There are two identifiable types of diseases in developing countries. Some, such as malaria, are specific to poor countries, but many others, such as cancer, have a high incidence in all countries. These differences give rise to quite distinct drug markets. In particular, for global diseases, pharmaceutical industry profits derived from having a monopoly over sales in poor countries make only a marginal contribution to total world-wide profit and therefore the incentives to invest in research. At the same time, even a small price increase due to such a monopoly in a poor country can greatly reduce the number of people able to purchase patented drugs and the welfare of those who do. This paper describes a policy that could improve on the current patent regime by acknowledging these differences in markets and what they imply for optimal patent protection. It allows protection to strengthen for diseases specific to developing countries where a clear argument can be made that some form of new incentives are warranted. At the same time, it effectively keeps protection at its current level in situations where increased profits are less likely to generate new innovation.
We are in the midst of a global expansion in the extent to which pharmaceutical innovations are protected by the patent system. Previously, most developing countries (LDCs) treated such innovations as non-patentable or at best offered only minimal protection for new manufacturing processes. Today, as the result of bilateral pressure and World Trade Organization membership requirements, they are in the process of implementing new patent laws that look very similar to those in the U.S. and Europe, granting full protection to all inventions in this area.
The public attention now focused on patents and the price of HIV/AIDS drugs in Africa has created an opening and a demand for creative thinking about ways to improve this new global system. Patent systems imply a tradeoff between prices and innovation which raises the question of whether the system be structured to elicit the same amount of innovation at a lower welfare cost? In answering this very basic question, it would be a mistake for international and domestic policy discussions to focus only on AIDS, despite the undoubted importance of this specific disease. The U.S. patent office granted over ten thousand patents related to pharmaceutical innovations in 1998, spanning thousands of diseases. No policy designed to address the current AIDS crisis is likely to be the best policy for the system as a whole, nor is it necessary to think in those terms. The AIDS epidemic is an international emergency of the first order. It can be treated as exceptional, and deserves its own policies.
An analysis of the implications of extending protection to additional countries is very closely analogous to that of granting protection for more years (see Nordhaus, 1968; Deardorff, 1992). Fundamental determinants of the optimal extent of protection are the degree to which the prospect of greater profits leads firms to increase research investment, and the degree to which further investment results in innovation of benefit to the public. These responses tend to decline at higher levels of R&D investment. Thus one can expect relatively more benefit from increasing protection where incentives are initially low.
From this perspective, it is important to recognize that there are two very different and identifiable types of drug markets. Some diseases are important worldwide, being found in both poor and rich countries, and therapies for such diseases have global markets. Others are more specific, with almost their entire market in the developing world (for example, malaria). Table 1 shows twenty diseases for which at least 99% of the global burden is in developing countries.
There has been almost no investment in the latter category outside of the public sector. Without protection in the developing world, there has been little prospect of profit anywhere and therefore little interest on the part of firms to invest in therapies for these diseases (see Lanjouw and Cockburn, 2001). The new regime may draw resources into the creation of drugs to prevent and treat diseases specific to poor countries. Of course, even with effective patent systems the group of LDC markets may not, by themselves, be very attractive given the prices that they can support. The goal of recent initiatives to “make a market” is to put more money into these poor country markets via a dedicated fund or tax credit to subsidize purchases of specified products (see Kremer, 2001, and World Bank, 1999, for details). This type of policy is appropriate for stimulating private investment in research on “Malaria”-type diseases: those which have small markets in the West, but which are of great importance in the developing world.
Consider, however, global diseases: those that are widespread in poor countries but also in rich countries. These diseases are the focus of the proposal described here. They have received less attention in development debates over intellectual property because they are not specific to LDCs. However, this does not mean that they are not important causes of disability and mortality amongst the poor. The first column of Table 2 indicates, for example, that cancer, heart disease, and diabetes together account for 16 percent of the total “disability adjusted life years” (DALYs) lost in a group of poorer countries with annual per-capita expenditure of just U.S. $1,250 (World Health Organization estimates. Similar percentages were found using mortality). This is four times higher than the share of their total burden coming from malaria. Not only are “rich country” diseases important in poor countries, they appear to cut across the income spectrum. Table 3, for example, presents data from a Pakistan health survey designed to gather information on the prevalence of strong risk factors for cardiovascular disease and cancer (see Pappas, et. al., 2001, for details). These data are unusual in having information from direct health examinations of the sampled individuals, rather than simply statements about disease incidence, together with at least some measure of household wealth. Fifty percent of the Pakistan population falls in the lowest defined asset owning group. The table shows that smoking among males is both widespread and significantly higher amongst the poor in Pakistan than the better off. Further, while those in the bottom half of the distribution have relatively lower rates of the risk factors associated with cardiovascular disease, the rates are still high with about a quarter suffering from hypertension and fifteen percent having high cholesterol. Other data exist giving self-reported, and therefore less reliable, disease incidence, but with better measures of household wealth.
Surveys in India, for example, found that of about 12,000 adult deaths in rural areas, 11% of those occurring in the lowest 20% of the all-India wealth distribution were ascribed to cancer or heart disease. This is well below the 35% rate in the highest quintile ascribed to these causes—but still a very substantial source of mortality (Deon Filmer, World Bank, personal communication). The evidence is not plentiful, but what evidence there is suggests that “rich country” diseases are widespread in poor countries, and that they are important among the poor and not just the relatively rich in those countries.
At the same time, almost all of the potential market for global diseases is found in the West. Return to Table 2. The second column gives rough measures of the relative market size in rich and poor countries based on disease incidence as measured by DALYs. The column figures are rich country DALYs divided by total DALYs for each disease, where rich and poor country DALYs are weighted by a rough estimate of their relative drug expenditure levels. On this measure, almost all of the market for cancer, heart disease and diabetes is in the rich countries. This is in stark contrast to malaria.
Tables 4 and 5 go directly to drug expenditure patterns. Like Table 2, the top panel of Table 4 suggests that poorer countries contribute little to total world expenditure on drugs for global diseases, but at the same time can be a significant major source of demand in some therapy areas (here parasitology). The bottom panel of Table 4 indicates, again, that a very significant share of the total spending by poor countries goes to global diseases even though their spending is of little importance in world demand for drugs for those diseases. Table 5 ranks selected major countries by their 1998 purchasing power parity adjusted per-capita GDP (those included are the largest LDC drug markets). We see each country’s share of total worldwide drug expenditure and an estimate of their individual shares of total worldwide spending on drugs for cardiovascular disease. These numbers are remarkably small. In particular, the subtotal in the middle of the table indicates that about 46% of the world’s population is found in countries representing less than 2% of total expenditure on drugs for cardiovascular disease.
In this paper I propose a policy that could improve on the current regime by acknowledging these differences in markets and what they imply for optimal patent protection. It allows protection to continue increasing worldwide in most areas of pharmaceutical innovation (as envisioned in TRIPs, the intellectual property section of the GATT treaty). In particular, and in contrast to other proposals being discussed such as indiscriminate compulsory licensing, it allows protection to strengthen for diseases specific to LDCs where there is a clear argument to be made that some form of new incentives are warranted. At the same time, it effectively keeps protection at its current level in situations where an increase in profits is less likely to generate new innovation. To do this, the policy requires inventors choose either to avail themselves of protection in the rich countries or, alternatively, in the poor countries, but not in both, whenever a patented product is for a global disease. Because the profit potential offered by rich country markets is far greater, firms will naturally relinquish those in poor countries. Thus the policy would lower the price of drugs for global diseases, and should be seen as a complement to policies that target poor-country specific diseases.
The following section suggests ways in which the policy could benefit both the world’s poor and research-based pharmaceutical firms. In particular, it addresses concerns over parallel imports and “low cost sources of supply”. Section III outlines a mechanism that gives a feasible way to present patentees with the desired choice between protection in either rich or poor country markets in the limited situations where their patents relate to products for specific global diseases. Economists and policy makers have been reluctant to differentiate protection across types of innovation despite the fact that there is a strong theoretical basis for doing so (and Article 27 of the GATT treaty explicitly requires non-discrimination). There are good reasons for this. The information needed to decide how best to differentiate is limited, and any differentiation must be on features both easily identified and hard to change or resources will be wasted as everyone tries to fit into the better class.
The mechanism described is simple to implement and has useful revelation and self-enforcement features that resolve these problems. Discussions of some of the important details are found in Section IV-VII. A brief discussion of some of the ways in which the proposed policy might be preferable to alternatives involving compulsory licensing and price control is in Section VIII.
II. Benefits for Firms and the World’s Poor
Firms have a legitimate concern about “low cost sources of supply” and seepage across borders, particularly into their major markets.
On the face of it, this proposal does not seem helpful in this regard since its intention is precisely to encourage low drug costs, in some areas, in poor countries. Firms may well object to it on these grounds. However, we must have “low cost sources” if we to have any hope of ensuring anything like the adequate availability of drugs to poorer people. The rich world will not supply levels of aid that would make purchases at U.S. prices feasible. Thus, the only appropriate response is to address the problem of seepage. If firms are confronted with substantial international arbitrage, then they will naturally respond by selling at a uniform price—one that is quite likely to be far higher than even than the monopoly prices appropriate to poor countries. They may decide not to launch drugs in the poorest countries altogether. To prevent this, efforts should be directed towards helping firms to separate markets. This is true regardless of whether the policy proposed here is implemented.
A first step in easing firms’ concern might be legislative confirmation that the U.S. does not have an international exhaustion of rights doctrine, in keeping with the more recent Federal Circuit Court interpretation of the law on exhaustion (see Adelman, et al., 1998). This would be a clear statement that holders of U.S. patents have the right to prevent products from coming into the U.S. from elsewhere, even if originally sold by their own licensees or subsidiaries.
The bigger issue, however, is the enforcement of rights in this area. Drugs are small and lightweight which makes it difficult to prevent products that have been sold cheaply in a country where consumers are poor from flowing back into markets where they are better off. The internet may greatly exacerbate this problem in the future. Consumers will be able to purchase drugs directly from around the world. Once LDC firms have developed sufficiently good reputations for quality that consumers feel comfortable with their products, one can easily imagine hundreds of thousands of packets crossing borders in separate envelopes in the regular post. Patentees will be hard pressed to identify such individual infringements and reluctant to enforce a separation of markets by suing their customers.
Internet sales also pose a safety threat to consumers. How is one to know that a web-based pharmacy is actually in North Carolina and not a counterfeit operation operating from overseas? (See http://www.fda.gov/ola/2000/internetsales.html for a discussion of current FDA concerns and efforts to combat this problem.)
It is difficult to see how the enforcement problems can be successfully resolved without better coordination and regulation of drugs at source. Thus the participation of poor countries in efforts to prevent illegal movements of drugs across borders will be key. The proposal described here is specifically designed to benefit developing countries, and in a way that would be very apparent to their populations. (This is contrast to the TRIPs agreement itself which, whatever its long run benefits in the form of new products, has engendered considerable resentment in LDCs.) It would seem reasonable to expect that they, in turn, make efforts to ensure that drugs priced for their consumers actually get to their populations and do not escape as exports to rich countries.
There are various ways that this might be done. One possible idea can be seen by analogy. The U.S. federal government taxes gasoline and diesel fuel at different rates depending on its intended use. This is difficult to enforce once distribution to users has occurred since the taxed and untaxed fuel looks the same. The solution has been to dye the untaxed fuel to make it more readily distinguishable.
Health authorities in all countries already specify features of drug appearance and packaging. One could ask poor countries that are candidates to be included under the policy to require that pharmaceuticals sold in their countries to be, for example, lime green. This would make it simpler to check bulk movements, and give consumers elsewhere a better chance of noticing that their drugs are not actually being manufactured in North Carolina, as they had supposed. There may be related and better ideas on how to use form and packaging to differentiate products—firms have considerable expertise in this area and their advice will be valuable here. But the point is clear. The fact that the policy encourages low prices in LDCs certainly implies the continued existence of “low cost sources of supply”. But the same policy also gives poor countries a positive reason to cooperate in resolving this looming, and extremely difficult, international enforcement problem. Seen from this perspective, the policy could help firms protect their more valuable markets.
In addition, the policy provides an alternative to untargeted policies now being suggested, such as across-the-board compulsory licensing of pharmaceutical patents or price control. Given the current climate of discontent with the new patent regime, and efforts to weaken it, some move away from the strongest level of protection will probably be necessary. This policy is be a controlled move designed to preserve incentives where they are most needed.
For precisely the same reasons the policy would benefit poor countries. They too stand to gain from the successful separation of markets. They too stand to gain from a policy that lowers prices on global diseases while at the same time maintaining incentives for firms to invest in products for diseases specific to poor countries. In addition, the policy involves only the patent laws and procedures in rich countries. Poor countries would continue to develop their patent systems fully and no questions would be raised about their compliance with WTO membership requirements. This would help shift international patent issues out of the realm of continuous dispute and put discussions on a more cooperative footing.
III. The Mechanism
I will first describe how the policy works in the simplest possible terms, leaving details to the discussion that follows. Assume, initially, that there are only:
- two countries, the U.S. (representing a set of rich countries) and India (representing a poor set);
- two diseases, Malaria and Cancer, the first representing a set with no U.S. market and the second a set with a very large U.S. market and a substantial but much smaller Indian market; and
- three companies, PharmaUS, CiplaIndia, USGeneric, where each represents a type of firm in the pharmaceutical market.
Bear in mind that patents are national in coverage. To obtain protection in France requires an application for a French patent. To obtain protection in Brazil requires an application for a Brazilian patent. Now, when an innovation is made in the U.S., the inventor is required to apply first for a U.S. patent. To make subsequent, foreign, applications the inventor is required to first obtain a “foreign filing license” from the US patent office (USPTO). This rule is in place for the purpose of protecting military secrets, and variants of it are found in patent regulations elsewhere.
The proposed policy is, very simply, to stipulate that when a patentee petitions for this license, he does so in the following form (exact language not important):
I, the undersigned, request a license to make foreign filings for patent no. X, with the understanding that this permission will not be used to restrict the sale or manufacture of drugs for “Cancer” in “India” by suing for patent infringement in “India”.
Again, obtaining a license is one of the steps that any U.S. patentee already must take in order to file abroad anywhere, including in Europe and Japan (see Section VII for further details). Requiring this declaration to obtain the license is the entire policy. A provision that already exists in the patent law is used to serve an entirely unanticipated purpose. The mechanism will work because other features of the patent law and pharmaceutical regulation can also be turned to serve this new purpose. These are discussed below.
Basic Outline of Why it Works
Consider the simplest situation. PharmaUS has a Cancer product protected by a single patent in the U.S. and in India. The company obtains marketing approval in both countries and sells the product. Now CiplaIndia (or USGeneric) enters the Indian market with its own version of the same product. PharmaUS can choose to do one of three things. First, it may continue to sell the product. Making this choice, it would need to lower its price to remain competitive with the new entrants. This is a strategy that multinationals have followed for decades in countries not offering them patent protection. On the other hand, PharmaUS might be uncomfortable selling at prices low enough to be competitive in India—perhaps because of international price comparisons—and it may choose to withdraw from the Indian market altogether. This is also a strategy that multinationals have followed. With this choice, PharmaUS would continue to exercise its rights in the U.S. market and entrants would supply the Indian market.
However, PharmaUS could make a third choice. The company has a valid patent in India, may sue CiplaIndia for infringement, and, if so, would win. Nothing prevents the company from choosing to protect its rights in India, on the basis of its patent there, in an Indian court, in exactly the same way that it would without the policy. But what happens then? At this point, either CiplaIndia or, more likely, USGeneric, can go to the USPTO and claim that, by attempting to stop CiplaIndia’s sales of the Cancer product in India, PharmaUS has rendered its U.S. patent unenforceable. This is so because, by taking this action, PharmaUS has falsified the declaration it made to the USPTO to obtain the foreign filing license. Patentees have a duty to deal with the PTO in good faith and failure in this regard is clear grounds for rendering a patent unenforceable.
Suppose now that the innovation had been for a Malaria product. Again PharmaUS could choose either to compete or to exit the market with the entry of CiplaIndia. Again its alternative is to sue for infringement. Now, however, the suit would give no grounds for rendering the U.S. patent unenforceable. The declaration made by PharmaUS to obtain its foreign filing license says nothing about Malaria.
So what is our result? In the case of a patent for a Cancer product, PharmaUS’s two choices are effectively between protecting its profits in the U.S. or in India, but not both, just as desired. It will not sue in India for infringements of Cancer product patents because it will not want to jeopardize its U.S. patents. Knowing this, CiplaIndia will enter the market and prices in India will fall. In the case of a patent for a Malaria product, PharmaUS’s two choices are effectively between protection in the U.S. or protection in both the U.S. and India. It will sue in India for infringements of Malaria product patents. Knowing this, CiplaIndia will avoid the suit by not entering the market—retaining the incentive for investment in Malaria products.
One might say, “With this policy PharmaUS may not even bother to get a patent in India for Cancer.” This is true and it is fine. One of two strategies will be followed. Either PharmaUS will continue to market its patented Cancer product in India, on a competitive basis, or it will leave the market to CiplaIndia and USGeneric. Both strategies have been followed by multinationals over the past decades in countries that have not granted them patent protection. Both LDC firms and developed country generics manufacturers have shown themselves to be adept at rapid imitation and entry. This was, after all, the point of pressing for TRIPs in the first place, as well as domestic legislation to control generic entry. Lanjouw (1998) presents evidence indicating that, over the past two decades, major patent drugs arrived on the Indian market typically within 7 years of their world launch, and often much sooner. Watal (2000) suggests an increase in arrival speed. For ten drugs launched in the U.S. after 1985, she finds an average time lag to availability in India of just two years. Thus, there does not appear to be any reason to be concerned about which strategy the patentee chooses to follow.
The mechanism is designed to be triggered by a lawsuit. Why do we go this route? Because when infringement suits are filed to prevent the sale of a product it is on the basis of a set of patents. In order to be successful in prosecuting his suit, the patent owning firm has an incentive to correctly announce which patents it believes best protect the product in question. This resolves the otherwise intractable problem of how to identify the use of particular patents. It allows the mechanism work without a bevy of scientists trying to identify patents that might someday be for Cancer.
Advantages of the Policy
- It does not contravene existing treaties (Paris Convention, Article 4bis; the TRIPs component of GATT, Article 27).
- It can be implemented unilaterally, although it would be most effective and acceptable to all parties if the EU, Japan and the U.S. were to move together. (Note: for simplicity, I will comment below as though only the U.S. implemented the policy. The comments would be equally true for other rich countries and one could read “France” or “Japan” in place of “U.S.” if those countries were to participate.)
- It does not require any changes whatsoever to new LDC patent systems or the development of their enforcement procedures. In fact, better functioning patent office and court systems in the LDCs will only improve the working of this policy. At a time when there is concern to nurture budding TRIPs compliance it seems a great advantage of this mechanism that it will not in any way “muddy the waters”.
- As detailed below, the mechanism relies almost entirely on the quality and reliability of U.S. institutions and not on those in the LDCs themselves.
- This policy would be fully controlled by the U.S. government. This is in contrast to the sanctioning of compulsory licensing by LDC governments, where pressure by local interests to expand coverage to all diseases will be difficult for the domestic government to resist.
- The mechanism does not require information that is clearly not available. In particular, and crucially, it does not require that patents be examined and identified as covering innovations for a particular disease. Such a task would be infeasible. Even ignoring the expense, at any moment in time the patent owner himself may not know the future uses of a patented innovation. The policy mechanism induces firms to volunteer the link between patents and products when the information becomes known and only as necessary.
- No one is told what to do. Incentives are aligned to make use of the greater information that firms have about the relative size of global markets for different products. They behave as desired without outside control or monitoring.
- Because it uses existing institutions and procedures, is largely self-monitoring and does not require the collection of information for each patent, the policy would cost very little to administer and enforce. One potentially important implication is that this policy need not be seen as an alternative to other policies within the constraints of fixed health or development budgets.
As noted in Section III, a case filing identifies the Indian patents that protect a particular product. This section considers the two remaining links that need to be made.
Linking Products to Diseases
One of the stated advantages of the mechanism is its reliance on U.S. institutions. But it is triggered by a court case in India. This may seem surprising. However, it is the filing of a suit that is the trigger—the effectiveness of the policy does not rely in any way on the subsequent legal proceedings in India. Using the Indian case for this purpose does raise two issues, however. First there must be a clear procedure for determining, on the basis of U.S. institutions, whether the Indian product which is the subject of the suit corresponds to a particular disease. CiplaIndia or USGeneric will always have an incentive to claim that a disputed product is for Cancer in order to render unenforceable the U.S. patent of PharmaUS, while the latter will claim all products are for Malaria.
I suggest the following. All products marketed in the U.S. are approved by the FDA for specific indications.
To render unenforceable PharmaUS’s patent, USGeneric must take the Indian product and apply to the USFDA for an abbreviated new drug approval (ANDA). In this, it would claim the Indian product’s equivalence to one already marketed in the U.S. with a Cancer indication. This procedure is exactly the same as that already followed for any generic on the expiry of a patented product so our own generic companies are well versed in following it through. If the USFDA issues tentative approval, or a preliminary letter of bioequivalency, the case that the Indian product is for Cancer is made and the U.S. patent rendered unenforceable.
At this point USGeneric or CiplaIndia can, and will, request final marketing approval from the USFDA, since obtaining access to the U.S. market was the point of rendering PharmaUS’s patent unenforceable. The bioequivalence report is the basis for that approval. Thus there is no net increase in resources expended by either the companies or the government as a result of using the USFDA ANDA process for our purpose. It also means that the FDA has a serious interest in the quality of the bioequivalence report as it has direct implications for the integrity of the U.S. system of safety regulation.
Linking Patents to Patents
The second issue that arises is that the Indian patents supporting the suit need to be linked to their U.S. equivalents. Fortunately, this is a standard output of international patent procedures. Having first filed in the U.S., a subsequent Indian application typically refers back to the U.S. application to establish the owner’s global priority over the innovation and the time limit for related foreign filings. The global links between patents covering the same innovation that are exposed by this process can be found in publicly available databases.
V. More Complex Settings
The simple situation described in Section III, where a single patent protects a single product, is rare. We next consider how the mechanism would work in more complex settings: with multiple uses of a single patent; multiple patents on a single pr