Wealth Effects and Demand for Probabilistic Insurance

In a previous task, individuals are equipped with different endowments. Then, they are confronted with an investment decision that is actually played out, resulting in different wealth levels. Afterwards, all subjects will answer the same set of questions regarding insurance demand with different default probabilities (but in randomized order).

2019-04-15

2019-06-19

Investigation of willingness to pay for probabilistic insurance coverage given different wealth levels

Subjects state their willingness to pay for three probabilistic insurance contracts. Default probabilities are given by 0%, 0.1%, and 1%. Previous research has shown that even the existence of very small default probabilities lead to a sharp decrease in individuals’ willingness to pay. In this experiment, we analyze different factors (such as wealth, cognitive effort, financial literacy, and other sociodemographic factors) influencing the demand for probabilistic insurance.

Most studies (such as Wakker et al. (1997), and Zimmer et al. (2009)) testing the hypothesis of probabilistic insurance so far used hypothetical payoffs. Zimmer et al. (2018) were the first to conduct an incentive-compatible experiment using real monetary payoffs. In their study, subjects were able to insure themselves against a potentially high loss. In contrast, the value to be insured in our study is relatively low. Subjects are exposed to the risk of a loss of 3,500 Talers. This occurs with a probability of 5%. Afterwards, three insurance contracts will be presented (in random order to avoid an order effect), which only differ in the probability of default, namely 0%, 0.1%, and 1% default probability. Subjects are asked to state their maximum willingness to pay. Since all subjects will be incentivized, they should state their true willingness to pay.
Prior to this insurance decision, subjects are equipped with different endowments and execute an investment task following Gneezy and Potters (1997) (see AEA RCT Registry “Relative Wealth Placement and Risk-Taking Behavior” [currently under review, we will edit this and replace it with a link given a positive review] for more details). As a result, two individuals can subsequently achieve the same wealth level, even though they have shown different levels of risk aversion previously. This way, we hope to disentangle wealth effects from other confounding effects such as risk aversion, cognitive effort, financial literacy, and other sociodemographic factors like gender.

All subjects will be recruited using hroot (Bock, Baetge, and Nicklisch, 2014). Further randomization will be done by oTree software.

Individual

No

18 experimental sessions

432 subjects

144 subjects beginning to state their willingness to pay for insurance given default probabilities of 0%, 0.1%, and 1%, respectively