Detail Econometric modeling of macroeconomic aggregates Bibliografi Author: Sims, Christopher A. (Advisor); Schorfheide, Frank ; Phillips, Peter C.B. (Advisor) Topik: ECONOMICS; GENERAL Bahasa: (EN )    ISBN: 0-591-90696-1     Penerbit: Yale University Press     Tahun Terbit: 1998     Jenis: Theses - Dissertation Fulltext: 9837280.pdf (0.0B; 0 download) Abstract In some areas of macroeconomics, quantitative model evaluation has proceeded without formal statistical methods because they lead to rejections of the interesting models. An example is the evaluation of dynamic stochastic general equilibrium (DSGE) models. This dissertation discusses an econometric methodology that can be formally applied if probabilistic candidate models are potentially misspecified. Positive prior probability is assigned to a more general reference model. In Chapter 2, based on an idea by Ingram and Whiteman (1994), a potentially misspecified DSGE model is used to motivate a prior distribution for the coefficients of a vector autoregression (VAR) to improve forecasts. Chapter 3 provides a Bayesian perspective on loss function based forecasting procedures. The candidate model is a simple parametric time series model and the reference model is comprised of a broad class of linear processes. The econometrician finds it too onerous to conduct a full Bayesian analysis with the reference model. Based on frequentist model checking, e.g., Box (1980) a choice is made between the Bayes predictor derived from the candidate model, and a predictor obtained through loss function estimation of the candidate model. The loss function predictor might lead to more precise forecasts if the data were generated from the reference model. Asymptotic risk properties of the proposed procedures are derived. Chapter 4 introduces a framework for the empirical evaluation and comparison of DSGE models. The candidate set is composed of DSGE models that are supposed to be assessed according to predictions of, for instance, unconditional moment properties of macroeconomic aggregates or responses to discretionary policy actions. A VAR serves as reference model. Unlike in Chapter 3, a full posterior analysis with the reference model is conducted. The DSGE models are evaluated according to the overall posterior risk associated with their predictions. The evaluation is coherent under misspecification, that is, low posterior probability of all DSGE models, and no misspecification. The framework provides a procedure to determine model parameters and criteria to weight different DSGE models. It can be implemented through Bayesian simulation methods. One application is an evaluation of a standard cash-in-advance model and a liquidity model. Opini Anda Klik untuk menuliskan opini Anda tentang koleksi ini! Lihat Sejarah Pengadaan  Konversi Metadata   Kembali

Copyright © 2006, 2007 Unika Atma Jaya, all rights reserved