Collaborative Research: The Role of Commitment in Dynamic Contracts: Evidence from Life Insurance

National Bureau Of Economic Research Inc, Cambridge MA

Investigators

Abstract

This project presents an empirical and theoretical analysis of the role of commitment in dynamic contracts. We consider an environment where payoff relevant information symmetrically evolves over time. Long term insurance is an example of this where the ability of parties to commit to long term contracts can have a profound effect on the design of contracts. When bilateral commitment is impossible, short term insurance is still provided but consumers may be left to suffer classification risk: the risk that future information about the agent's type will prove unfavorable hence increasing future premiums. This issue has important welfare and policy implications. Contract theory has received little empirical attention. Models are difficult to test. They tend to be very stylized, and the nature of the question involves relations between non-observables, making measurement very difficult. The goal of the project is to look at the life insurance industry which is an ideal environment for testing models of optimal dynamic learning under one-sided commitment. First, life insurance is a prototypical example of an environment in which agents receive information sequentially about their health state that may render previous actions (contracts) suboptimal for one or both parties. Second, contractual data is available. Third, life insurance contracts are offered in several varieties ideally suited to analyze of the effects of commitment and renegotiation. This variety of contracts is used to test the implications of the model. To our knowledge no previous work on the dynamics of contracts has used direct contract information. The existing models on one-sided commitment are adapted to obtain empirical predictions on the design of life insurance contracts. A preliminary empirical analysis of the observed contracts is then presented. This has the following purposes: (i) testing the theory, (ii) evaluating the potential inefficiencies arising from lack of commitment, and (iii) studying the way the industry copes with the problem. The predictions of the model are supported in the data. First, as predicted by the model, virtually every offered contract involves some degree of front-loading. This finding is in sharp contrast with what would be expected under competition and lack of commitment (and in light of the health insurance experience), i.e., a sequence of short term contracts that leave the insured subject to reclassification risk. Second, in line with the predictions of the model, front-loading (lock-in) is associated with lower lapsation, and in turn with better risk pools. Finally, according to our numbers, the long-term level contracts capture a good part of the gains from long term insurance. It is worth noting that the industry achieved this partial solution to the problem of reclassification risk without need of regulation (e.g., no imposition of guaranteed renewability). Lack of renewability and the absence of insurance of classification risk are some of the concerns raised about health insurance. Understanding the life insurance experience may contribute to the policy debate over health insurance and help in understanding some aspects of the (mis)functioning of the health insurance market. The health care market suffers from a variety of other problems. The analysis of this project may isolate those problems that are solely due to the lack of bilateral commitment and hence to understand better the source of inefficiency in health insurance.