Long-Term Health Insurance: Theory Meets Evidence from Germany

The potential welfare improvement from long-term health insurance contracts that insures policyholders against contemporaneous health expenditure shocks and the reclassification risks have been recognized since Cochrane (1995) and Pauly, Kunreuther and Hirth (1995). While this is a relatively moot issue in countries with national health insurance, the US lacks a private long-term health insurance market. This paper builds on a recent literature on the welfare effects of individual long-term health insurance contracts. In the current policy environment, policymakers essentially face a trade-off between (a) community-rated spot contracts with adverse selection and individual mandates, like in the ACA Exchanges that insure 12 million people; and (b) risk-rated long-term contracts with one-sided commitment and consumer lock-in, like in the individual private market in Germany that fully insures 8 million people.

The goal of this paper is to study the workings of a real-world long-term insurance market. Among the only two countries in the world with long-term private health insurance markets, Chile and Germany, the German market is particularly suited to test the implications of the optimal long-term health insurance contracts. Plans in Germany constitute a pure financial contract, since the provider network and provider charges are uniform across all plans. On the contrary, provider network and charges vary across plans in Chile. Thus, Germany provides a particularly clean test ground for the theory, which abstracts away from the complications of network heterogeneity and medical care charge differences.

In the first part of the paper, we build on existing static and dynamic insurance models such as Handel, Hendel, and Whinston (2015), Harris and Holmstrom (1983), Hendel and Lizzeri (2003) and, most related, Handel, Hendel and Whinston (2017) and derive optimal insurance contracts under scenarios (a) and (b) given the health risk evolutions and income dynamics among the insuree population for Germany.

In the second part of the paper, we use a unique panel of 400,000 individual long-term health insurance contracts from one of the biggest German private insurers to empirically test how close the German real-world long-term market is to the theoretical optimum. The German claims panel data contain the universe of clients from this insurer, where the oldest enrollee has been a client for 85 years and has been enrolled in the same health plan for 40 years. We observe all underwriting information, premium payments, submitted claims, actual plan choices and plan characteristics, and internal plan switching within a large menu of plans. The data span seven years from 2005 to 2011.

The third part simulates counterfactual scenarios and compares them to the theoretically optimal contracts of the first part and the real-word contracts of the second part. In particular, we simulate the welfare effects of a spot market with and without community-rating and adverse selection. To the extent that the theoretical optimal contracts differ from the observed plans, we also hope to provide insights regarding the sources of the differences.