firms into research—or too little—by giving an original patentee overly broad rights to control future inventions. In theory, the system could be tailored by varying the value of a patent up or down whenever private R&D investment is too low or too high.
As a means of encouraging optimal levels of R&D, though, the patent system is a very blunt instrument. A patent's value for a given invention is a function of its length (the number of years it is in force), its breadth (the span of its claims allowed by the patent examiner), and a firm's reputation and resources for enforcing the patent. In practice, these adjustments to patent value either are not practical or not yet well-enough understood to use patents to fine-tune R&D investment levels. The statutory length of a U.S. patent—seventeen years—has been essentially unchanged since 1790. The more useful policy instrument is patent breadth, which affects a firm's ability to control its current inventions and any future ones they inspire. Patent breadth can be varied on an individual basis, and economic research has begun to show how a patent's breadth relates to its value.
For economists to estimate the optimal level of R&D, they would need better measures of R&D productivity than are currently available. Patents are a promising source of data for this exercise, but they have several limitations. One, which I suggested earlier, is that not everybody patents their inventions. Trade secrecy tends to be preferred to patents, especially in fields where inventions become obsolete quickly and do not need the lengthy protection of a patent. My research is aimed at another problem—that not all patents are equally valuable. To calculate productivity, economists need to know the value of outputs: a simple count of issued patents would not estimate accurately even the total output of patented innovations. A simple count would value minor innovations (a new type of door latch, say) equally with major ones (the DNA sequence for EPO). Economic research has shown that if, instead, patents are weighted by something correlated with their relative values, the resulting weighted sum much better represents the value of patented outputs.
New patents drive firms' market values
My approach to estimating an individual patent's value is to measure the change in a firm's market value when it receives a patent. This technique, called an event study, often has been used to determine the effects of specific, significant events on firms affected by them, such as product recalls, new regulations, or mergers. While other approaches are available, the event study can help estimate the effect of a particular patent on, say, rival firms researching the same drug. In effect, an event study compares changes in firm values with the values that would have been expected had the event not occurred.
Estimating the value of a patent based on movements in a firm's stock price relative to the market
Often these changes are measured over just a few days, as I have done. A crucial assumption of this technique is that all relevant information is reflected immediately in the stock prices of affected firms. Assuming no other significant events have occurred around the time a patent is issued, the change in a firm's value from that patent event represents the value today of all net income the patent is expected to generate in the future. (Future income is discounted because one dollar "tomorrow" is less valuable than one dollar today.) The market's patent valuation is a forecast, but one that is based on the market's historical experience with patents.
The precise technique by which estimated patent values are teased out of changes in a firm's stock prices is somewhat involved. In essence, the estimate is the difference between the actual change in a firm's value around the time it receives a patent and the change that would have been expected had there been no patent. This expected change depends on what the market as a whole has done over this period, since individual stocks move in characteristic ways relative to the market. Take a firm whose stock price tends to move with the market. If its value increases by 2 percent in a week in which it receives a patent, and the market average climbs 1 percent over that same period, the estimated value of the patent would be something like the difference of these numbers, in this instance 1 percent of the firm's value. If the market average had instead dropped by 1 percent, the patent's value would be 3 percent of this firm's equity.
The results of my study confirm some commonsense notions about patent value, but there have also been some intriguing surprises. With my focus on biotechnology patents, I constructed a sample of more than 550 patents issued to the twenty largest domestic biotechnology firms—and an additional seventy patents issued to other companies in these firms' research areas. I found that, for a typical patent in this sample, a firm's value increased by 1.2 percent (net of the market change) in the first two days after it received the patent. That is an annual rate of 141 percent; for a firm of median size in this sample, the patent is estimated to be worth $2.4 million.
I then looked at how a firm's value changed when it received a key patent, one that experts have tended to identify as being among the most influential biotechnology patents. Strikingly, for firms holding key patents, the estimated values average an astonishing $11 million. These high prospective valuations appear to have held up nicely into the future, in the sense that patents of above-average estimated value have demonstrated a greater likelihood of being cited by later patents, compared with citation rates for the less valuable patents in my sample. Citations, which are made whenever subsequent patent applications build on what is contained in the cited patent, have been shown elsewhere to correlate with the actual value of an invention.
My research estimated the direct effects that patents have on rival firms as well. By examining the contents of each patent, and by knowing which biotechnology firms are researching which compounds, I was able to identify all of the rival firms in each of seventeen selected research areas. These firms may be in competition for the same patent, but, more importantly, they are after shares of the same market. I found that when a firm received a patent in one of these research areas—this describes about a quarter of the patent sample—the values of its rivals dropped by an average of 0.6 percent ($1.3 million). Here, too, an influential, key patent has a much larger effect, causing a 2.4 percent ($5 million) decline for each rival. These numbers represent the direct effect of the patentee's expected market power, although I found, by measuring effects from patent applications, that the other firms actually appear to benefit from knowing that the prospective patentee has been successful in its line of research (which through scientific exchanges is evidently somewhat familiar to all of them).
The remaining question is to what extent biotechnology patents enable firms to control their particular markets. Within the research areas I have studied, I find some limited support for the notion that a key patent depresses the future research output of rival firms. Interestingly, only in the case of EPO did a patent appear to have forced out all rival firms. In most cases, where one firm wins a valuable patent, its rivals have appeared to carry on with their research, but at a less intense level. These firms all have research going on in multiple areas, so they may decide to shift their resources to areas of more immediate promise. On the other hand, by not actually exiting the market where another firm has received a strong patent, these firms perhaps are maintaining their research in that area, in hopes they eventually will be able to develop a patentable, second-generation product.
Only once did a firm's winning of a valuable biotechnology patent appear to force out all rival firms. Usually, rivals carried on their research, but at a less intense level.
Looking ahead
Increasingly, government investments of all kinds are coming under scrutiny by a Congress intent on lowering the federal budget deficit. Among the most prominent of spending programs currently in the sights of Congress are two that involve government coinvestments with private firms performing advanced-technology research. These programs are derided by their critics as subsidies for private industry: the National Institute of Standards and Technology's Advanced Technology Program (ATP) and the U.S. Department of Defense's Technology Reinvestment Project (TRP). In both programs, funded firms are not required to repay those public investments even should they ultimately bear fruit. The purpose of the programs, however, is to create large public benefits (such as an enhanced standard of living and increased U.S. competitiveness) by funding ventures considered too speculative for private venture capital markets to be willing to bear the risk.
Supporters of federal investments in private R&D, including the Clinton administration, will have to rebut the critics by demonstrating positive net payoffs from federal investments in ATP and TRP. To date, the administration has no hard evidence to support this contention; the patent methodology I have presented here can be used to value some of the benefits from these programs. Valuing the patents that have resulted from public investments in private R&D will count only a fraction of the total benefits that may result from these programs. Still, if the values of the patents alone are greater than the public investments that brought them about, this would provide compelling evidence for continuing such programs. (If the patents appear less valuable, program supporters would need additional evidence.) If repayment is politically necessary, the estimated values of the resulting patents could be used as a basis for determining the appropriate share of any privately realized profits to be returned to the public coffers by program beneficiaries.
A patent system's primary benefits—overcoming the public goods problem and stimulating technological change—clearly are not limited to these virtues. Patents are invaluable as technology indicators, public disseminators of new knowledge, and building blocks for future inventions as well.
David Austin is a fellow in RFF's Quality of the Environment Division. The research and issues discussed in this article are detailed in RFF discussion paper 94-36, "Estimating Patent Value and Rivalry Effects: An Event Study of Biotechnology Patents."
A version of this article appeared in print in the May 1995 issue of Resources magazine.
A version of this article appeared in print in the May 1995 issue of Resources magazine.
