Cost-Sharing

What it is

The belief behind cost-sharing is widely held in development and health policy. If a net, a pill or a service is free, the thinking goes, people take it casually and do not use it, and the people who most need it are not picked out. A positive price, even a low one, is meant to filter for those who value the product.

Three quantities get run together in this argument, and the experiment separates them. Uptake is how many people take the product at a given price. Use is how many of those who took it actually use it. Targeting is whether the people who take it are the ones who need it. Cost-sharing is defended on the second and third, on the ground that a price improves use and targeting, and it is attacked on the first, on the ground that it reduces uptake. A good test needs a randomised price, a measure of use, and a measure of need, and the Kenyan clinics offered all three.

The argument also depends on the size of the price. A symbolic price is not the same as a market price, and the experiment's top price of $0.75 was the price at which nets were then sold to pregnant women in Kenya, so it was a real policy rather than a token.

Jessica Cohen and Pascaline Dupas tested this in a field experiment in Kenya, published in the Quarterly Journal of Economics in 2010. They randomised the price at which pregnant women could buy long-lasting insecticide-treated nets at prenatal clinics. They report three findings. Women who received free nets were not less likely to use them than women who paid subsidised positive prices. Women who paid higher prices appeared no sicker, in terms of measured anaemia, than the prenatal clients in the control group, so price did not select the women who needed a net more. And cost-sharing considerably dampened demand: uptake dropped by 75 per cent when the price rose from nothing to $0.75, the price at which nets were then sold to pregnant women in Kenya.

The authors combined the estimates in a cost-effectiveness analysis of infant mortality that included both private and social returns to using a net. Given the large positive externality of widespread net use, they concluded that free distribution was both more effective and more cost-effective than cost-sharing.

In effect

For a health programme, the finding moves the burden of proof. A price needs to be justified by what it screens, and in this setting it screened nothing and cost uptake.

A 2007 evaluation of a Kenyan deworming programme by Michael Kremer and Edward Miguel found the same direction. A drug cost-recovery scheme reduced take-up by 80 per cent. Both results involve a health product with benefits that reach beyond the person who takes it, which is the case where free has the strongest argument.

Dupas has also studied what happens after a subsidy ends. Her 2014 paper in Econometrica, "Short-Run Subsidies and Long-Run Adoption of New Health Products: Evidence From a Field Experiment", takes up the question of whether a low price today changes what people will pay tomorrow. That paper is cited here by title only and was not read.

For anyone reading about pricing psychology, cost-sharing is where the zero price effect meets the real world. The experiments on chocolates show the pull of zero in small choices. The Kenyan results show how steeply demand falls when a price appears, with something that matters at stake.

What it does not say

It does not say that subsidies are always better given away. In a 2015 American Economic Review study, Cohen, Dupas and Schaner found that a very high subsidy on anti-malarial drugs sold over the counter dramatically increased access, but nearly one half of subsidised pills went to patients without malaria. Targeting matters for goods that can be wasted or misused.

It does not isolate zero from cheapness. The bed-net and deworming comparisons set nothing against a positive price, so they show demand is very sensitive at the bottom of the scale. Only the chocolate ladder, in which two cents to one cent changed little and one cent to nothing changed a great deal, separates zero from a small price.

It does not extend automatically to consumer goods. The results come from clinics and schools in Kenya, with health products and large external benefits. This page was written from abstracts and the original papers were not read in full.

---