One of the most controversial subjects in global health is the topic of user fees for health services and commodities. Ever since Nancy Birdsall, David de Ferranti and John Akin declined to rule out user fees as a useful source of health financing way back in 1987, the World Bank has been pilloried for "advocating user fees" in the health sector, which the Bank has never done. By setting out explicit criteria for setting user fees in the 2004 World Development Report, the Bank resisted continuing political pressure to ban them outright and further fueled the debate. Among the donors, DFID (see also this white paper) has most adamantly rejected user fees as either effective or legitimate for improving access to health care by the poor. On the other hand, the social marketing of health care commodities, from condoms to antibiotics, is increasingly popular among actors such as PSI and KfW and presumes that the optimal price of these commodities, while heavily subsidized, is not free.
The obvious argument against user fees is based on the fundamental economic proposition that demand curves slope downward* -- e.g. that the number of people willing to purchase a product or service declines as its price increases. It follows that lower prices should result in more health care utilization than higher prices, and that zero prices would be even better. However, a fascinating new study by Nava Ashraf, James Berry, and Jesse Shapiro of the market for home water purification solution in Zambia finds that in fact demand curves seem to flatten out as the price approaches zero, and may even slope upward -- or, in plain English, that the act of paying a small amount (up to 18 cents, in this case) actually increases use over distributing Clorin free of charge.
The authors posit two reasons why people might behave contrary to the simple law of demand. First, the price effectively targets the distribution of the health commodity to those least likely to waste it. Second, people who have paid more for a product may have a greater psychological commitment to using it. They find strong statistical support for the first of these effects and weak support for the second.
If these results could be generalized to other health commodities, like bednets to prevent malaria, or to health services such as curative health clinic visits, the suggestion would be that sufficiently small user fees do little to discourage utilization; they might even increase it by stimulating the supply of health care quantity and quality. Although 18 cents -- the highest price that can be charged for Clorin in Zambia without reducing utilization -- seems like a tiny sum to people in rich countries, this would be enough to substantially motivate a distributor of a health commodity such as bednets or condoms, and a similarly small sum might help fill a financing gap in health clinics.
Obviously, this has huge implications for policymakers in developing countries, where cutting subsidies could actually improve overall health outcomes while motivating both public and private providers at the periphery of health care systems. Let's hope that the debate on user fees for health care is enriched by more such randomized controlled studies and that they look in more detail at supply and financing effects as well as demand-side effects.
*As an aside, the existence of upward-sloping demand curves has actually been in the news a lot this week due to recent evidence proving the heretofore purely theoretical case of Giffen goods.
On January 9th, we at CGD had the pleasure of hosting Jessica Cohen for a presentation of her Brookings Working Paper authored jointly with Pascaline Dupas and entitled Free Distribution vs. Cost-Sharing: Evidence from a Malaria-Prevention Field Experiment in Kenya. This paper uses a randomized-controlled trial to test the hypothesis that people who pay more for a commodity, in this case a mosquito bed net, are more likely to use it then people who pay less. In contrast to Ashraf et al's results for the water purification treatment, Cohen & Dupas found no such effect. What's going on?
Well, as is always the case with rigorous impact studies, generalizing is a problem. While Ashraf et al delivered their water purification product to the consumers, Cohen & Dupas sold or gave away the bed net product to expectant mothers after they had walked to a clinic. One can imagine that the same woman who would accept free chlorine (perhaps to please the salesman at the door), but then not use it, might refuse to carry home a free bed net unless she wanted to use it. The substantial bed net promotion campaign that had gone on for years in the region of Western Kenya where the study was conducted may have disseminated so much information about the utility of bed-nets that the women did not need to use the price as an indicator of quality. Or the prices charged for bed nets in this experiment might simply not have been large enough to trigger cognitive dissonance.