The skin I live in

What if offering people protection makes take bigger risks?

OK – it has been a worthwhile break. Let’s do this.

Ever since a spate of randomized controlled trials revealed that circumcision can functionally reduce a man’s chance of contracting HIV, global health officials have been been pushing mass circumcision campaigns as much as possible in sub-Saharan Africa. Concerned that the assumption that the internally-valid results from the RCTs may not hold in `the real world’, I wrote a fairly critical post a few years ago. My main concern was that no one was properly accounting for post-treatment behavioural changes.

I was particularly worried about something called the Peltzman effect, a concern that economists have about offsetting behaviour: if you reduce the riskiness of a particular activity for someone, then they have an incentive to do that activity more. While academics have had some trouble reliably identifying Peltzman effects in the real world, it is still a valid concern: while those undergoing an experimental treatment may not change their behaviour before more information on the effectiveness of that treatment is known, men who know that circumcision reduces their chance of contracting HIV might react by simply having more sex.

Someone has finally taken a look at this – Nicholas Wilson, Wentao Xiong and Christine Mattson have a paper specifically looking for the Peltzman effect in a group of Kenyan men who participated in one of these RCTs. They subdivide the sample based on their pre-treatment beliefs on the effectiveness of circumcision, then look at risky sexual behaviour in two follow-ups six and twelve months later. The punchline?

Contrary to the presumption of risk compensation, we find that the response due to the perceived reduction in HIV transmission appears to have been a reduction in risky sexual behavior. We suggest a mechanism for this finding: circumcision reduces fatalism about acquiring HIV and increases the salience of the trade-off between engaging in additional risky behavior and living longer. We also find what appears to be a competing effect that does not operate through the circumcision recipient’s belief about the reduction in the risk of acquiring HIV.

At first glance, this sounds like good news: those who said they believed that circumcision was effective reduced their risky behaviour – the exact opposite of what one would expect from a Peltzman-type relationship. Wilson et al. chalk this up to a sort-of income effect: now that their baseline risk of contracting HIV is lower (therefore raising potential life expectancy), subjects who have been circumcised decide to invest more in life quality, avoiding dangerous behaviours.

But wait – this nugget of good news is actually obscuring something else: those who didn’t believe that circumcision worked increased risky sexual behaviour – they were less likely to use a condom during intercourse and had more sexual partners. The authors don’t have any good explanations for this behaviour, other than suggesting that being circumcised might make men more attractive to other women, since it signals that they have a lower chance of having HIV. I’m not sure how plausible this is, unless men are making their circumcision status public, this is the sort of information that is usually only revealed very, very late in the game. How could we get around this? Randomly reveal men’s circumcision status (sounds fun, but unethical – these two sometimes go hand in hand)?

Continue reading

Cash-on-delivery grants in Indonesia

In a new paper, Benjamin Olken, Junko Onishi and Susan Wong examine the effects of a cash-grant programme in Indonesia, randomly allocated (across sub-districts), but with some receiving a cash-incentive bonus, conditional on performance:

This paper reports an experiment in over 3,000 Indonesian villages designed to test the role of performance incentives in improving the efficacy of aid programs. Villages in a randomly-chosen one-third of subdistricts received a block grant to improve 12 maternal and child health and education indicators, with the size of the subsequent year’s block grant depending on performance relative to other villages in the subdistrict. Villages in remaining subdistricts were randomly assigned to either an otherwise identical block grant program with no financial link to performance, or to a pure control group. We find that the incentivized villages performed better on health than the non-incentivized villages, particularly in less developed provinces, but found no impact of incentives on education. We find no evidence of negative spillovers from the incentives on untargeted outcomes. Incentives led to what appear to be more efficient use of block grants, and led to an increase in labor from health providers, who are partially paid fee-for-service, but not teachers. On net, between 50-75% of the total impact of the block grant program on health indicators can be attributed to the performance incentives.

Cash-on-delivery aid critics and proponents should give this a good read. Of course, it won’t settle any arguments, but provides some interesting evidence. A couple of thoughts:

  • The authors find that there are no Milgrom & Holstrom-type external spillovers (i.e. you start paying me to publish more so I start publishing in worse journals) and deduce that non-targeted indicators might have benefited from the extra effort. One caveat: they only look at non-targeted indicators within the domains of health and education, so we can’t say what happened to indicators outside of this domain that went unobserved or unreported.
  • Prior to the intervention, the optimal allocation of funds was unknown, but it seems that the villages solved it: the extracted inefficient spending on school materials and re-allocated it to health. As far as black box solutions go, this is great – but note that the only thing that might have kept educational performance constant was the addition of the performance measures.
  • The performance measures were relative to other sub-districts, so there was no set `target’ that villages had to meet, so not as much scope for threshold effects (slacking off because you confident you will meet the target or that the target can never really be met).

One question we should be asking ourselves when we try and tie this to cash-on-delivery aid: how might a village respond differently to incentives than a national government?

Hat tip to Chris Blattman