Predicting Multivariate Two-Part Health Outcomes
Edward W. Frees, Xiaoli Jin, Xiao (Joyce) Lin
Wisconsin School of Business, University of Wisconsin-Madison, WI, United States

In the paper, we explore multivariate two-part models of health outcomes. We use data from the US Medical Expenditure Panel Survey to explore outcomes that come in two parts. In the first part, zero expenditures correspond to no use of health care services. For the second part, a postive expenditure corresponds to an amount of utilization. Outcomes are multivariate, the five components being (i) office based, (ii) hospital outpatient, (iii) emergency room, (iv) hospital inpatient, and (v) home health expenditures. For each individual in the survey, we base our predictions on age, gender and many more covariates that have been explored extensively in the healthcare literature. To assess the utilization of all expenditure types, we use multivariate binary regression models. For the amount of utilization, we use generalized linear models for marginal distributions of each expenditure type with elliptical copulas for the association among expenditure types.

Keywords: Copulas

Biography: Edward W. (Jed) Frees is a Professor of Business at the University of Wisconsin-Madison and is holder of the Assurant Health Insurance Professorship of Actuarial Science. He is a Fellow of both the Society of Actuaries and the American Statistical Association. Professor Frees was the Editor of the North American Actuarial Journal and is an Associate Editor for Insurance: Mathematics and Economics. In addition to over 50 journal articles, he has published 3 books. The most recent is a 2010 book entitled “Regression Modeling with Actuarial and Financial Applications”, published by Cambridge University Press.