A randomized test of welfare

In the Economist, no less. Consider:

When the results were in, the team found that the unpaid women had suffered more than twice the HIV infection rate experienced by the paid women over the course of the 18 months of the experiment, and four times the infection rate of genital herpes. Intriguingly, there was no difference between the infection rate suffered by those required to go to school and those who received the money unconditionally. Whether the actual amount of money mattered was not clear. For that to emerge a larger sample would be needed. What is abundantly clear, however, was that the money did make women behave differently. They had younger boyfriends than those in the control group, and had sex less frequently.

What should be noted is that this was a randomized experiment so you can actually infer causality.  I am positive Mark will have a lot more to say about this experiment.

But let me note, for the record, that this is the opposite result of what conventional thinking would yield about giving young people cash subsidies.  It's also notable that requiring school did not change the good results so unconditional transfers are not inferior to conditional transfers.  Are we sure that a social safety net would result in worse outcomes?  What about giving grants to college students?