Keeping up with the Medici! Three essays on social comparison, consumption and risk

Abstract

This thesis investigates the influence of social comparison (i.e. comparing ones outcomes with others such as neighbours, colleagues, etc) on consumption and risk-taking. The first essay (joint with Robert H. Frank and Adam S. Levine) shows how income growth of the top ranks of the income distribution can decrease overall saving rates through a so-called expenditure cascade. As the higher incomes increase their consumption, those ranked below them also increase their consumption in order not to fall behind too much, which causes those ranked below them to increase consumption as well, etc. These consumption cascades can thus lower saving rates throughout the income distribution. We provide empirical evidence for this phenomenon by showing that several proxies related to financial distress (bankruptcy rates, divorce rates, commute times) are positively associated with increases in inequality. The second essay argues shows that social comparison can induce risk-taking. It shows that with comparison convex preferences social comparison would induce both more risk taking and a preference for negatively correlated gambles. With a laboratory experiment we show that although only a third of subjects display the preference for negatively correlated outcomes typical of comparison-convex utility, those subjects take a lot more risks in a social setting, resulting in significantly higher overall risk-taking. This would both explain the puzzling amount of portfolio under-diversification among households, as well as excessive risk-taking among financial professionals in the run-up to the financial crisis. Finally, the third essay experimentally investigates whether subjects focus on rank or social distance when comparing their outcomes. In the theoretical literature both specifications have been used. No support for a social rank effect is found, but a higher social reference point is found to be positively associated with more risky choices, thus lending credibility to the social distance utility hypothesis.