Essays on unemployment and expectations in macroeconomic models

Chapter one begins with work on unemployment fluctuations that argues (1) The job separation rate rises during recessions and (2) The value of non-market activities is low compared to market activities. This paper evaluates the ability of search and matching models to explain these facts. I find models with i.i.d shocks to matches do not generate separation rate volatility. I consider two extensions: inefficient separations and permanent shocks. Both explain separations even with a low value of unemployment. The two models fit the unemployment, vacancy, finding rate and separation-rate moments well. Fluctuations in the separation rate are shown to be an important source of unemployment fluctuations. Chapter two notes that in the permanent income model, consumption volatility depends crucially on the stationarity of income. Additionally, non-stationary and near non-stationary income processes are practically indistinguishable in the US time series. Therefore, I model learning about the income process. The model matches several features of the US aggregate consumption data including: (1) the variance of consumption relative to income, (2) the rise and fall of this ratio and (3) estimated breaks in consumption variance. Additionally, beliefs about the income process substantially explain consumption changes. In chapter three I find that lagged unemployment changes distort consumers' expectations of future unemployment changes by studying unemployment expectations from the Michigan Survey of Consumers. This distortion results in insufficient pessimism at the beginning of a recession and excessive pessimism at the end of a recession. More people expect unemployment to rise when it is falling at the end of a recession than expect it to rise when it is rising at the beginning of a recession. This occurs despite the fact that a VAR correctly predicts the sign of unemployment changes. The paper documents that least squares learning or real time expectations do little to help explain these facts. However, delayed updating of expectations can address some of these puzzles and extrapolative expectations addresses these puzzles the best. Further analysis of the survey data indicates higher income or education are only slightly correlated with expectational errors and the errors significantly affect buying attitudes.