Topics in Economic Dynamics and Time Series

点击翻译按钮获取中文摘要

Research will be completed on three different topics:First, the project develops a new approach to characterizing nonlinearities in weakly dependent time series. Nonlinear principal component decompositions are used to represent the evolution of continuous-time Markov processes and to summarize the correlation structure. These decompositions help us understand better the evolution of relative prices across locations and the nonlinearities induced by transaction or transport costs. This work draws on three literatures: functional principal component analysis; quadratic-form modeling of Markov processes; and nonlinear, stochastic modeling of real exchange rates. Nonlinear models that are strongly dependent or nonstationary are also considered.Second, the investigator continues his study of economies in which private agents and policy-makers confront model misspecification. The project explores and implements solution methods for nonlinear, stochastic equilibrium models. It studies monetary models in which the misspecification concerns of macroeconomic policy makers differ from those of private agents. It builds from insights in the time-series econometrics literature on misspecification and filtering to investigate alternative formulations of robustness in the decision-making of private agents. Finally, the project studies connections between robustness, adaptive learning and information- based models of inertia.Third, the project examines the observable implications of dynamic models under uncertainty with human capital and portfolio investment. Implications for time series and cross sections of asset returns are deduced and tested. The role of heterogeneity in the stochastic structure of labor income also is considered. Finally, the project investigates the role of financial market structures on the shadow valuation of human capital.