Toggle Main Menu Toggle Search

Open Access padlockePrints

A comparison of the grade of membership measure with alternative health indicators in explaining costs for older people

Lookup NU author(s): Paul McNamee


Full text for this publication is not currently held within this repository. Alternative links are provided below where available.


An important focus in empirical econometric studies of health care concerns estimation of the relationship with health. This paper describes a non-parametric technique, the grade of membership (GoM) procedure, which is used to construct summary indicators of health. Health status is also assessed using three alternative measures commonly employed in the literature: an index of problems of activities of daily living (ADL), an index of health conditions and overall self-rated health. Using data from a sample of frail older people, the relationship between costs and different measures of health are estimated using ordinary least squares (OLS) and instrumental variables (IV) methods. The results show that health influences costs more strongly than other variables, an effect which persists after controlling for potential endogeneity. It is shown that health can be usefully classified using GoM techniques. An alternative measure, based on an index of ADL, also performs well. Measures based on overall self-assessed health and the number of health conditions present were less satisfactory. Although health indicators derived from GoM techniques may be more theoretically attractive, in terms of capturing more fully the multidimensional nature of health, simpler alternatives such as those based on ADL indices may offer practical computational advantages. Copyright © 2003 John Wiley & Sons, Ltd.

Publication metadata

Author(s): McNamee P

Publication type: Article

Publication status: Published

Journal: Health Economics

Year: 2004

Volume: 13

Issue: 4

Pages: 379-395

Print publication date: 01/04/2004

ISSN (print): 1057-9230

ISSN (electronic): 1099-1050

Publisher: John Wiley & Sons Ltd


DOI: 10.1002/hec.833

PubMed id: 15067674


Altmetrics provided by Altmetric