This dissertation examines how agents process information and update their beliefs in two different contexts. In the first two chapters we consider dynamic decision problems under perfect information. In the last chapter we consider static, strategic interactions with common knowledge but imperfect information. To tackle our first set of questions we design an experiment analogous to the dynamic consumption problem with stochastic income that households solve in standard macroeconomic models. In the first chapter we show that our subjects condition on past actions in the absence of informational frictions or switching costs. We argue that subjects do so to economize on scarce cognitive resources and develop a model of inattentive reconsideration that fits our data. An implication of our model is that inertia is state- dependent. In the second chapter we revisit the longstanding problem in empirical macroeconomics of excess sensitivity of consumption to income in our experimental data. We find that excess sensitivity arises from two distinct channels. The first channel is an overreaction of households to the arrival of income that is independent of their wealth level. The second is increased excess smoothness with respect to wealth when households receive news about future income. The third chapter examines the scope for persuasion in global games. We consider a central bank with a commitment technology that chooses a robustly optimal persuasion strategy. We show that such a policy can reduce and even eliminate multiple equilibria in such games because it updates agents beliefs so that coordination motives become irrelevant. This suggests that central bankers are better served from influencing the markets through announcements rather than direct intervention.
| Identifer | oai:union.ndltd.org:columbia.edu/oai:academiccommons.columbia.edu:10.7916/d8-p21h-g711 |
| Date | January 2019 |
| Creators | Zorrilla, Oskar A. |
| Source Sets | Columbia University |
| Language | English |
| Detected Language | English |
| Type | Theses |
Page generated in 0.0018 seconds