Econometrics Workshop - Anna Mikusheva
Tue, 04/11/2017 - 11:40am
498 Uris Hall
"Optimal Tests with a Functional Nuisance Parameter (joint w/Isaiah Andrews)"
We consider the problem of optimal hypothesis testing in potentially nonlinear moment equality models without assumptions on identification. Building on our previous result that this may be considered as a testing problem with an infinite-dimensional nuisance parameter, we obtain a tractable parameterization for the model under both the null and alternative. We then study weighted average power optimal conditionally similar tests for independent weights on the structural parameters and a functional nuisance parameter, with Gaussian process weights for the functional parameter. Building on results from the theory of Gaussian processes, we show that a natural further restriction on the weight for the functional parameter yields an intuitive, computationally tractable class of optimal tests.
Event Categories: Econometrics