Measuring is knowing? Eliciting preferences in the general population

Preferences are fundamental characteristics of individuals and have been shown to be predictive of decisions in a variety of domains. For example, people’s willingness to take risks (risk preferences) has been shown to affect investment and occupational decisions. The willingness to defer immediate gratification (time preferences) is related to saving and retirement decisions, and caring about others (social preferences) affects donating behavior and attitudes toward redistribution. This thesis is about measuring preferences in a general population sample of the Netherlands. Preferences are elicited with methods that ask people to make actual decisions, usually with real (financial) incentives, from which preferences are inferred (revealed preferences), and methods that ask people to state their own perception of their preferences (stated preferences). The thesis contributes to a better understanding of (i) the validity of measures, (ii) the stability of measures after experiencing life events and during a crisis, and (iii) differences between self-employed workers and employees in their preferences.

Chapter 2 examines whether measures from different risk preference elicitation methods correlate with each other (convergent validity) and with field behavior (external validity). Previous literature suggests that stated methods perform better than revealed methods in terms of convergent and external validity when it comes to measuring risk preferences. One critique of this literature is that measurement error is often not properly accounted for. Measurement error can occur, for instance, because of varying attention and focus of participants. A contribution of this study is that we correct measurement error using a method that was recently proposed in the literature. We find that the correlation between methods improves when controlling for measurement error. This provides an indication that not accounting for measurement error can partly explain the lack of convergent validity among revealed risk preference elicitation methods found in previous studies. At the same time, we find clear differences between stated and revealed methods when it comes to their external validity. Revealed methods do not correlate well with risk-related field behavior, even when controlling for measurement error. Stated methods correlate with most types of risk-related field behavior and correlations are of economic significance. Thus, measurement error appears insufficient to explain why the external validity of incentivized risk preference elicitation methods is generally found to be low.

Chapters 3 and 4 investigate the stability of preferences after personal life events and during the COVID-19 crisis, respectively. Stable preferences are often implicitly assumed, but it is important that this assumption is validated empirically. Chapter 3 examines the effect of (recent) marriage, divorce, and parenthood on risk, time, and social preferences. The findings suggest that there are only some short-lasting effects of personal life events on preferences. Importantly, however, the results from revealed and stated preference methods largely do not coincide. Chapter 4 examines the effect of the COVID-19 crisis on preferences. Preferences were measured right before and over a one-year period during the COVID-19 pandemic. The findings from this study suggest that preferences remained remarkably stable throughout the pandemic. The results from both studies are encouraging from a theoretical and a practical point of view as they support the assumption of stable preferences. However, more research is needed to investigate why the literature on these topics is far from conclusive.

Chapter 5 compares the preferences and traits of self-employed workers and employees. The study contributes to question “Who are the self-employed?”. The question has been studied before but remains relevant because the labor market is constantly changing and the characteristics of self-employed workers vary across countries and time. The results show that self-employed workers state to be more patient compared to employees but behave equally or less patiently in revealed preference methods. In addition, self-employed workers in our sample are found not to differ from employees in terms of self-control and financial literacy, contrasting results from previous studies. Other findings suggest that self-employed workers are more willing to take risks (as inferred from both stated and revealed methods), are more optimistic, and have lower trust in institutions but higher trust in other people.