Essays in macroeconomics

Defence date: 23 October 2019Examining Board: Prof. Ramon Marimon, European University Institute, (Supervisor); Prof. Juan Dolado, Universidad Carlos III; Prof. Gaetano Gaballo, HEC, Paris; Prof. Thomas Sampson, LSEThis thesis contains three chapters. The first two consider deviations from rational expectations for understanding the unprecedentedly long period of a binding zero lower bound (ZLB) since the Great Recession. There I show that if agents are adaptively learning, Central Banks can use forward guidance to guide them through the novel economic environment. In the third chapter I take a more long-run structural outlook to study the interplay of skills, technologies and complementarities for understanding differences in labour market outcomes across OECD countries. The first chapter studies the effects of forward guidance (FG) from a novel perspective. Instead of considering FG as a promise for future actions or providing better forecasting, the Central Bank (CB) in the model is giving a signal about its own reaction function. The CB uses FG as a communication device to signal a policy change. The main findings are that clear communication increases welfare compared to no communication, yet vague messages prove ineffective. The second chapter considers the ZLB as an informational curtain for adaptively learning agents as they cannot observe the path of the interest rate. In a model I show that this results in expectations disagreement between the agents and the CB, consistent with the data. The disagreement coupled with the learning of the agents results in explosive dynamics. Forward guidance is shown to restore stability at the ZLB by preventing spurious expectational drift. The third chapter studies the relationship between returns to skill and assortative matching. Using the PIAAC cognitive skills dataset I show that: returns to skill are systematically related to industrial sorting; high-skilled industries have more assortative matching of workers from all occupations; and more developed countries have less mismatch. I further build a model to illuminate the mechanism. I find that rich countries experience a trade-off of lower overall mismatch but higher crosssectoral mismatch, yet due to higher search frictions poorer ones end up being more mismatched overall