Course theme

Type of course

Banner
Banner

Combining choice experiments and agent based modeling to analyze preference uncertainty

  • 8 December, 2023
  • 09:45
  • VU University Amsterdam, Institute for Environmental Studies (IVM)
  • Prof. R. Brouwer
  • Dr I. Logar

In Switzerland, flooding occurs with a relatively high frequency and can cause serious damages. Protective measures from both public and private sources can help mitigate consequences when prepared. Decisions to invest in flood protection measures are, however, associated with uncertainties, stemming from both the chance-based occurrence as well as unfamiliarity with making such a choice. The aim of this PhD thesis is to explore the impact of such uncertainties on protective decisions and the stability of underlying preferences. To this end, an online survey with a discrete choice experiment (DCE) is conducted among a representative sample of the five German speaking cantons of Switzerland with the largest population. The DCE elicited respondents’ willingness to pay for flood protection depending on flood risk reduction, measure type (public or private) and price. In addition, a lottery game was used to elicit their risk attitudes. To better explore the dynamics of outcome and preference uncertainties on preference forming and resulting decisions, an agent-based model (ABM) is developed and used to simulate the choice process in a DCE while relaxing behavioral assumptions from the expected utility theory. The thesis adds to the existing literature by developing a dynamic decision-making framework assuming preference uncertainties. The framework is modeled as an ABM and parameterized with the data from the DCE. The overall result is a dynamic outlook on preference formation and an attempt to predict preference stability over time. As the evaluation of risk might be subject to risk preferences, the thesis tests a variety of behavioural assumptions about decision making under uncertainty in the context of flood risk reductions. Comparing predictions from the Expected Utility Theory (EUT) and the Prospect Theory (PT) the results show that including risk attitude parameters, even when collected in a financial lottery context, can increase the explanatory power of the choice model. Best predictions were achieved when applying assumptions based on PT, suggesting that respondents weight risk and the reduction thereof in a non-linear manner. Preference uncertainties stemming from unfamiliarity with the aspects of the decision can also influence a decision. Survey respondents that had previous flood experience, talked to peers about possible flood protection measures or received information about frequency of flooding in their region at the beginning of the survey were more likely to choose alternatives with additional flood protection measures. Another source of uncertainty is the difficulty in weighing choice attributes. Respondents that stated a strong certainty about their choices also made less random choices, suggesting consistency between self-reported and inferred choice certainty. The thesis combines knowledge gathered in the previous parts by parameterizing the ABM with the elicited preferences and risk attitudes from the DCE and the lottery game. To enhance the dynamic nature of the ABM, a Bayesian framework is used to simulate decisions for the represented agents. The approach predicts how preferences, as elicited in the CE, are likely to develop over time when exposed to repeated flood risk protection decisions. Importantly, the simulation results suggest no major change in the average welfare estimates for protective measures. The heterogeneity of the sample however increases because of self-perception learning but decreases due to social interaction learning. The model results show that preference learning appears to be strongest when the taken decisions and the related trade-offs are more difficult.

Shape
Researcher

Markus Glatt

readmore
shape Banner
We provide a disciplinary and multidisciplinary research programme aimed at advanced understanding of environmental problems and advanced training of PhD candidates in this field.
More information
shape Banner
Here you will find a selection of international events related to the socio-economic and natural sciences of the environment.
View our events