Comparing Many Treatments
Event type: Seminars28 February 2019 from 12.30 pm - 2.00 pm
Speaker: Wilbur Townsend – Predoctoral Fellowship, Harvard University
Abstract: This paper demonstrates how to compare a large number of mutually exclusive treatments, given access to local average treatment effects between pairs of those treatments. It derives an impossibility theorem which demonstrates the maximal heterogeneity of treatment effects such that a (cardinal or ordinal) index on the treatments can be constructed, with comparisons using the index being independent of irrelevant alternatives. The theorem also implies how similar the populations identifying the local average treatment effects must be, given heterogeneous treatment effects. After presenting that restriction, the paper shows how it can be tested, and then shows how the postulated index can be estimated. It also shows that an existing estimator of such a treatment index requires even stricter restrictions on the heterogeneity of treatment effects.
About Wilbur Townsend: Wilbur Townsend currently holds a Predoctoral Fellowship in Harvard University's economics department, where he studies the relationship between social capital and economic mobility. He previously worked at Motu, where he studied unemployment, earnings dynamics and technology. Townsend has published research in Economics Letters, the Journal of Economic Behaviour & Organization, and the Journal of Business & Economic Statistics. He holds degrees in economics, mathematics and philosophy from Victoria University of Wellington.
Tea and Coffee will be provided prior to the start of the Seminar. RSVP is not required.