Bridging the Gap Between Revealed and Stated Preferences in Flood-prone Housing Markets

This paper serves as the first step in unravelling the interplay of various biases on the gap between the two estimation techniques with a special focus on the value of simulation techniques in this analysis. Thus, we highlight only on two aspects – violations from the expected utility framework and spatial correlation of property attributes that may confuse risk assessments. The paper's aim is to quantitatively explore (1) the market-level consequences of various behavioural models' assumptions regarding risky choices at the individual level, and (2) the effect of the market-level bias that individual preferences for neighbourhood quality imply. By pursuing this dual goal, we contribute to understanding the gap between individual choices, on which stated preference methods rely, and observed market outcomes based on revealed preferences methods. We consider three types of behavioural biases that are known to distort markets: subjective weighting of probabilities, loss aversion, and correlation between amenities and risks.

We proceed with an overview of confounding factors that affect flood-risk analysis on the micro and macro level. Then, we dive more into the details on how ABMs can alleviate the impact of these confounding factors and combine the strengths of revealed and stated preference approaches. The paper proceeds by briefly describing the model and the data that is used for its calibration and validation. We discuss the different sub-models that are systematically compared in the results section. Next, we explain how the analysis is done, and we conclude by discussing the results and their implications.