Supplementary Appendix to: Nearest Comoment Estimation with Unobserved Factors

Abstract

Full paper is available at: https://ssrn.com/abstract=3087336.

In this supplementary appendix to the paper Boudt, Cornilly and Verdonck (2019), we first provide a brief R tutorial for the proposed NC estimator. Then, we go into more detail about the shape of its influence function. Additionally, we study the MSE in the case where n / p is constant and consider the sensitivity of the NC estimator with respect to model misspecification. We extend the simulations in the main paper by showing the performance of the proposed AIC and BIC selection criteria. We include three goodness-of-fit statistics and examine their coverage in the simulation settings of the main paper and explain how the theory changes when the mean of X is assumed to be known. Furthermore, we give examples of moment expansions that are relevant in economics and finance, define mean-variance-skewness-kurtosis efficient portfolios and give the element-wise formulas for the moments under a factor model.

Keywords: higher-order multivariate moments; latent factor model; minimum distance estimation; risk assessment; structural equation modelling

JEL Classification: C100; C130; C510