Essays in bounded rationality and economic experiments

Abstract

This thesis studies bounded rationality in three different contexts through the lens of economic experiments. The first two essays focus on preferences, which constitute the foundation of economic theory. In modern economics, preferences are assumed to have a consistent underlying structure based on a small number of axioms. However, empirical evidence suggests that preferences are not always well-defined. The first essay studies the descriptive and predictive power of the axiomatised expected utility theory and its alternatives, specifically, Tversky & Kahneman (1992)'s cumulative prospect theory and Bordalo, Gennaioli & Shleifer (2012)'s salience theory. We conduct a Lab experiment with binary choice questions over lotteries and find that both alternatives race closely and outperform expected utility. The second essay examines the economic importance of anchoring. Anchoring is proven to be robust in the psychology literature, but the quantitative economic signicance of the phenomenon has not been given enough focus. We conduct a systematic synthesis of experiments examining the effects of numerical anchors on willingness to pay (WTP) and willingness to accept (WTA), and find that the effect of anchoring is relatively smaller than previously believed. Another key aspect of bounded rationality is the premise that people have limited computational power, and examining the economic and political implications of this is a fundamental task of modern social scientists. My third essay studies these implications in a voting environment with biased polls. In our experimental design, there is a strict subset of voters that is informed about the quality of the candidates, and polls serve to communicate this information to uninformed voters. Voters in the treatment group are presented with biased poll results, which favour systematically one candidate. The result shows that voters fail to infer the biased rules behind information revelation and account for it, since voters in the treatment group consistently elect the candidate favoured by polls more often than in the unbiased control conditions.

More information

Identifiers

ORCID for Zacharias Maniadis: orcid.org/0000-0002-3225-0835

Catalogue record

Export record