Average partial effects in multivariate probit models with an application to immigrants’ ethnic identity and economic performance

Content Provider	Semantic Scholar
Author	Bruno, Giovanni S. F. Dessy, Orietta
Copyright Year	2014
Abstract	We extend the univariate results in [Wooldridge, J. M. (2005): “Unobserved heterogeneity and estimation of average partial effects,” in Identification And Inference For Econometric Models: Essays In Honor Of Thomas Rothenberg, ed. by D. W. K. Andrews, and J. H. Stock. Cambridge University Press, New York] to multivariate probit models, proving the following. 1) Average partial effects (APEs) based on joint or marginal response probabilities are identified in multivariate probit models with general conditionally independent latent heterogeneity (LH), provided that the error covariance matrix in the structural model is unconstrained beyond normalization. If this caveat is not met, identification requires that the covariance matrix for the LH components be restricted. This finding is substantial since in most coded routines for multivariate probit models it is not possible to adjust the form of the covariance matrix to the parametric structure of the latent regression model. Stata's biprobit, mvprobit, mprobit and cmp or Limdep's BIVARIATE PROBIT and MPROBIT are cases in point. 2) Conditionally independent LH does not assure identification of APEs based on conditional response probabilities, unless additional independence restrictions are maintained. 3) The dimensionality benefit observed by [Mullahy, J. (2011): “Marginal effects in multivariate probit and kindred discrete and count outcome models, with applications in ∗Bocconi University, Milan †Cà Foscari University, Venice and Catholic University of Milan
File Format	PDF HTM / HTML
Alternate Webpage(s)	https://www.stata.com/meeting/italy14/abstracts/materials/it14_bruno.pdf
Language	English
Access Restriction	Open
Content Type	Text
Resource Type	Article