More health care utilization with more insurance coverage? Evidence from a latent class model with German data

Abstract

We analyse the impact of optional deductibles, private supplementary health insurance and income on the demand for health care utilization, measured as the number of physician visits with data from the German Socio-Economic Panel (SOEP). With a set of newly available variables for the years 2002, 2004 and 2006 that measure individual health more accurately and including risk-attitudes towards health we find that possible endogeneity of the insurance choice is not a problem. A latent class approach that takes into account the panel structure of the data reveals that especially individuals who have few doctor visits, the low users, respond strongest to insurance status and income. In this group we find that more insurance increases the demand for physician visits and there is a pro-rich inequity in health care utilization. No such effects are found for the high users.

Keywords:

• health insurance; deductibles; supplementary insurance; count data; latent class panel model

JEL Classification::

• I11; I18; G22

Acknowledgements

I am grateful to Daniel Baumgarten, Stefan Felder, Annika Herr, Hendrik Jürges, Annika Meng, Alfredo Paloyo, Wolfram Richter, Reinhold Schnabel and Nicolas Ziebarth for valuable comments on earlier versions of is article. Moreover, I want to thank an anonymous referee for improving this article. This article benefited from discussion at the EEA annual meeting 2008 in Milan the 2nd RGS doctoral conference in Essen and the annual meeting of Verein für Socialpolitik 2008 in Graz. Financial support by the Leibniz association is gratefully acknowledged. All errors are my own.

Notes

¹ Recently they were used in different specifications by, e.g., Deb and Holmes (2000), Deb (2001), Gerdtham and Trivedi (2001), Deb and Trivedi (2002), Jimenez-Martin et al. (2002), Atella et al. (2004), Bago d’Uva (2005, 2006), Bago d’Uva and Jones (2009) and Zheng and Zimmer (2011). Latent class models also became increasingly popular in other fields of economics, see, e.g. Greene and Hensher (2003) or Shen (2009) for applications using transport mode choice data.

² However, in 2004, a co-payment of 10 Euro for the first visit in a quarter was introduced. No effects of this reform on the demand for doctor visits are found by Augurzky et al. (2006) and Schreyögg and Grabka (2010). Since they use the same data set as we do, we do not take this reform into account. Also, since our latest year of analysis is 2006, the reform of 2007 is of no importance for this study.

³ The data used in this article were extracted using the Add-On package PanelWhiz v2.0 (Nov 2007) for Stata. PanelWhiz was written by Dr John P. Haisken-DeNew ([email protected]). The PanelWhiz generated DO file to retrieve the SOEP data used here and any Panelwhiz Plugins are available upon request. Any data or computational errors in this article are my own. Haisken-DeNew and Hahn (2006) describe PanelWhiz in detail.

⁴ Questions concern, among others, bodily pain, stress, ability to carry out everyday tasks, phases of melancholy, etc.

⁵ The question in the SOEP is: ‘How would you rate your willingness to take risks with your health?’.

⁶ We used the ml command in Stata and drawed on the code provided in Jones et al. (2007). Various starting values were used to rule out local maxima of the likelihood function.

⁷ Risk-aversion concerning financial matters is asked for in the same way as the risk-aversion concerning health. The question about the attitudes towards co-payments was asked only in 2002. As in the case of risk-aversion, we assume that this is a time-invariant preference.