Sectoral prices and price-setting.

摘要

This dissertation explores the price-setting behavior of firms both theoretically and empirically. The first portion constructs a theoretical model of price-setting in which firms are rationally inattentive: they cannot perfectly attend to all sources of uncertainty. By accommodating multiple sources of uncertainty within the model, it is possible to reasonably calibrate key parameters of the model. This bolsters the case for rational inattention as a microfounded alternative to ad-hoc mechanisms in order to generate price-stickiness and it not only allows for multiple sectors but demonstrates why their introduction is important. The second portion contributes to the empirical literature exploring disaggregated price series. Taking into account the lessons from the theoretical model, a combination of dynamic factor and unobserved component models are applied to explicitly model heterogenous dynamic processes for sectoral prices. The key finding is that models with enforced homogenous dynamics are outperformed under a variety of criteria. More importantly, models with enforced homogenous dynamics can generate erroneous conclusions with respect to the speed of price responses to aggregate and idiosyncratic shocks. A large body of recent empirical work on price-setting, including the empirical exercise described above, estimates a dynamic factor model using a relatively simple and partially non-parametric method. This method is valid in large samples, but alternative parametric methods exist that may be more efficient in small samples. The final portion of this dissertation compares methods for the estimation of dynamic factor models, including non-parametric, classical, and Bayesian techniques. The results of a Monte Carlo experiment validate the use of the partially non-parametric method, but find that the Bayesian approach may provide weakly superior results.