I agree that no adequate, feasible alternative to patents exists for encouraging R&D by manufacturers of pharmaceutical drugs. But my agreement should not be understood as indicating approval of the present U.S. patent system. Patents are a source of great social costs, and only occasionally of commensurate benefits. The social cost of patents that was traditionally emphasized by economists is the wedge that a patent drives between price and marginal cost. Generally, we expect that competition will compress price to marginal cost and that this is efficient because it means that no one willing to pay the marginal cost of a good is deflected to a substitute that might cost more to produce, yet look cheaper simply because the price of the good was above marginal cost. By preventing duplication, a patent reduces competition and so may enable the producer to charge such a price. But the objection to patents that is based on the wedge is superficial because the producer may have incurred costs that do not affect his marginal cost. Suppose the cost of R&D that the producer must incur to bring a drug to market is $100 million, but the marginal cost (the cost added to total costs by each unit of output) is a constant $1 a pill. Then no matter how many pills the producer sells at $1, he will never recover his upfront investment. The real concern about patents is the costs imposed on inventors themselves. There is the cost of searching the records of the patent office to make sure you're not going to be infringing a patent, but more important is the transaction cost involved in obtaining a license from an existing patentee. Invention is a cumulative process; a new invention is usually an incremental improvement on an existing one. So the more patents that are “out there,” the greater are the costs involved in negotiating for a license from every patentee whom the new inventor may arguably be infringing. Because a patent can be obtained without even a prototype, because patent examiners are overworked and it takes less work to approve than to reject a patent application, and because the U.S. Court of Appeals for the Federal Circuit, which reviews patent validity, is extraordinarily pro-patent, the number of issued patents has grown steadily in recent decades. There is concern that some fields are so blanketed with patents (which may be owned by firms that do no production at all—whose business plan centers on demanding license fees under threat to sue for patent infringement) that innovation is actually being impeded. Most firms don’t actually want patents; for those firms, the costs involved in obtaining licenses from patentees are not offset by the prospect of obtaining license fees on their own patents. There are many alternatives to patents for protecting one’s investment in R&D, and they are often cheaper and more effective. They include: trade secrecy; the advertising value of, and the consumer loyalty generated by, being the first to produce a popular product; the fact that marginal cost may increase steeply with output, so that a price equal to marginal cost may cover the fixed costs of invention after all; the fact that it may be costly and time-consuming for competitors to duplicate the invention exactly; and, related to the last point, the learning curve—if costs of production fall over time as the producer learns more about how to make the product at least cost, the first firm in the market will tend to have a cost advantage over competitors, who arrive later. But if other companies are busy getting patents, you may have to patent defensively, and the patent thicket thickens. The pharmaceutical-drug industry is the industry that can make the strongest case for needing patent protection. The investment required to bring a new drug to market is very great, in part because of the many “dry holes.” And it may take years before the new drug can be sold, which shrinks the effective term of the patent (if it takes 8 years to bring the drug to market, the effective term of a 20-year patent is only 12 years). This not only reduces the revenue from the patent; but because the costs of the upfront investment are incurred years before revenues commence, those revenues must, because of the time value of money, exceed the upfront costs in order to be fully compensatory. In addition, once the drug is in production, it is easily duplicated by competitors, and the marginal cost is very low at all feasible output levels, so that with free entry the original producer would not be able to recoup his R&D investment. That said, I am skeptical about the length of the patent term for pharmaceuticals. Congress has tacked on to the normal 20-year patent term (which until 1995 was only 17 years) an additional term of up to 5 years for the time it takes a pharmaceutical manufacturer to get a new drug approved by the Food and Drug Administration. In addition, the expiration of a pharmaceutical patent does not extinguish the patentee’s ability to obtain a higher price than the generic substitutes that come on line when his patent expires, because there may be substantial consumer and physician goodwill attached to the trademark of the patented drug—consumers, even physicians, may distrust generics and prefer the original brand even at a higher price. Indeed, there is evidence that when a patent expires the ex-patentee will actually increase price, ceding the low-price end of the market to the generics. His overall profits will be lower but may still be substantial. Against this it may be argued that the fact that the drug companies apparently do not have excess profits show they need every bit of patent protection they have. Not necessarily. Competition for a profit opportunity may transform expected profits into costs. Suppose the drug companies believe that the invention of some new drug will yield the successful inventor a $1 billion net profit. The prospect will induce heavy expenditures on being first (the aggregate expenditures may actually exceed $1 billion). The result is that none of the companies, or the industry as a whole, may have abnormal profits. Now suppose that as a result of a shortening of the patent term, the prospect for the successful inventor is for making only an $800 million profit. Less will be spent on the patent race. Yet consumers as a whole may be better off, because the investment saved may have greater value elsewhere in the economy. The entire patent “prize” goes to the firm that crosses the finish line first, and so a firm might spend a huge amount of money to beat its nearest rival by one day even though the value to the public of having the invention one day earlier might be negligible. This danger is greater, the bigger the prize. Shortening the patent term would reduce this potential waste by reducing the revenue from a patent; it would also reduce the transaction costs of licensing, because more inventions would be in the public domain.