Two challenges motivate Cohen and Dickens in “Adoption of Over-the-Counter Malaria Diagnostics in Africa.” First, individuals with malaria use the wrong malaria drug to treat their illness. They use monotherapies rather than combination therapies, specifically artemisinin combination therapies (ACTs), and the use of monotherapies is more likely to lead to drug resistance. Second, individuals use malaria treatments even when they do not have malaria. Specifically, individuals with fevers take malaria treatment even if they do not have malaria and either an antipyretic or antibiotic would be more effective. This too exacerbates drug resistance.
The favored policy response to the first problem, suboptimal malaria treatment, has been to subsidize the cost of ACTs. Unfortunately, this subsidy does not solve the second problem. Indeed, it may worsen it—a point to which I will return later. The proper policy response to the second problem, excessive malaria treatment, is to get individuals to take rapid diagnostic tests (RDTs) to verify that they have malaria before they take malaria treatments. Of course that is easier said than done. In a pair of papers, Cohen has taken up the question of how one can get individuals to take RDTs.
In a separate paper with Dupas and Schaner, Cohen reports on the results of an experiment in which individuals were randomized to subsidized ACTs and RDTs at different prices. The salient findings are two. First, subsidizing the price of ACTs appears to increase the degree of ACT use by individuals—especially older children and adults—who do not have malaria. Second, demand for RDTs is relatively inelastic. Specifically, demand is the same whether RDTs have zero price or a price equivalent to almost one-seventh of the subjects’ daily wage. The results suggest that, if local pharmacies offer consumers RDTs for sale, those RDTs will be purchased, and the second problem—overuse—will be solved.12
This volume’s chapter by Cohen and Dickens takes up the natural question that follows: under what conditions will firms offer RDTs for sale, at least to the same extent that a social planner would want them to? The long answer is that it depends on several factors, including the beliefs of drug sellers and individuals about the prevalence of malaria and the externalities from excessive use of ACTs. But the useful normative policy proposal that emerges is that appropriate subsidies for RDTs may encourage RDT use and solve the problem that malaria drugs are overused.
In this comment I want to highlight two points that Cohen and Dickens make but do not stress and yet are very important for policymakers to understand. Moreover, I want to raise some more complications that they ought to consider in future research.
The first point I want to stress is that the policy designed to get people to use ACTs rather than monotherapies—ACT subsidies—exacerbates the second, overuse problem. By reducing the gap between the price of ACTs and the drug that individuals should take (antipyretics or antibiotics) if they know they do not have malaria, ACT subsidies also reduce the incentive of individuals to use RDTs and identify the proper drug to treat their illness. Indeed, to the extent that ACTs are more effective at treating malaria than monotherapies because they are less likely to be resistant, they will actually worsen the overuse problem after equating the price of ACTs and monotherapies. The implication is not that ACT subsidies are a bad idea. Rather, it is that the return to such subsidies is lower than expected.13
The second point is that a critical factor in evaluating the efficacy of any subsidy for RDTs is determining how they affect both sellers’ and consumers’ beliefs about malaria prevalence. As Cohen and Dickens acknowledge, if monopoly sellers think that malaria prevalence is lower than consumers think it is, then they would be reluctant to sell RDTs (or would require a higher subsidy to sell RDTs) because, through RDTs, consumers may learn that prevalence is lower and thus they may demand fewer ACTs. What I want to stress is that even if monopolist sellers were uncertain whether consumers thought prevalence was higher than it actually is, the risk that they might would actually encourage monopolists to at least delay selling RDTs. Once consumers learn that malaria risk is lower than they previously thought, that belief cannot be reversed. Thus the decision to sell RDTs has real option value.
The problem is even thornier if the monopolist seller starts wondering why an NGO or the government is subsidizing RDTs. If everyone who currently sought treatment actually had malaria, then there would be no need for RDTs. RDT subsidies are only required if individuals underuse ACTs or if they overuse it. If they underuse ACTs, an alternative solution is to further subsidize ACTs. If they overuse it, the RDT subsidies are required. Thus it is plausible that sellers will infer from RDT subsidies that malaria is lower than consumers suspect. But this very signal will discourage monopolist sellers from offering RDTs in their stores. The one consolation, however, is that this should not affect the behavior of competitive sellers.
Beyond this point I want to recommend some topics for future research on RDT subsidies. The model that Cohen and Dickens present is purposely simplified to convey the basic intuition behind an RDT subsidy. All the comments that follow are meant to complicate that model to make it more realistic and help craft a more appropriate subsidy.
First, and most important, the present model assumes that individuals believe the RDT works. If they are uncertain of RDT accuracy, then they will have lower demand for RDTs. This has two consequences. One is that it is important to model how individuals update their beliefs about the accuracy of tests. From Gentzkow and Shapiro (2006), we know that individuals will judge tests partly by their priors and hence will be slow to learn about the accuracy of tests—at least without successful use of antimalarials to verify tests. Another consequence is that slow learning will require higher subsidies to encourage individuals to use RDTs.
A second topic for research is whether the subsidies for RDTs are so large that firms (or consumers) will face a negative price for RDTs. That raises the problem that governments and NGOs must monitor the use of RDTs; otherwise firms or consumers will simply order and dispose or take duplicative tests just to obtain income from the subsidy. That will increase subsidy costs without benefit.
Third, the present model assumes that individuals do not currently purchase diagnostic tests. But the fact is that they do. Buying an antimalarial is also the purchase of a diagnostic test. If the antimalarial does not work, people know either the antimalarial does not work or they do not have malaria.14 As a result, the product choice they face is not an antimalarial or a test (the RDT). Rather, it is an antimalarial with a diagnostic test or a diagnostic test by itself (the RDT). This will change the equilibrium price for antimalarials, the demand for RDTs, and the magnitude of the subsidy required for the RDT.
Finally, the present model assumes that all individuals have identical beliefs about whether they have malaria and identical valuation for a cure conditional on having malaria. Of course both values will vary among the population. As a result, sellers face a downward-sloping demand for ACTs and RDTs even among people with fevers or with malaria. So a monopolist will sell fewer RDTs than the social planner desires and fewer than a competitive firm would sell, even if there were common knowledge about aggregate malaria prevalence and no externalities from mistreatment, contrary to the conclusion at the end of Sect. 7.4.
In summary, the chapter by Cohen and Dickens in this volume, combined with the companion piece by Cohen, Dupas, and Schaner, is an important step in addressing the problem of antimalarial overuse. The lesson—RDTs must be subsidized along with ACTs—is an important one for policymakers to learn. Further work is required to fine-tune the RDT subsidy amount, but that should not detract from the main lesson.