Non-parametric hypothesis testing procedures and applications to demand analysis (Q580848)

This paper proposes a hypothesis test that a (possibly vector-valued) regression function g lies in a particular family of functions \({\mathcal F}\), not necessarily a finite-dimensional parametric family, where \({\mathcal F}\) is a compact subset of an appropriate topological space of continuous functions. Under the assumption of independently and identically distributed disturbances with mean 0 and known covariance matrix \(\Sigma\), the test is shown to be consistent for a broad range of alternatives and an upper bound on the significance level is established. The application on which this paper is focused, is a non-parametric approach to demand- and factor analysis. In particular, the authors describe asymptotically valid and consistent non-parametric tests of utility maximization and utility maximization with homothetic utility.