Hostname: page-component-7b7567c8c-48dgx Total loading time: 0 Render date: 2026-11-18T09:56:49.052Z Has data issue: false hasContentIssue false

Actuarial Applications of Multivariate Two-Part Regression Models

Published online by Cambridge University Press:  02 April 2013

Edward W. Frees*
Affiliation:
University of Wisconsin - Madison
Xiaoli Jin
Affiliation:
University of Wisconsin - Madison
Xiao Lin
Affiliation:
University of Wisconsin - Madison
*
*Correspondence to: Edward W. (Jed) Frees. E-mail: jfrees@bus.wisc.edu
Rights & Permissions [Opens in a new window]

Abstract

This paper synthesizes and extends the literature on multivariate two-part regression modelling, with an emphasis on actuarial applications. To illustrate the modelling, we use data from the US Medical Expenditure Panel Survey to explore expenditures that come in two parts. In the first part, zero expenditures correspond to no payments for health care services during a year. For the second part, a positive expenditure corresponds to the payment amount, a measure of utilization. Expenditures are multivariate, the five components being (i) office-based, (ii) hospital outpatient, (iii) emergency room, (iv) hospital inpatient, and (v) home health expenditures. Not surprisingly, there is a high degree of association among expenditure types and so we utilize models that account for these associations. These models include multivariate binary regressions for the payment type and generalized linear models with Gaussian copulas for payment amounts.

As anticipated, the strong associations among expenditure types allow us to establish significant model differences on an in-sample basis. Despite these strong associations, we find that commonly used statistical measures perform similarly on a held-out validation sample. In contrast, out-of-sample risk measures used by actuaries reveal differences in the association among expenditure types.

Information

Type
Papers
Copyright
Copyright © Institute and Faculty of Actuaries 2013 
Figure 0

Table 1Joint Counts Among Types of Events and Total Event Counts

Figure 1

Table 2Dependence Ratios Among Types of Events

Figure 2

Table 3Odds Ratios Among Types of Events

Figure 3

Table 4Summary Statistics of Expenditures by Event Types

Figure 4

Table 5Spearman Correlations Among Five Types of Expenditure

Figure 5

Table 6Counts by Number and Type of Events

Figure 6

Table 7Association Test Statistics From Logistic Regression Fits

Figure 7

Table 8Spearman's Rho after Controlling for Covariates

Figure 8

Table 9Odds Ratios Estimates and t-statistics

Figure 9

Table 10Copula Parameters Estimates

Figure 10

Table 11Copula Parameters t-statistics

Figure 11

Figure 1Comparison of Predictive Distributions. The smooth density is the predictive distribution from the independence model. The rectangular histogram is from the dependence model. The arrow marks the actual held-out 2007 expenditures.

Figure 12

Table 12Out-of-Sample Risk Measures

Figure 13

Table 13Covariate Description and Percentage of Positive Expenditures by Level of Covariates

Figure 14

Table 14Marginal Logistic Regressions for Five Types of Events

Figure 15

Table 15Marginal Regressions for Expenditure of Five Types of Events Assuming Gamma Distribution

Figure 16

Table 16Out-of-Sample Statistics