Three essays in the estimation of dynamic macroeconomic models

This dissertation consists of three essays on Bayesian estimation of dynamic macroeconomic models. The first essay is focused on explaining the observed high persistence of hours worked in a standard real business cycle model, while two other essays are about exploring macro-finance interface by linking a sticky price macro model with the yield curve data. In chapter 1, which is the joint work with Yongsung Chang and Frank Schorfheide; I estimate two versions of dynamic stochastic general equilibrium (DSGE) models in which hours worked are either stationary or non-stationary depending on the persistence of labor supply shock. If firms can freely adjust labor inputs, the data support the latter specification. Once we introduce frictions in terms of labor adjustment costs, the overall time series fit improves and the model specification in which labor supply shocks and hours worked are stationary is preferred. In chapter 2, I study the time variation of the federal reserve\u27s target inflation between 1960 and 2004 using both macro and yield curve data. A New Keynesian DSGE model in which the target inflation follows a random-walk process is estimated. I compare estimation results obtained from both macroeconomic and yield curve data, to estimates obtained with macro data only, in order to determine what the yield curve tells us about target inflation. In the joint estimation, the estimated target inflation is found to be much higher during the mid 1980s than the corresponding one in the macro estimation. Also, when private agents have to filter out the random walk part of the target inflation from the composite target inflation, some part of the decline in the target inflation during the early or the mid 1980s seems to be perceived as temporary. In chapter 3, a nonlinear version of a sticky price DSGE model is used to identify macro factors behind the movement of the yield curve. In empirical finance models, yield curve dynamics are often determined by statistical latent factors such as level, slope, and curvature . By imposing structural restrictions from a DSGE model, I obtain clear economic interpretations for statistical term structure factors. I propose new closed-form solutions for bond yields to make the estimation of a nonlinear macro model with the yield curve data practically feasible. The likelihood-based approach built on particle filtering is applied in a Bayesian way. The main findings are as follows. First, the level is accounted for by a persistent monetary policy shock, while the slope and curvature are driven by a markup shock and a transitory monetary policy shock, respectively. Also, the decline of the yield term premium after the early 1980s is captured by a persistent monetary policy shock. Second, there is a tension in improving macro implications of the model (e.g., too volatile inflation) and term structure implications (e.g., too low term premia) simultaneously. It turns out the nonlinear analysis does not resolve this problem because nonlinear terms in macro dynamics increase term premia at the cost of amplifying the volatility of inflation