The generic insurance product involves an agent giving up a certain amount of money ex ante some risky event in the expectation of being given some money in the future if something unfortunate occurs. It is immediate that has something to say about the positive and normative evaluation of insurance from the perspective of the insured. Risk preferences play a role, and there is now a rich body of theory and empirical evidence that some agents behave differently than the standard Expected Utility Theory. Time preferences play a role, as the premium is typically paid prior to the stream of expected benefits in the future, and again there is a rich body of theory and empirical evidence that some agents deviate from the standard Exponential discounting model. Subjective beliefs play a role, as perceptions of loss probabilities need not be the same as actuarial assessments, nor need they be updated over time consistently with Bayes Rule. Trust plays a role, as there can be fine print, lawyers, and outright corruption in any contract, leading to nonperformance risk. Finally, some have questioned whether models of probabilities and time-dependent and state-dependent payment structures are sufficient to describe the insurance decision completely. Some have argued for an explanatory role for affective, or emotional, components in the decision framework for insurance demand. These moving parts combine to allow one to explain a wide range of possible behaviors, and the general scientific challenge is how to rigorously identify each of their roles. Doing so, in a structural manner, clearly matters for normative policy design. Do we need better products, better decisions about insurance products, or a healthy mix of both? Some years ago, we started an annual series of workshops on Behavioral Insurance and Risk Management, alternating between Munich (MRIC, the Munich Risk and Insurance Center, Ludwig-Maximilians-Universitat) and Atlanta (CEAR, the Center for the Economic Analysis of Risk, Georgia State University). This symposium of the Journal of Risk and Insurance reflects research in this vein. Research in behavioral insurance approaches this generic characterization of the insurance product by allowing a wide range of assumptions about behavior. It is possible to identify five clear methodologies with behavioral insurance, with surprisingly little overlap to date. 1. Theorizing With Behaviorally Motivated Assumptions. Some of the earliest behavioral insurance research involved the use by theorists of alternative assumptions to the familiar list of neoclassical economics. As behavioral economics offers specific alternative models of decision making with respect to risk, uncertainty, ambiguity, time, and learning, one naturally finds a reconsideration of previous results. 2. Empirics With Hypothetical Surveys. A large body of research involves the use of hypothetical surveys of choice and valuation of insurance products that are hypothetical. There are several ways in which questions can be hypothetical. They could be questions that are about real products, but posed hypothetically in the sense that there are no monetary consequences to the respondent for one answer or another. Or they could be questions about hypothetical products altogether. Thought experiments, an important staple of theorists, fall into this category. 3. Empirics With Laboratory Experiments. The use of controlled, salient payments depending on the response is the hallmark of an experiment, and laboratory experiments are conventionally defined as those undertaken with a convenience sample of students for small stakes. 4. Empirics With Naturally Occurring Data. In some settings, there exist naturally occurring data on salient choices that allow one to draw inferences about behavior, without any additional incentives provided to subjects or without any randomization applied. …
Read full abstract
Feb 1, 2017
Dujuan Guan + 3
Open Access