Abstract

In Chapter 1 I show that analysts whose forecast revisions were more exaggerated (than granted) in the past make recommendation changes that lead to lower abnormal returns than their peers; consistent with the idea that their recommendation revisions are also more exaggerated. Interestingly analysts' characteristics derived from past earnings forecasts remain informative about future returns to recommendation revisions even after controlling for information obtained from past recommendations, which suggests that in finite samples past forecasts can provide valuable information about same-analysts' recommendations.

In Chapter 2 I study the performance of an unsophisticated individual employing the so called '1/n strategy' in a sample of DC pension plans, finding that for typical investors, the 'out of sample' impact of following this heuristic is insignificantly small. This stands in contrast to estimates based on calibrated models or obtained assuming that the data generating process is known with certainty. Using this simple heuristic does not seem necessarily irrational then, especially in consideration of the time and effort usually required to make 'optimal' investment decisions.

In Chapter 3 I use a gravity model framework to evaluate the extent to which trade declines following sovereign defaults are the result of sanctions imposed by creditor countries. I find that, in the aftermath of defaults, there seems to be no evidence of a larger decline in bilateral trade with creditor countries affected by the default. This would imply that the declines are not due to punishments imposed by these creditor countries. The analysis does not yield evidence of broader punishment strategies including a league of major creditors (not just those affected by the default) either. These results would contradict the predictions of the trade sanctions theory of sovereign borrowing.