Essays in Macroeconomics and Macroeconometrics

Abstract

This thesis contributes to macroeconomics and macroeconometrics. Chapters 2-4 study the role of producer heterogeneity for business cycles and macroeconomic development. Chapters 5-6 provide inference for structural vector autoregressions. <br /> Chapter 2 examines the role of time to build for business cycles. We document that time to build is volatile and largest during recessions. In a model with producer heterogeneity and capital adjustment frictions, the longer time to build, the less frequently firms invest, and the less firm investment reflects firm productivity. Longer time to build thus worsens the allocation of capital across firm. In the calibrated model, one month longer time to build lowers GDP by 0.5%. <br /> Chapter 3 investigates the role of uncertainty fluctuations. We exploit highly disaggregated industry-level data to study the empirical importance of various transmission channels of uncertainty shocks. We provide testable implications for the interaction between various frictions and the job flow responses to uncertainty shocks. Empirically, uncertainty shocks lower job creation and raise job destruction in more than 80% of industries. In line with theory, these responses are significantly magnified by the severity of financial frictions. In contrast, we do not find supportive evidence for other transmission channels.<br /> Chapter 4 re-examines the importance of misallocation for macroeconomic development. We ask whether differences in micro-level factor productivities should be understood as a result of frictions in technology choice. We document that the bulk of all productivity differences is persistent and related to highly persistent differences in the capital-labor ratio. This suggests a cost of adjusting this ratio. In fact, a model with such friction can explain our findings. At the same time, the loss in productive efficiency from this friction is modest.<br /> Chapter 5 studies structural VAR models that impose equality and/or inequality restrictions on a single shock, e.g. a monetary policy shock. The paper proposes a computationally convenient algorithm to evaluate the smallest and largest feasible value of the structural impulse response. We further show under which conditions these values are directionally differentiable and propose delta-method inference for the set-identified structural impulse response. We apply our method to set-identify the effect of unconventional monetary policy shocks. <br /> In Chapter 6 we study models that impose restrictions on multiple shocks. The projection region is the collection of structural impulse responses compatible with the vectors of reduced-form parameters contained in a Wald ellipsoid. We show that the projection region has both frequentist coverage and robust Bayesian credibility. To address projection conservatism, we propose a feasible calibration algorithm, which achieves exact robust Bayesian credibility of the desired credibility level, and, additionally, exact frequentist coverage under differentiability assumptions.