Essays on identification and estimation of macroeconomic models.

详细信息   

• 作者：Choi ; Jinho.
• 学历：Ph.D.
• 年：2011
• 导师：Escanciano, Juan Carlos,eadvisorLeeper, Eric M.ecommittee memberPark, Joon Y.ecommittee memberWalker, Todd B.ecommittee member
• 毕业院校：Indiana University
• Department：Economics
• ISBN：9781124659596
• CBH：3456450
• Country：USA
• 语种：English
• FileSize：2650479
• Pages：149

文摘

This dissertation investigates the identification and estimation of dynamic macroeconomic models. For model identification, much of the macroeconometrics literature relies on linear variation, assuming that correlations are sufficient statistics. In contrast, this dissertation proposes that for a wide class of macroeconomic models nonlinear variation is an important source of identification and can be motivated by economic behavior such as nonlinear monetary policy. The first essay (joint with Juan Carlos Escanciano) proposes new methods for identification and estimation of the hybrid New Keynesian Phillips curve (NKPC) exploiting nonlinear variation in inflation dynamics. Unlike classical linear methods, our nonlinear methods provide evidence of point identification of the NKPC with U.S. postwar data. For estimation, we propose a generalized spectral estimator (GSE) accounting for nonlinear variation and all lags in the sample, while permitting a focus in a particular band of frequencies. We find that the forward-looking component and the inflation inertia are both quantitatively important and statistically significant in explaining the short-run inflation dynamics. The second essay extends the GSE approach to nonlinear models to solve the problem of lack of global identification of GMM methods. A Monte Carlo study of consumption-based capital asset pricing model investigates the finite-sample performance of the new estimator, exhibiting the robustness of our estimation to the dimension of instruments. The third essay (joint with Bin Chen and Juan Carlos Escanciano) proposes a test for invertibility or fundamentalness of VARMA models generated by non-Gaussian independent shocks. We prove that in these models Wold innovations are serially dependent if and only if the structural shocks are non-fundamental. This simple but powerful characterization suggests an empirical way to assess invertibility. We propose a test based on a generalized spectral density to check for serial independence in the Wold innovations. A Monte Carlo study is conducted to examine the finite-sample performance of our test. In an application to the classical bivariate VAR model for GNP and unemployment of Blanchard and Quah (1989), we find evidence of non-invertibility, which may be explained by nontrivial dynamics in productivity as suggested by Lippi and Reichlin (1993).