What a small price screens for.

There is a second half to the argument, and it is about the person rather than the sorting. Pay for something and you value it, so you are more likely to use it. Get it for nothing and it ends up in a cupboard. Most people find both halves of this obvious, which is exactly the sort of thing worth checking.

Malaria nets turn out to be a clean place to check it. A treated net does not cost much, and whether it is actually hanging over a bed is something you can walk in and see. So the prices got randomized: across 20 prenatal clinics in western Kenya, 16 of which handed nets out at all, some women were offered a net free and others were asked for a small amount, with the amount varying from place to place.

That is a collapse rather than a trim, and it happened over an amount of money most people reading this would not stop to pick up off the pavement.

Which leaves the other half of the argument, the part about valuing what you pay for. If the price were sorting for intent, the women who paid should have been more likely to hang the net. They were not. Women who received a net free were no less likely to use it than the women who paid.

The study went looking for the sorting in one more place, too. If a price selects for need, the women paying the higher amounts ought to have been the sicker ones. On measured anemia, which is one of the signals of malaria, they were not: the women who paid more looked no sicker than the average woman walking into those clinics.

Put the two results beside each other and they say something sharper than either one on its own.

If the price were sorting people by whether they would use the net, the women who paid should have used theirs more. As far as this study could see, they did not. So whatever the price was sorting on, it was not showing up as use. And in a place where sixty cents is real money, the study did not leave the rest to guesswork. The women who paid the top price were more likely to have paid for transport to the clinic, more likely to be literate, more likely to be wearing shoes, and more likely to report owning animals. The authors keep a hedge on it, because the sample is small enough that most of those gaps fall short of significance on their own, but their reading is that selection under cost-sharing happened at least partly along wealth lines.

That is the uncomfortable version. The filter installed to catch the person who would waste the net was, as far as this study could see, catching the person who could not spare the money. And with no detectable difference in use between the two groups, there is nothing here to suggest the nets that never got handed out were rescued from waste. They were just never handed out.

The shape to notice is that a barrier filters on whatever it is denominated in.

Deposits, application fees, co-pays and minimum balances are priced in money, so they filter on who has money. The form that eats an afternoon is priced in time and attention, so it filters on who has time and attention. Nearly every one of them was defended as sorting for seriousness, and the uncomfortable part is that they are not the same filter as one another. Swap one for the other and you lose a different group of people. What this study is evidence about is the cash kind, and about the cash kind it is fairly blunt: the person the filter was built to keep out was not the person it caught.

So the question worth asking is not whether a barrier filters. Everything filters. It is what the barrier is denominated in, whether that is the thing you meant to measure, and whether anybody has ever checked.