Varying random coefficient models

Abstract This paper analyzes unobserved heterogeneity when observed characteristics are modeled nonlinearly. The proposed model builds on varying random coefficients (VRC) that determined by nonlinear functions of regressors and additively separable unobservables. proposes a novel estimator the VRC density based weighted sieve minimum distance. main example bases Hermite which yield numerically stable estimation procedure. shows inference results go beyond what has been shown in ordinary RC models. We provide each case rates convergence also establish pointwise limit theory linear functionals, where prominent is potential outcomes. In addition, multiplier bootstrap procedure to construct uniform confidence bands. A Monte Carlo study examines finite sample properties it performs well even associated far from being heavy tailed. Finally, methodology applied analyze income elasticity demand for housing.