This is the goal of a dissertation Yang et al. (2022) in and out health economics. They start with a good review of the literature and divide the approaches around risk taking into type types. I briefly summarize the literature below.
direct method. These approaches may ask individuals about their attitudes toward risk or the likelihood of engaging in risky behavior.A paper by Domen et al. (2011) show that this immediate question provides an estimate of risk appetite, which is useful for predicting an individual’s actual willingness to take risk. This straightforward approach is easy to implement in surveys.
indirect method. These methods usually take the form of lottery tickets. Holt and Laurie (2002) measure risk appetite for financial outcomes in this way; Gneezy and Potter (1998) Use lottery tickets to see myopic loss aversion, and Eckel and Grossman (1998) Look at the difference in risk appetite between men and women.However, it is estimated financial Risk appetite has been shown to be a poor predictor of risk appetite for healthy behaviors (see Anderson and Mailer 2008; Slack et al. 2012). However, many studies have implemented methods to examine risk appetite for health states:
- Arrieta et al. (2017) Using Holt and Laury style methods, but with different distributions over survival (see figure below)
- Atma et al. (2013) Test whether survival distributions matter, whether they are framed at absolute levels compared to expected survival gains and losses (it turns out that people are risk averse for both gain and loss survival).
- Galizzi et al. (2016) Examining risk and time preferences between patients and physicians found that physicians placed less emphasis on future outcomes than patients, but physicians and patients had similar risk preferences. Galizzi uses a variation of Holt and Laury’s multiple price list method, and the result is to live for a few days and be completely healthy.
- Krieger and Meyerhof (2012) Examining patient preferences for health status (i.e. sick versus not sick) and monetary outcomes combining risk aversion and caution. They found that “risk aversion and prudence in medical decision-making reduced the (mean) treatment threshold by 41% relative to risk neutrality (from 50.0% to 29.3% prevalence). Risk aversion accounted for 3/3 of this effect. 4, prudence accounts for 1/4.”
- Ruard et al. (2018) Using Attema et al. (2013) Semiparametric utility heuristics, using deterministic equivalence to binary prospects involving lifetimes.
- Van der Poel and Ruggieri (2008) It was found that individuals “tend to be risk-averse for gambling involving an immediate risk of death and risk-seeking for other healthy gambling”.Follow-up studies Ruggeri and Van der Poel (2012) found similar results.
Yang et al. The paper goes on to examine risk appetite for health status and risk appetite for risky behavior using survey methods. Direct methods are direct. It asks:
How would you rate your willingness to risk your health?please write a number
In the box below, between 0 and 10, where 0 means “not at all ready to take risks” and 10 means “completely ready to take risks”. “
The indirect derivation of risk appetite is based on a standard approach to gambling, with one option presenting a risk of good and bad health, while the other gives a certain health outcome.
The authors also aimed to verify how well these risk appetites correlated with risky behaviors. Risk behavior outcomes were defined as smoking status/regular smoker; frequency of drinking five or more alcoholic beverages; and overall course of non-adherence to medication. Using direct and indirect methods, they found:
…risk appetites derived using direct methods are better predictors of health-related risk behaviour than risk appetites derived using indirect methods.
This finding is not surprising, as the topics covered in the direct approach are more closely related to the risk behavior analyzed. Furthermore, while risk appetite certainly plays a role in risky behaviors, so does the utility of those behaviors. It is quite possible that risk appetite for health status is somehow related to utility for risky behavior. For example, risk-averse people may generally be cautious people, but may enjoy risky behavior (eg, drinking or smoking) as a stress reliever. Alternatively, if risk-averse people are risk-averse for side effects, they may be less likely to take a full course of the drug. Therefore, the finding that direct heuristics are better at predicting risky behavior is not surprising, but it does not suggest that indirect methods are not a better measure of risk appetite than health. Certainly more research is needed on this topic.
Arrieta et al. (2017) Method
Galizzi et al. (2016) Method
Ruard et al. (2018) Method
Yang et al. (2022) Method
