We introduce a novel adaptive targeted treatment assignment methodology for field
experiments. Our Tempered Thompson Algorithm balances the goals of maximizing
the precision of treatment effect estimates and maximizing the welfare of experimental
participants. A hierarchical Bayesian model allows us to adaptively target
treatments. We implement our methodology in Jordan, testing policies to help Syrian
refugees and local jobseekers to find work. The immediate employment impacts
of a small cash grant, information and psychological support are close to zero, but
targeting raises employment by 1 percentage-point (20%). After four months, cash
has a sizable effect on employment and earnings of Syrians.
|