Economic models require a formal treatment for individual preferences and expectations. Preferences are often assumed to be stable (and measurable) while expectations of the future are perfectly rational. Empirically, however, there is little evidence for stability of preferences or perfect rationality of expectations. This dissertation assesses the causes and consequences of these phenomena. The first two chapters identify novel channels leading to instability of revealed preferences in laboratory and field experiments, and the third chapter assesses the consequences of expectation bias on the outcome of policy in a general equilibrium model.
Motivated by the extreme events of the financial crisis, the first chapter explores conditions under which preferences may become unstable, or more specifically, are revealed to be more or less risk-averse in response to extreme events. The results support existing literature showing that extreme events can have an economically and statistically significant effect on revealed risk aversion. The chapter provides evidence that investors' own experiences play a key role in shaping revealed risk preferences and the weighting of past observations when forming expectations. The results suggest that experiencing severe negative or positive returns leads subjects' revealed preferences to become closer to risk neutrality, while subsequent asset allocation is affected primarily by subjects' own returns relative to the market and not by the market experience itself, suggesting that agents' performance relative to a benchmark is what matters in shaping expectations.
The second chapter, co-authored with Brian Giera and Biruk Tekele, focuses on risk-taking decisions of Micro-enterprises in developing countries, whose small businesses are known to exhibit high marginal rates of return to investment, but then owners fail to reinvest earned income back into their business to capture these unexploited profits. Recent studies have focused on behavioral biases as an explanation for this behavior, with an attention on mental accounting and loss aversion which could be dampening shop owners ability to grow as their aversion to loss and narrow temporal bracketing lead to under-investment in risky products. Using data from a lab-in-the-field experiment with micro-entrepreneurs in Ethiopia, we first replicate \cite{gneezy1997experiment} and find that our sample does not display any myopic loss averse tendencies when bracketing temporally. We extend their experiment by allowing subjects to invest in a cross-sectionally framed set of assets with equivalent payoff and risk structure as the original temporally framed experiment. In our cross-sectional treatments we find a 46\% increase in the amount allocated to the risky asset, suggesting that attitudes towards risk allocations could be affected by adjusting the perceived investment frame.
Finally, the third chapter assesses the importance of expectation bias in quantifying the effects of macroeconomic policy outcomes. First, I explore the relationship between expectation bias and monetary policy shocks and find that during the post financial crisis period, monetary surprises have a significant effect on expectation bias, which I construct structurally and estimate using option prices on the S
amp;P 500 Index. I then explore the effect of this induced change in expectation bias on the outcome of monetary policy in a general equilibrium theoretical model. The model results show that the effect of monetary policy on aggregate outcomes is highly sensitive to the policy's induced effect on expectations. If monetary tightening causes a decrease in optimism then monetary authorities can do more with less, i.e. achieve a greater effect from a one-percentage point increase in the interest rate relative to the rational expectations equilibrium. The opposite is true if tightening causes an increase in optimism.
