On the third Thursday of every month, we speak to a recent graduate about their thesis and their studies. This month’s guest is Dr Stefan Lipman who has a PhD from Erasmus University Rotterdam. If you would like to suggest a candidate for an upcoming Thesis Thursday, get in touch.
Title
Decisions about health: behavioral experiments in health with applications to understand and improve health state valuation
Supervisors
Werner Brouwer, Arthur Attema
Repository link
https://repub.eur.nl/pub/130911/
Why did you choose to use behavioural experiments in your research?
The use of behavioural experiments was already, to some extent, predetermined by the project on which I was hired, which centred around applications of behavioral economics to health. In this field, experiments are, to my knowledge, the main tool of discovery. It is also a testament of my undergraduate studies (I hold a BSc. and MSc. in Psychology), in which most research I encountered as well as the methods courses I followed were focused on experimental work. Switching disciplines when I started my Ph.D. studies in health economics was already quite a challenge so, to date, I have focused on applying the methods I’m familiar with. As much as I appreciate work on observational or panel data, I lack the econometric skills to run such a project. Also, each time I come up with a research question, my first instinct is not to consider: ‘Where can I find the data to answer this?’ but rather: ‘How can I collect data to find this out myself?’. It’s a different mindset; it took me a while and a lot of discussion with colleagues to appreciate how different exactly.
Do people make decisions about health and non-health outcomes in the same way?
This depends on how we define ‘the same’! In my research, I have studied decision-making for both health and monetary outcomes. A clear commonality is that the type of decision-making that economists consider ‘rational’ is hardly ever observed. Instead, in my dissertation, we showed that many behavioral insights apply across both domains. Consider loss aversion, for example, which refers to individuals’ tendency to assign a larger weight to losses than to gains of the same size. In this paper, I measured loss aversion for monetary outcomes (e.g. gaining or losing 100 euro), while in two other papers (here and here) we measured it for life duration (e.g. gaining or losing two years of life). Comparing results across these papers shows that, in both domains, most people are loss averse, with median estimates also largely similar. However, my dissertation also shows differences.
In Chapter 4 we studied preference reversals and found these to be more common for health outcomes than for monetary outcomes. It is also important to note that my work has not necessarily studied how people decide about health and non-health outcomes; this requires a different approach. For example, it has been shown that decisions about health may involve entirely different (neural) processes than deciding for non-health outcomes, and individuals may be more likely to ignore risks for health than for non-health outcomes.
What does your research tell us about the way people trade-off length and quality of life?
Trade-offs between length and quality of life are crucial for the quality-adjusted life year (QALY) models that underlie a lot of my work. In particular, I have studied such trade-offs in the context of health state valuation, i.e. the process of finding out the value or utility associated with health states such as life in a wheelchair. My research has shown that individuals are willing to trade-off length and quality of life, but do so reluctantly. That is, they are loss averse and are relatively unwilling to give up health compared to their current reference-point. Hence, when trading off length and quality of life, it appears to matter what reference-point one takes. For example, we found evidence suggesting that individuals take their subjective life expectancy into consideration in such trade-offs; when considering to trade-off length and quality of life, the age someone expects to attain is taken into consideration.
Do your findings imply that we should adopt new methods for health state valuation?
In the second part of my thesis, I explored differences between the commonly used standard gamble (SG) and time trade-off (TTO) methods. These methods, when applied to value the same health state, will yield different results. This could be problematic if economic evaluation could come to depend on the method we use. In a recently published paper, I’ve shown that SG would generally yield higher valuations than TTO, but also that both methods’ valuations are considered too high by respondents themselves. Hence, SG and TTO (to a lesser extent) are biased upwards.
My research suggests this is not necessarily a problem with the methods, but rather the theoretical framework applied to derive valuations from them. These frameworks, as they are applied in practice, allow researchers to easily derive these valuations from decisions about health, but they also place unrealistic demands on these decisions. For example, we assume that individuals consider each year worth the same as the next, whereas most people will consider this year of larger importance than a year in the future. Furthermore, because each year is worth the same, there is no room for loss aversion. In my dissertation, we showed that by measuring how people decide about health, and incorporating this information in health state valuation, the differences between TTO and SG may disappear.
Hence, I suggest we don’t necessarily need more methods, we need to relax the strict assumptions we use with TTO and SG. In practice, this means that to interpret peoples’ decisions in health state valuation, we need to know more than just their preferences between health states. We need to know how they feel about (losing) length of life!
How can your work be used in the design of health policies?
The results presented in this dissertation imply that assuming everyone decides rationally about health misrepresents actual decision-making. As such, policy aimed at improving decisions with health consequences can learn from behavioral economics. In particular, reference-points matter, loss aversion is relatively stable and extends to decisions about health, and people overreact to small probabilities while underreacting to large probabilities (i.e. probability weighting). Indeed, such insights have already reached policymakers, as the past decade has seen a large increase in attention for behavioral public policy, through the use of nudges or other policies inspired by behavioral insights.
My work, however, suggests that policymakers should recognize the inherent heterogeneity in how individuals decide about health, both within and between individuals. For a single individual, decisions about health may differ depending on whether risks or delays are involved, and health preferences could depend on how they are elicited. The findings of this dissertation furthermore suggest that large differences exist in, for example, loss aversion or probability weighting between individuals. Hence, health policies that aim to benefit from these behavioral insights – e.g. nudges, or information campaigns and financial incentives that are ‘behaviorally inspired’ – may have heterogeneous effects.Although my somewhat disappointing (null) results suggest that much work remains needed in this area, my research suggests that ‘tailoring’ interventions to this heterogeneity could be a promising way forward (see, for example, the work on personalized nudges, by: Peer et al., 2019). Such personalization could potentially improve the (cost-)effectiveness of health policies, by attempting to fit policies to individuals’ heterogeneous preferences.
Has your PhD research influenced the way that you make decisions about health?
Fortunately, I haven’t had to make explicit trade-offs between my length or quality of life, as one would have to do when severely ill. Of course, as I also argue in the introduction of my dissertation, we all make decisions about our health every day, or at least decisions that will at some point affect our health. Doing a Ph.D. has certainly influenced my decisions about health, as the structure of full-time work has helped me exercise much more consistently. In fact, my pride about the research completed pales in comparison to how I feel about the 5 to 10 kg muscle mass I’ve gained since 2016. However, I wish I could write here that my research and expanded knowledge of decision fallacies has enabled me to ban all unhealthy habits from my life, but that would be a flagrant lie. If anything, I’ve become more mindful of the trade-offs that I make, as befits an economist.
