A Flexible Information Theoretic Approach for Inference of Multiple Regression Function and Marginal E¤Ects

We develop an estimation procedure of multiple regression function and its derivatives (formarginal e¤ects) based on multivariate maximum entropy (ME) density estimation. The relevantmoments and cross moments conditions are selected to be inlcuded in the inferred distribution.Finite sample and empirical examples show that our proposed ME approach produce an esti-mated joint density leading to superior estimation and prediction of the regression function andits derivatives