A Purchase Fund for New Vaccines: Rationale and a Proposed Design *

Transcription

1 Executive Summary A Purchase Fund for New Vaccines: Rationale and a Proposed Design * Michael Kremer ** Harvard University and Brookings Institution This Draft: October 4, 1999 Malaria, tuberculosis, and HIV kill approximately 5 million people a year. Recent scientific advances give ground for optimism about the scientific prospects for vaccine development. Yet vaccine research remains minimal relative to the burden of these diseases. Public funding for malaria research, for example, amounts to about $84 million annually, or $77 per malaria fatality, compared to expenditures of $789 per fatality for asthma. This paper explores the potential role of a vaccine purchase fund that would pre-commit to purchase vaccines for these diseases and distribute them in developing countries. Such a fund would both create incentives for vaccine research and ensure that if vaccines were developed, they would reach people in developing countries. Public, global action is needed to spur vaccine development and ensure that if vaccines are developed, they are widely distributed. Most countries subsidize vaccines. These public subsidies are justified because people who take vaccines not only benefit themselves, but also benefit others by interfering with the chain of disease transmission. Moreover, if vaccines were sold under patent at a monopoly price, the market would be inefficiently small. Large public vaccine purchases offer the opportunity to bring down the price per dose by enlarging the market, potentially benefiting both vaccine producers and consumers. Global coordination is needed because otherwise each individual country has an incentive to free-ride on the research incentives provided by other countries. Finally, action to encourage vaccine development is needed now, because once vaccine developers have invested large amounts in developing a vaccine, governments face the temptation to use the implicit threat of price controls or compulsory licensing to bargain for prices that cover manufacturing, but not research costs. Anticipation of this discourages vaccine research. Programs to encourage vaccine research can be divided into push programs that pay for research inputs, for example, through grants to researchers, and pull programs that pay for a viable vaccine. Push programs are attractive for funding basic research, but the development of the biotech venture capital industry has greatly broadened the scope for pull programs in stimulating the later, more applied stages of vaccine research. Researchers with strong financial * I am grateful to Daron Acemoglu, Philippe Aghion, Martha Ainsworth, Susan Athey, Sara Ellison, Rachel Glennerster, Eugene Kandel, Jane Kim, Sendhil Mullainathan, Margaret Ronald, Andrew Segal, Scott Stern, and especially Jeffrey Sachs and Amar Hamoudi for suggestions. This paper is part of a Harvard Center for International Development project on vaccines conducted jointly with Jeffrey Sachs. I thank the National Science Foundation and the MacArthur Foundation for financial support. The views expressed in this paper are my own, and not theirs. ** Department of Economics, Littauer 207, Harvard University, Cambridge, MA 02138; mkremer@fas.harvard.edu.

2 incentives to develop a marketable vaccine are likely to focus more on this task, rather than other goals, such as publishing academic articles. Moreover, paying for vaccines, rather than research costs, gives pharmaceutical firms and scientists strong incentives to self-select only those research projects that have a reasonable chance of leading to a vaccine. With pull programs, the public pays only if a vaccine is actually developed. The paper also proposes a design for a vaccine purchase fund. Comprehensively specifying vaccine eligibility and pricing in advance is likely to be impossible, since vaccine quality depends on many different features, including side effects, efficacy against different variants of the disease, and the need for boosters. Fund administrators will therefore need to use some discretion in determining eligibility of candidate vaccines for purchase, as well as the price that would be paid for these vaccines. Such discretion creates a temptation for the fund to take advantage of vaccine developers by refusing to pay a price adequate to cover research costs. To limit this temptation, the committee making eligibility and pricing decisions should include industry representatives as well as representatives from the scientific and public health community, and it should be insulated from political pressures through long terms. To the extent possible, decisions should be decentralized, and the fund s procedure should be structured in advance. I propose the following three-step process: 1) Candidate vaccines would have to be approved by some regulatory agency, such as the U.S. FDA; 2) Candidate vaccines would also have to meet a market test developing countries wishing to purchase vaccines using fund resources would have to contribute a 20% co-payment, and would be required to draw down an account that they would hold within the fund; 3) In addition to the base price, bonus payments for vaccines should be linked to vaccine effectiveness, perhaps as measured by the estimated number of lives or DALYs saved by the vaccine. In order to avoid offering a price for vaccines that is either too low to spur adequate research or unnecessarily high, the fund should specify both a base price and a schedule according to which the vaccine price would rise over time until a vaccine had been developed. This procedure mimics auctions, which are often an efficient procurement method when costs are unknown. To avoid creating an incentive for vaccine developers to withhold vaccines from the market while waiting for the price to rise, the price should rise at a rate only slightly greater than the interest rate. The number of children born each year in low- and middle-income countries with high prevalence of HIV, malaria, and tuberculosis is 103 million, 52.6 million, and 81 million respectively, so purchasing vaccine at $10 per immunized child would cost approximately $500 million to $1 billion annually, depending on the disease. Further expenditures would be needed initially to vaccinate the backlog of unvaccinated people. A private foundation could combine push and pull programs by spending 5% of its assets each year on grants for basic research, while pledging its principal towards a vaccine purchase fund.

3 Vaccine Purchase Fund Kremer 1 Malaria, tuberculosis, and HIV kill an estimated 5 million people each year. In the past fifty years, these diseases have claimed six times as many lives as all wars [WHO, 1999b]. Virtually all malaria cases, and more than 95% of new HIV and tuberculosis cases, are in developing countries [WHO, 1994; UNAIDS, 1998; WHO, 1998], although the spread of drug resistance threatens increased infection and death in developed countries as well. Almost 90% of malaria cases and 70% of new HIV infections occur in sub-saharan Africa [WHO, 1999a; UNAIDS, 1998]. 1 Recent scientific advances have increased the potential for development of effective vaccines. Yet, despite the huge toll of these diseases, they are the subjects of very little vaccine research. This is both because these diseases primarily affect poor countries, and because vaccine markets are severely distorted. One way to create incentives for vaccine development would be to establish a vaccine purchase fund which would pre-commit to purchase vaccines and provide them to developing countries at an affordable price. This paper sets forth the economic rationale for a purchase fund for new vaccines and proposes methods for establishing eligibility of candidate vaccines for purchase and for setting a purchase price. Section 1 of this paper provides background information on malaria, HIV, and tuberculosis, and reviews the current state of scientific progress towards vaccines. Section 2 discusses the economic rationale for pre-commitments to purchase vaccines. Sub-section 2.1 argues that public, global action is needed to support the development and wide distribution of vaccines, and that this action should take place now, before vaccines have been developed. Almost all countries subsidize vaccination, and these subsidies are justified for three 1 Although all these numbers come from the UN and the UN-affiliated World Health Organization, they may not have been estimated using consistent methods.

4 Vaccine Purchase Fund Kremer 2 reasons. First, individuals have inadequate incentives to take vaccines, since those who take vaccines benefit not only themselves, but also benefit others by breaking the cycle of infection. Second, the chief beneficiaries of vaccination are often children, and public action to protect the welfare of children is warranted. Finally, monopoly pricing of vaccines would lead to an inefficiently small market. Large public purchases could make both vaccine producers and consumers better off than under monopoly pricing by reducing the cost per dose, but expanding the market. Global action is needed because otherwise each country would have an incentive to free ride off the research incentives provided by other countries and take advantage of a vaccine once it was developed. Action should be taken now, rather than after a vaccine is developed, because once a vaccine is developed, governments and international organizations will face a temptation to negotiate for the vaccine prices which cover manufacturing costs, but not research costs, and the anticipation of this deters vaccine research. Sub-section 2.2 examines the appropriate role of various mechanisms for encouraging research. It argues that push programs, which pay for research inputs, are appropriate for basic scientific research. However, the development of the biotech venture capital industry has increased the scope for pull programs, which pay for successful research output, for example, by promising to purchase vaccines if they are developed. Such programs are likely to have an important role in the later, applied work of actual vaccine development. Pull programs provide strong financial incentives for researchers to select appropriate research projects and to focus on developing a marketable vaccine, rather than pursuing more diffuse scientific goals. With a pull program, the public pays nothing unless an actual vaccine is developed. Section 3 discusses how a vaccine purchase fund could work. It examines how a system could be designed with enough flexibility to allow purchasing decisions to depend on vaccine

5 Vaccine Purchase Fund Kremer 3 quality, while still providing enough commitment to potential vaccine developers to induce them to invest in research. Section 3.1 argues that after candidate vaccines have been developed and tested, some discretion will be needed to determine eligibility of vaccines for purchase and the amount that the vaccine purchase fund would pay for these vaccines. It is impossible to precommit to a comprehensive set of eligibility and payment rules for vaccines that have not yet been developed. Given that the fund will need to use discretion in determining vaccine eligibility and pricing, it is important to structure the decision-making process so as to limit temptations for the fund to take advantage of vaccine developers by insisting on low prices once developers have already sunk millions of dollars into research. If potential vaccine developers anticipate that this will happen, they will not invest at all. Section 3.2 outlines a system in which potential vaccines would first have to receive regulatory approval and meet some other minimal eligibility requirements. They would then be subject to a market test: nations wishing to purchase vaccines would need to provide a 10-20% co-payment and spend down an account assigned to them within the fund. Since these requirements are fairly minimal, an important part of compensation for vaccine developers should take the form of bonus payments tied to vaccine effectiveness. Section 3.3 discusses the tradeoffs in setting a vaccine price when the cost of developing a vaccine is unknown. It suggests that one way to avoid paying either too little or too much for a vaccine would be to specify a schedule according to which the price will increase over time until an appropriate vaccine is invented. This procedure mimics auctions, which have been found to be efficient procurement methods in many situations in which the purchaser does not have full information about the cost of the product, and there is insufficient competition to simply observe a market price. Section 3.4 reviews a number of other design issues that would arise with vaccine purchase pre-commitments.

6 Vaccine Purchase Fund Kremer 4 The conclusion briefly considers the potential cost of vaccine purchase pre-commitments, the politics of such a program, and the role that could be played by private foundations in helping to establish a vaccine purchase fund. Appendix 1 argues that a vaccine purchase fund is superior to other pull mechanisms designed to increase incentives for vaccine research. Rewarding vaccine developers with extensions of patents on other pharmaceuticals is an inefficient way of financing vaccine research. Cash prizes for research are economically similar to a vaccine purchase fund, but are likely to be politically less attractive, and thus provide less credible commitment to vaccine developers. Research tournaments are not well suited to a context in which there is no guarantee that a vaccine could be developed within a fixed time period. Paying more than necessary for existing vaccines is an expensive and potentially ineffective way of encouraging research in future vaccines. Appendix 2 uses techniques from the economic theory of auctions to examine how vaccine prices should be determined. The idea of an HIV Vaccine Purchase Fund was proposed by a coalition of organizations coordinated by the International Aids Vaccine Initiative at the 1997 Denver G8 summit. Since then, the idea has been explored by the World Bank Vaccine Task Force [World Bank, 1999]. Kremer and Sachs [1999] and Sachs [1999] have advocated the establishment of a fund in the popular press. This paper draws on a number of earlier papers, including the vaccine work of Batson [1999], Mercer Management Consulting [1998], and Milstien and Batson [1994], and the broader academic literature on research incentives, including Dutton [1984], Green and Scotchmer [1982], Guell and Fischbaum [1995], Johnston and Zeckhauser [1991], Lanjouw and Cockburn [1999], Lichtmann [1997], Rogerson [1994], Romer [1993], Russell [1998], Scotchmer [1997], Shavell and van Ypserle [1998], Taylor [1995], and Wright [1983]. The contribution of this paper lies in setting forth the economic rationale for pre-committing to

7 Vaccine Purchase Fund Kremer 5 purchase vaccines, and in proposing procedures for determining eligibility and pricing. 1. Background on Malaria, HIV, and Tuberculosis This section reviews the burden of the major infectious diseases and discusses scientific prospects for vaccines The Burden of Malaria, HIV, and Tuberculosis UPDATE Estimates of the burden of infectious disease vary widely, but it is clear that the burden is huge. The World Health Organization estimates that each year there are 300 million clinical cases of malaria, and 1.1 million deaths from malaria [WHO, 1999a]. Malaria is particularly likely to kill children and pregnant women. Resistance is spreading to the major drugs used for treating malaria and for providing short-term protection to travelers. Each year, approximately 8 million people contract tuberculosis disease, and 1.9 million people die from it. More than 98% of these deaths occur in developing countries [WHO, 1999a]. With up to 17% of tuberculosis infections resistant to all five major anti-tubercular drugs, the spread of resistance poses a threat to developed as well as developing countries [WHO, 1997]. The existing BCG vaccine, which is distributed widely, provides only temporary and imperfect protection against tuberculosis. 2 More than 33 million people are infected with HIV worldwide, over 95% of whom live in developing countries. In 1998, about 2.5 million people died of AIDS, 80% in sub-saharan 2 The vaccine seems much more effective in some places than others: trials in Britain suggest effectiveness up to 80%, while those in the southern United States and southern India suggest close to zero effectiveness.

8 Vaccine Purchase Fund Kremer 6 Africa, and approximately 5.8 million people were newly infected, 70% in sub-saharan Africa [WHO, 1999a; UNAIDS, 1998]. New life-extending HIV treatments are far too expensive for most individuals and governments in low-income countries. Since people with compromised immune systems are especially vulnerable to tuberculosis, the spread of HIV is contributing to the spread of tuberculosis The Prospects for Vaccines Vaccines have proved effective against many other infectious diseases, and in the long run, they are likely to be the most effective and sustainable way to fight malaria, tuberculosis, and HIV. In the 1970 s, smallpox was eradicated through a vaccination program. Currently, almost 80% of the 130 million children born worldwide each year are immunized with vaccines against diphtheria, pertussis, tetanus, polio, and measles, and with the BCG vaccine [WHO and UNICEF, 1996]. The question of whether vaccines could be developed against malaria, tuberculosis, and HIV remains open, but there is reason to be optimistic. A recent National Academy of Sciences report [1996] concludes that development of a malaria vaccine is scientifically feasible. Candidate vaccines have been shown to protect against malaria in several rodent and primate models. Moreover, the human immune system can be primed against natural malaria infection. People who survive beyond childhood in malaria endemic areas obtain limited immunity which protects them against severe malaria, although not against parasitemia and milder illness. Since vaccines prime the immune system by mimicking natural infection, vaccines may similarly provide protection against severe disease. Recently, candidate vaccines have been shown to induce protection against tuberculosis

9 Vaccine Purchase Fund Kremer 7 infection in animal models. The example of the existing BCG vaccine suggests that the human immune system can be primed against tuberculosis infection. It may be particularly difficult to stimulate an immune response which can defeat AIDS, since the disease overwhelms the immune response in the majority of patients. However, recent evidence suggests that some people never contract HIV despite continuous exposure for long periods of time, and that others do not contract AIDS despite being infected with HIV for many years. This may provide a model of protection on which to base HIV vaccine development efforts. A number of candidate HIV vaccines protect monkeys against infection and induce immune responses in humans. Nonetheless, formidable scientific and technological obstacles remain in the way of development of malaria, tuberculosis, and HIV vaccines. All three diseases have many variants and evolve rapidly, making it difficult to design vaccines which are effective against all variants and remain effective for a long time. Moreover, scientists lack information on the specific immune response mechanisms against each of these diseases and the correlates of protection variables for which scientists can test in experiments to determine if a vaccine is likely to be effective. Recent advances in immunology, biochemistry, and cloning have given scientists new tools to understand the immune response to these diseases, detect possible correlates of protection, and develop better animal models. Genetic sequencing of the organisms causing tuberculosis, HIV, and malaria is either complete or far advanced, which should help scientists rationally design vaccines against these diseases. This may help scientists create vaccines which target many different antigens, and thus are more effective in the face of genetic diversity. 2. The Need For Public, Global Action to Support the Development and Distribution of

10 Vaccine Purchase Fund Kremer 8 Vaccines This sub-section argues that public, global action is needed now to encourage the development and wide distribution of vaccines. In the vast majority of countries, governments buy vaccines on behalf of the population, and then distribute them at a low or zero price. There are several good reasons for these government subsidies. First, individuals who take vaccines not only benefit themselves, but also help break the chain of disease transmission, thus benefiting the rest of the population. Second, the chief beneficiaries of vaccines are often children, who are not able to make decisions on their own behalf. Third, consumers may not immediately comprehend the effectiveness of a vaccine, as the benefits of vaccines, unlike most drugs, are not evident until considerably after they are is taken. Finally, because vaccines require high research costs, but typically have low marginal production costs, if vaccines are produced under patent and sold at monopoly price, production will be inefficiently low. Large government purchases may benefit both vaccine producers and consumers relative to monopoly pricing, because they can increase the market for vaccines, and thus bring down the cost per dose. To see why such programs can benefit everyone, even taking into account the need to raise tax revenue to pay for the program, note that those consumers who would have been willing to pay the monopoly price are better off, as long as the tax on them is less than the monopoly price. This is feasible, since the price per dose with a large government order is less than the monopoly price. The vaccine developer can be made better off, because the larger market created by government purchases more than makes up for the lower price per dose. The consumers who valued the vaccine at more than the production costs, but less than the monopoly price, can also be made better off, as long as the value they place on the vaccine is greater than the increase in taxes. Not only is there a strong case for public purchases of vaccines, there is also a strong case

11 Vaccine Purchase Fund Kremer 9 for global action to encourage vaccine research. In the absence of global coordination, each country would have an incentive to free ride off the research financed by other countries while providing little or no R&D financing and intellectual property rights protection itself. Individual countries could simply rely on the intellectual property rights protection provided by other countries to stimulate vaccine research, but bargain with vaccine developers to obtain the lowest possible price for themselves. This problem is particularly severe for countries that are economically small enough that they reap only a small fraction of the worldwide benefits of research. This may help explain why pharmaceutical prices are controlled at prices approximately ½ of United States levels in the European Union, while in Japan, they are controlled at ¼ of U. S. levels [Robbins and Freeman, 1988]. The world s three leading infectious diseases primarily affect small developing countries, which have even less reason to internalize the benefits of drug development than the European Union or Japan. Prices for vaccines in developing countries are often a tiny fraction of U. S. prices. Historically, developing countries have not provided much protection for intellectual property rights (IPR) for pharmaceuticals. Several large developing countries, including India and Brazil, have recently agreed to enhance intellectual property rights for pharmaceuticals, but only under intense trade pressure from the United States. It remains to be seen whether the promised IPR protection will be enforced, and many pharmaceutical firms are skeptical. South Africa has threatened to impose compulsory licensing of HIV treatments [Wall Street Journal, 1999]. Weak IPR protection makes it harder for developers of vaccines against diseases affecting developing countries to recoup their investment. Note that if IPR are enforced globally, the same arguments which suggest that national vaccine purchases are more efficient than individual purchases under monopoly pricing also

12 Vaccine Purchase Fund Kremer 10 suggest that global purchases are potentially more efficient than national purchases under monopoly pricing to individual countries. If vaccine developers charge a monopoly price, some countries could not afford to purchase the vaccine. The vaccine developer, the poor countries, and the rich countries all could potentially be better off if vaccines were purchased in sufficient quantity for the entire world at a price below the monopoly price. As long as the rich countries paid no more than the monopoly price they would have paid otherwise, and the poor countries pay less than the amount at which they value the vaccine, but more than the actual production cost, everyone could be made better off than they would under monopoly pricing. (Alternatively, the market could reach efficient size if vaccine developers charged each person (or nation) a separate price based on what they were willing to pay. In the pharmaceutical industry, this is known as tiered pricing, and economists typically call it price discrimination. In fact, pharmaceutical firms do charge different prices to different countries. However, tiered pricing is limited by the possibility of resale and by political considerations. Politically, it may be difficult for pharmaceutical firms to justify charging high prices in one country when they charge lower prices in another country. 3 A global purchase fund, with national contributions based on GNP, may be more sustainable politically.) Finally, action to encourage the development and wide distribution of vaccines should not just be public and global, but should take place now, before vaccines have been developed. This is because vaccine research is very expensive, but once vaccines have been invented, they can usually be manufactured at low cost. After vaccines have already been invented, governments 3 At a congressional hearing, Senator Paula Hawkins asked a major vaccine manufacturer, How can you justify charging nearly three times as much to the United States government as you did to foreign countries, and then the next year again also substantially below federal U.S. prices? [Mitchell, Philipose, and Sanford, 1993]. Since then, U.S. manufacturers have not submitted bids to UNICEF to supply vaccines. When Clinton announced his childhood immunization initiative in 1993, he said, I cannot believe that anyone seriously believes that America should manufacture vaccines for the world, sell them cheaper in foreign countries, and immunize fewer kids as a percentage

13 Vaccine Purchase Fund Kremer 11 therefore have a temptation to try to obtain them at a price that would cover manufacturing costs, but not research costs. Governments are in a strong bargaining position at this point, because they are monopsonistic vaccine purchasers, because they regulate vaccines, and because they can threaten to violate intellectual property rights of the vaccine developer, for example, by imposing compulsory licensing of the vaccine to alternative suppliers without paying appropriate royalties. Thus, while in theory government purchases of vaccines could make both vaccine producers and consumers better off, in practice, they are often used as a vehicle to make transfers from vaccine producers to consumers. Since potential researchers anticipate this redistribution, they invest less in research than they would otherwise. Due to all these factors, very little vaccine research has been conducted over the last twenty years. Recent scientific advances have increased vaccine research, but it is still a fraction of overall pharmaceutical research. Of $24 billion in worldwide annual pharmaceutical research expected in 1999, less than $2 billion goes to vaccines and biologicals. Total pharmaceutical sales in 1997 were $104.6 billion, of which $4.8 billion were vaccines and biologicals. Despite the likely scientific potential for vaccines, current vaccine research is paltry relative to the burden of malaria, tuberculosis, and HIV. According to a Wellcome Trust study, public and non-profit malaria research amounted to about $84 million in fiscal year 1993, with vaccine research making up only a small fraction of the total. This amounts to about $77 per malaria fatality, compared with research expenditures of up to $789 per fatality for asthma. Private sector spending on malaria is unknown, but is generally considered to be lower than public spending. Less is known about total expenditures on tuberculosis research, but the United States National Institutes of Health, one of the world s leading funders of basic research, has been of the population than any nation in this hemisphere but Bolivia and Haiti. [Mitchell, Philipose, and Sanford, 1993].

14 Vaccine Purchase Fund Kremer 12 spending around $65 million per year on tuberculosis research, compared with $2.7 billion on cancer research [NIH, 1999]. Applied AIDS research is overwhelmingly oriented towards treatments which would be appropriate for people with AIDS in the United States, rather than towards vaccines appropriate for less developed countries. The multi-drug treatments for HIV are not feasible for poor countries, since they cost $10,000-16,000 a year [PhRMA, 1999] and need to be taken in perpetuity according to a precise protocol, which is unlikely to be realistically achievable in developing countries. To the extent that vaccine research is conducted, it is oriented towards the HIV strains common in rich countries. Most candidate HIV vaccines tested worldwide are based on clade B, the strain of the virus transmitted in the United States, Europe, Australia, and Southeast Asia, rather than the clades most common in Africa, where two-thirds of new infections occur. It is uncertain whether a vaccine developed for one clade would protect against other clades.. A very crude estimate suggests that the social benefits of vaccines are likely to exceed the private benefits obtained by vaccine developers by many orders of magnitude. Since potential vaccine developers will only consider these private benefits in setting their research budgets, incentives for vaccine research are far too small. For example, consider the benefits from a vaccine which reduced malaria mortality and morbidity by 80%. A standard way to measure the burden of the disease is the number of Disability Adjusted Life Years (DALYs) lost to the disease. The WHO recently estimated that malaria creates 39.3 million DALYs per year [WHO, 1999a]. Assuming that the marginal manufacturing and delivery cost for this vaccine would be about $1.50 per child vaccinated, 7 and that the 52.6 million children born annually in low-income and 7 The addition of hepatitis B and yellow fever vaccines (which are relatively expensive) increased the $15 cost of the Expanded Program of Immunization by 15%. Almost all of this constitutes manufacturing and distribution costs,

15 Vaccine Purchase Fund Kremer 13 lower-middle-income countries with high enough prevalence to make vaccination cost-effective were inoculated, mass administration of this vaccine would cost about $3.51 per DALY saved. In its 1993 World Development Report, the World Bank defined a highly cost-effective health intervention as one which cost less than $100 per DALY saved. (A moderately cost-effective intervention was one which cost less than $1,000 per DALY saved). Even under the conservative valuation of $100 per DALY saved, the annual net social benefit of this vaccine would be $96.49 per DALY saved, equivalent to $2.17 billion, or about $41.29 per immunized child. At $1,000 per DALY, a malaria vaccine s benefits would be approximately $22.6 billion annually. Note that these figures do not take into account the potential economic benefits of reducing malaria prevalence outlined in Gallup and Sachs [1998]. The revenues generated from a malaria vaccine would almost certainly be far less than this social benefit, limiting the incentives to invest in vaccine research. To give some indication of this, while 23% of vaccine doses produced globally are sold to UNICEF [Russell, 1997], these doses generate only $240 million of revenue 6% of the $4 billion annual global vaccine revenue. The total developing country market for childhood vaccines is $200 million annually. The combined cost of the six vaccines in the standard Expanded Program on Immunization (EPI) package is about $0.50 per dose [Robbins and Freeman, 1988]. Of course, a vaccine under patent would likely generate greater revenues. However, when the hepatitis B vaccine was first introduced and priced at $30 per dose, it was not used widely in developing countries. [Muraskin 1995; Galambos 1995]. 4 rather than profits for pharmaceutical firms, since those vaccines are no longer under patent and are subject to tiered pricing. It seems reasonable to guess that the manufacturing and distribution cost of adding one more vaccine would be 8% of $15, or $ Even if the entire pharmaceutical budget in many African countries went to malaria vaccines, the benefit to a vaccine developer would be far less than the social benefit. The social benefits of a malaria vaccine in 25 African countries for which health expenditure data are available would be $648 million annually, even using conservative assumptions on

16 Vaccine Purchase Fund Kremer 14 The economic rationale for spending foreign aid funds on vaccine research seems extremely strong. Vaccine research is an international public good, whereas some aid funds are currently spent on programs which are not even public goods within a country, let alone international public goods. The next section discusses the roles of various methods of encouraging such research The Roles of Push and Pull Mechanisms in Encouraging Vaccine Research The literature on vaccine research distinguishes between push and pull programs. Push programs provide direct funding for vaccine research, for example, through grants to academics, in-house research in government laboratories, or public equity investments in vaccine research. Pull mechanisms increase rewards for development of a vaccine, for example, by promising to purchase a vaccine if it is developed. Roughly, the distinction is between paying for research inputs and paying for research outputs. This section argues that push programs are likely to be well suited to basic research, but that the development of the biotech venture capital industry has made pull programs attractive for the later stages of the vaccine development process. Historically, there were relatively few sources of finance available for commercial pharmaceutical research. Programs designed to encourage such research therefore had to finance research inputs ahead of time rather than simply paying for research outputs and relying on DALYs. (This number needs to be updated, however.) These countries spent $1.7 billion on health in 1996 [World Bank, 1998]. Generally, pharmaceuticals account for between 10-30% of recurrent costs in health [World Bank, 1993]. This would amount to $ million. (Of course, pharmaceutical expenditures could change in response to vaccine development.)

17 Vaccine Purchase Fund Kremer 15 investors to provide finance. 5 However, with the rise of the biotech venture capital industry in recent years, it is now much easier for scientists with potentially lucrative ideas to obtain finance. This suggests that there may be more scope for pull mechanisms today than twenty or even thirty years ago. It may be necessary to reevaluate methods of supporting research in light of the changed financing environment. For some types of research, the traditional push approach continues to be attractive. There are several reasons why it may be appropriate to finance basic research with grants, rather than rewards linked to research outcomes. First, basic research is inherently risky and it may therefore be best to partially insure researchers against the risk of failure. Second, the difficulty of evaluating the quality of basic research output makes paying for research output more problematic. It is not simply a matter of testing if a vaccine works, or a product sells. Finally, basic research, by definition, is as valuable for the information it provides to other researchers as for the products it produces. With strong financial incentives tied to product development, researchers would be more inclined to keep their research results private longer in order to have an advantage in the next stage of research. Grant-funded academics and scientists in government laboratories have career incentives to publish their results quickly. The strong financial incentives associated with pull programs are more important in the later, more applied stages of vaccine production. Scientists are likely to be much better informed about their work and about the prospects for vaccines than research administrators, and this creates moral hazard and adverse selection problems in the absence of strong financial incentives. In particular, many academic and government researchers have career incentives and intellectual 5 Some vaccines have been developed primarily in the public sector, and only later licensed out to the private sector for production. For example, the meningococcal meningitis vaccine was developed almost entirely at the Walter Reed Army Institute of Research, and a hepatitis B vaccine was designed by the Hepatitis B task force [Muraskin, 1995]. However, it is not clear that the development of these vaccines in the public sector reflects so much the suitability of

18 Vaccine Purchase Fund Kremer 16 interests that orient them to fundamental science. The later, more applied portions of the vaccine development process include some activities that are not particularly interesting intellectually, but are quite expensive. For example, techniques for developing sufficient quantities of candidate vaccines in sufficient purity for clinical trials must be developed and implemented. Animal models for the disease must be created. Vaccine trials in the field must be conducted. Nobody wins a Nobel Prize for these activities, even though they are important steps in vaccine development. In the absence of strong financial incentives, some researchers may therefore be tempted to use funds earmarked for applied vaccine research and development to finance their time while they devote much of their effort to pursuing more theoretical scientific objectives. By purchasing vaccines, governments pay for research output rather than research inputs, and thus provide strong financial incentives to actually develop a marketable vaccine. 6 Perhaps the chief advantage of strong financial incentives, however, is that they help not only in motivating researchers, but also in selecting research projects. Advocates for particular diseases and scientists working on the disease may have an interest in exaggerating the scientific potential for advances on that disease. Researchers working on a particular line of research similarly have an interest in exaggerating its promise. Politicians and government scientific administrators may have trouble deciding which diseases are worth working on, and which vaccine approaches, if any, are worth pursuing. They may wind up financing ideas with only a tiny probability of success, or worse, failing to fund promising vaccine research because they do the public sector for this task as the barriers facing private-sector vaccine development. 6 Subsidizing private research through targeted tax credits is subject to problems similar to those affecting grants to researchers. Currently, U.S. companies are eligible for a 20% research and development tax credit. A bill recently introduced in the United States Congress, the Lifesaving Vaccine Technology Act, proposes increasing this credit to 30% for research on vaccines for diseases that kill more than one million people a year. The bill would also allow money-losing firms to pass some of the credit on to their investors, making biotech startups working on these diseases more attractive to investors. Like direct funding of research, tax credits may be difficult to target. Firms doing research with only indirect implications for these diseases might try to claim eligibility for the credit.

19 Vaccine Purchase Fund Kremer 17 not have confidence that its backers are presenting objective information on its prospects. 7 Purchase pre-commitments or other strong financial incentives tied to research outputs have an advantage over research grants precisely because the scientific potential for vaccines is difficult for those outside the field to assess. With an assured vaccine purchase, taxpayers pay nothing unless and until an effective vaccine is produced. Pharmaceutical firms contemplating pursuing a line of research and scientists contemplating joining biotech ventures in exchange for stock options will take the risk if they believe the scientific prospects are worth pursuing, and otherwise will not invest their money and time. This argument applies even more strongly to the public, which has even less information about the scientific prospects for vaccines than politicians. Opinion polls suggest that the public believes that most foreign aid is wasted, but that they would be willing to support foreign aid if they felt the money was reaching the intended beneficiaries [Kull and Ramsay, 1997]. Under a system of grants to research laboratories, the public would not know whether funds are being wasted, thus making support for vaccines less politically appealing. If no public funds are spent until actual vaccines are purchased, the public can be much more confident that its funds are not going to waste. Pull programs let a hundred flowers bloom, while push programs centralize research decisions. While decentralization may lead to some duplication of effort, it also means that mistakes by a single decision-maker will not block progress towards a vaccine. Overall, it seems reasonable to have a two-pronged approach in which public sector and grant-funded university laboratories, with their weak financial incentives, focus on basic research, while strong financial incentives are created for private institutions to focus on the later stages of 7 Equity investments are subject to a similar adverse selection problem. Those researchers who believe that they have identified projects with very high expected net present value will be least inclined to dilute their equity stake, while

20 Vaccine Purchase Fund Kremer 18 vaccine development. The basic structure of institutions designed to help push vaccine development, such as the peer-review process, is already established. The remainder of this paper examines how a vaccine purchase fund could be designed to create pull incentives for the private sector to develop vaccines. (Appendix 1 argues that a vaccine purchase fund is likely to be superior to alternative ways of creating pull incentives for vaccine development, including cash prizes, research tournaments, buying existing vaccines at higher prices than would otherwise be necessary, and allowing vaccine developers to extend patents on other, more commercially profitable pharmaceuticals.) 3. The Design of a Vaccine Purchase Fund This section sketches a proposed design for a vaccine purchase fund. Section 3.1 argues that some discretion will be needed in determining eligibility and payments for vaccines after vaccines have been developed and tested, since it is likely to be impossible to comprehensively set forth rules ahead of time. However, decision makers at a vaccine purchase fund might be tempted to use this discretion to take advantage of vaccine developers by ratcheting up vaccine requirements or offering a low price after the vaccine developer had invested hundreds of millions of dollars in developing a vaccine. Procedures for determining eligibility and pricing should be designed to limit this temptation. Section 3.2 outlines a proposed method for determining vaccine eligibility and pricing. Vaccines would first have to receive regulatory approval and meet some minimal technical conditions. They would then be subject to a market test. Developing countries wishing to purchase vaccines would have to provide a co-payment and would have to draw on an those who are least confident about their research prospects will be most inclined to seek equity investment.

21 Vaccine Purchase Fund Kremer 19 account established for them within the vaccine purchase fund. To provide incentives for development of effective vaccines, bonus payments would be linked to vaccine quality, perhaps explicitly to the number of lives or DALYs saved by the vaccine. Section 3.3 discusses an auction-like procedure designed to avoid offering either too high a price or too low a price for vaccines. Sub-section 3.4 discusses other issues that would arise in establishing a vaccine purchase fund Discretion in Determining Vaccine Eligibility and Pricing Some discretion would be needed to determine vaccine eligibility and payments for vaccines after vaccines have been developed and tested. This is because before candidate vaccines had been developed, it would probably be impossible to fully specify all the requirements for eligible vaccines and an entire system of prices depending on vaccine characteristics. Such characteristics would have to include: the required number of boosters and the ages at which the boosters must be taken. This is important because if too many boosters are required, fewer people will bring their children in to receive the full course of immunization. whether the vaccine can be given along with vaccines that are already widely administered, making delivery much cheaper, or whether a separate schedule would be necessary; the storability of the vaccine under various temperature conditions. Heat stable vaccines are easier to distribute in rural areas of developing countries. The extent of antigenic diversity against which the vaccine is effective. A vaccine might work in some areas, but not in others.

22 Vaccine Purchase Fund Kremer 20 vaccine side effects on different sub-populations; the length of time for which protection would be provided; the effect of the vaccine on people who do not comply perfectly with the delivery development of natural limited immunity, and thus reduce the survival prospects of people who receive an initial dose but not a booster. in the case of malaria, the effect of the vaccine on the mosquitoes which spread the individual who takes it. It may be difficult to encourage people to take such vaccines. whether the vaccine protects against morbidity, as well as mortality, and if so, how would subjects have to be followed to determine the length of protection? How many separate studies in different regions would need to be conducted to assess?; the number of different antigens targeted by the vaccine, and, more generally, the likely extent to which resistance to the vaccine would develop. a long list of manufacturing characteristics: the acceptable fault rate, the shelf life Mis-specifying eligibility and pricing rules could reduce misdirect research incentives away from appropriate vaccines, or vitiate research incentives altogether. For example, if the

23 Vaccine Purchase Fund Kremer 21 interfered with the development of natural immunity and provided only temporary protection. Such a vaccine might merely postpone malaria deaths from childhood to young adulthood. If more useful lines of research. On the other hand, a committee developing specifications might set standards for an ideal vaccine that was difficult to develop in practice, which would discourage 90% reduction in mortality in all localities, potential vaccine developers might not pursue a candidate vaccine that would be likely to yield 99% protection in most regions, but only 85% regions, no vaccine at all might be developed. In order to provide appropriate incentives for researchers, the fund should pay more for superior vaccines; a vaccine with 90% efficiency is than one requiring boosters every five years. However, specifying a full schedule of prices based on characteristics of hypothetical vaccines is likely impossible. fund will have to exercise discretion after vaccines have been developed and tested. However, once the vaccine developer has invested hundreds of millions of dollars in research and developed developer. Sobel [1995] outlines the travails the inventor of the chronometer went through in trying to obtain the reward established by the British government for a method of determining solution, and insisted on improvements and multiple trials, creating repeated delays, until the king intervened on behalf of the chronometer's inventor. Note that once a vaccine had already been

24 Vaccine Purchase Fund Kremer 22 insist on vaccine improvements and to conserve the fund s resources so they could be spent on other worthy health objectives. If pharmaceutical executives suspect that the committee will It is therefore important to design the fund so as to limit tendencies for it to take advantage of vaccine developers. appropriate decision makers, creating incentives for the fund to develop a reputation for fair dealing by having the fund pre-commit to purchase vaccines against several diseases, the fund board by laying out fairly transparent rules for determining eligibility and pricing. The first step would be to appoint decision-makers who are insulated from political pressures by long terms and who have strong preferences for the action that the authority would appointed as central bankers. This suggests that members of the fund s board should be comfortable with the notion that pharmaceutical firms must cover their research expenditures. and UNICEF have a culture of trying to purchase vaccines at the minimum possible price, and there is a history of antagonism between these institutions and the pharmaceutical industry. As 8 major purchaser is in some ways similar to the problem of inducing firms to conduct research and development on weapons for which they expect governments will be the major purchaser. In each case, the government must convince they have already sunk their investments in research. Procurement rules for the U.S. Department of Defense do not instruct procurement officers to purchase orders at the lowest possible price, but instead to purchase at a price that manufacturing costs, which in turn provides an incentive for firms to invest in R&D to produce attractive products that allow them to win procurement contracts. Rogerson [1994] suggests that this serves as a reputational mechanism for institution, with a well-developed reputation about how it treats contractors, and contractors can count on the desire of the Defense Department to maintain a reputation for the future, because the continued existence of the Defense