We know malaria is a big problem and we know fake drugs are a big problem. What do you get when you put them together? Bad news. A recent paper by Martina Bjorkman-Nyqvist, Jakob Svensson and David Yanagizawa-Drott (ungated version here) shows how bad this problem is in Uganda, and provides an innovative way to deal with it.
Let's start from some facts from the paper. Malaria kills 660,000 to 1.2 million people a year, mostly in sub-Saharan Africa. Most people in Africa, when they get what they think is malaria, diagnose and treat themselves. Given that there are a bunch of febrile illnesses out there, this means folks often have imperfect information about what they really have. Moreover, a lot of folks don't have complete information about how malaria is transmitted. When Bjorkman-Nyqvist et. al. (henceforth BSY) query survey respondents, 99% nail mosquito bites as a mode of transmission, but 85% also cite eating "infected mangos", 60% cite contaminated drinking water, and 50% cite contact with an infected person as modes of transmission.
But there are effective treatments out there for malaria. Recommended by the WHO is artemisinin-based combination therapy or ACT (for a review of diagnosis and treatment, see Jed's post from a year ago). Alas, a fair number of drug shop owners lack scruples. BSY send out a trained set of covert shoppers to buy ACT for a "sick uncle." They then utilize handheld scanners to perform Raman spectroscopy on the pills (this is cool not only because it uses laser beams, but also because it allows them to test individual tablets pretty rapidly). They find that 36.8% of the outlets are selling fake drugs -- and 19.4% of all drugs are fakes. And consumers have some inkling that this is the case -- their survey work shows that there is a correlation between individual's belief that the local druggist is selling fakes and the fact that he or she actually is -- but these beliefs are far from perfect. Finally, lest we think the market might sort out this unobserved quality issue, BSY cite papers that show that price dispersion is big enough that higher price can't signal higher quality (making me question my developing country medicine purchase choices).
Inspired by Akerlof's market for lemons paper, they build a model and an intervention to deal with this set of problems. The model leads to some basic predictions. First, since consumers can at least partially infer quality from health outcomes, quantity demanded, quality and price are increasing in expected quality. However, the more naive folks (i.e. folks who don't have great knowledge about transmission) there are, the more fake drugs will be sold.
Enter the intervention. Working with two NGOs (BRAC and Living Goods), a community health promoter (CHP) will be sent to half of the villages (randomly selected) in which BSY are surveying druggists and households. These CHPs sell a different brand of ACT (among other health related goods) that is authentic and priced below the market price. CHPs cannot sell to stores and this is monitored. Coming back to the model, the prediction here is that good quality may drive down (if not out) bad, particularly when consumers have better information about the disease. Of course, since the CHPs are selling at a lower price, existing druggists will lower their prices. These two effects together should also increase consumption.
So what do they find? Before getting to the main results, they lay out some interesting (if not surprising) correlations. First, beliefs about fake drugs are correlated with demand: when mothers believe that the local druggist is selling fakes, kids get 0.7 fewer tablets (off an average of 6.7). Second, druggists are more likely to sell fakes in markets where there are higher levels of misinformation about transmission modes.
Now what does the intervention do? Having the NGO offer real drugs in town has a big impact on the incumbent druggist: there is a 20 percentage point in the chance that a drug shop sells fake ACTs and the share of fake drugs drops by 11 percentage points. Prices also drop from about 8910 Uganda shillings to 7100-7400 shillings (note that the NGOs were selling ACT at around 7000 shillings). In terms of demand, there are no significant effects on the extensive margin, but the average household buys around 2.6 more pills per sick kid. In addition, household beliefs in treatment areas that the local druggist is selling fakes also fall. BSY also utilize baseline household level misinformation about malaria transmission to look at the heterogeneity of treatment effects. Their results indicate that when the share of consumers with bad information is one standard deviation above the mean, there are no improvements in quality when the NGO comes to town. This is a sobering note: not even subsidized markets (at this level anyhow) can overcome consumer ignorance.
So what do these results tell us? Given that over half of the villages had a monopolistic drug seller at baseline, these results are showing us that the provision of authentic goods and the introduction of competition lead to better quality, lower prices, and higher demand in the market overall. This is a result that's interesting not only for its implication for health, but how we might solve other lemon-infused markets. As Akerlof wrote in 1970: "there is considerable evidence that quality variation is greater in underdeveloped than in developed areas." And this paper gives us a start on thinking about dealing with this in multiple realms.