Researchers from the Universities of Eastern Finland and Bern have tested economic choice models in health-care data. Regret-based modeling showed potential in predicting individuals’ health-care choices. This research asks one of the fundamental questions in economics: how individuals make choices. If we can find an answer, we can better understand the systematics of consumers’ choices.

A unique aspect of the study was that Swiss medical admission data was transferred outside Switzerland for the first time in the country’s history. The results were published in Applied Economics.

How people choose a service or commodity can be modeled in several different ways. In this paper, the researchers tested the performance of two different choice models. The utility maximization model is based on the assumption that when choosing, an individual tries to maximize their utility, whereas the regret minimization model assumes that an individual tries to minimize the regret of not choosing an alternative service or commodity.

Comparing these two models, the researchers focused especially on how sensitive the models are to individual heterogeneity, meaning how they deal with the differences between individual preferences and intrapersonal random variation in preferences. The difference between the models is how they evaluate various characteristics/attributes of the alternative goods or services.

The data were collected from Swiss hospital registers for 2007–2012. The choices people made were between public, semi-private, and private beds. The data covered seven selected illnesses and altogether almost 800,000 individual hospital episodes. The statistical methods used in the comparison of the two approaches were multinomial logit models in regard to the variances in utility and regret function, goodness-of-fit, and predicted marginal effects of additional payment.

The models’ performance testing indicates that random regret-based models perform better. Hence, regret-based choice modeling captures signi?cant loss or gain in choice situations. Thus, it can be used as a complementary to conventional utility-based modeling. In health care, this is relevant also because of the outcome uncertainty of all health-care services. It is unlikely that the axioms of the expected utility theory remain consistent in all health-care choice situations.