Generalized Bayesian Estimation, Forecasting, and Policy Analysis in Dynamic Stochastic General Equilibrium Macroeconomic Models

动态随机一般均衡宏观经济模型中的广义贝叶斯估计、预测和政策分析

• 批准号: 0617803
• 负责人: Francis Diebold
• 金额: --
• 依托单位: University of Pennsylvania
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2006
• 资助国家: 美国
• 起止时间: 2006-07-01 至 2011-06-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0617803&HistoricalAwards=false
• 关键词: Generalized; Bayesian; Estimation; Forecasting; Policy

This project develops and applies tools for estimation, forecasting, and policy analysis in dynamic stochastic general equilibrium macroeconomic models. These methods are fully structural and fully Bayesian, shrinking vector autoregressions in the direction of the prior mean. These methods are unique in that although the model is estimated over all frequencies, the shrinkage occurs only at frequencies selected by the user, such as business cycle frequencies. Although superficially resembling the band-spectral methods of several decades ago, they are fundamentally different and facilitate a tight integration of macroeconomics and econometrics in estimation, forecasting, and policy analysis - using the same model, and acknowledging misspecification, throughout.The intellectual merit of the work is high, as the problems addressed - acknowledging misspecification while finding a workable middle ground between the unappealing extremes of pure calibration and classical maximum-likelihood estimation - have eluded solution for decades and are widely acknowledged to be both highly challenging and crucially important for compelling empirical application of modern macroeconomic models.The broader impacts of the project are substantial and several-fold. First, it will contribute directly to teaching and learning via mentoring and collaborating with graduate students. Second, it will reach out to underrepresented groups via broad web-based dissemination of all research results. Third, it will enhance infrastructure for research and education by establishing a variety of collaborations: between disciplines (by deepening our understanding of the econometrics / macroeconomics interface), between researchers and nations (by utilizing national and international coauthorships and joint projects), and between academia and other communities including government and policy organizations (by facilitating and accelerating knowledge transfer from academia). This knowledge transfer is crucially important, because the problems on which the proposed research focuses are precisely those that have hindered widespread application of modern macroeconomic models in government and policy organizations. Hence if the intellectual merit of the research is high, its value is nevertheless much greater than that associated purely with its intellectual merit. Indeed, the research will significantly push modern macroeconomic modeling, forecasting, and policy analysis toward routine application, producing improved policy, and ultimately improved general performance of the macroeconomy.

该项目开发和应用动态随机一般均衡宏观经济模型中的估计，预测和政策分析工具。 这些方法是完全结构化和完全贝叶斯的，在先验均值的方向上收缩向量自回归。 这些方法的独特之处在于，尽管模型是在所有频率上进行估计的，但收缩仅发生在用户选择的频率上，例如商业周期频率。尽管表面上类似于几十年前的带谱方法，但它们有本质上的不同，并且促进了宏观经济学和计量经济学在估计、预测和政策分析中的紧密结合--自始至终使用相同的模型，并承认规格错误。这项工作的智力价值很高，作为解决的问题-承认错误的规定，同时找到一个可行的中间地带之间的不吸引人的极端纯校准和经典的最大似然估计-几十年来一直没有得到解决，被广泛认为是极具挑战性的，对于现代宏观经济模型的强制性经验应用至关重要。首先，它将通过与研究生的指导和合作直接促进教学和学习。第二，它将通过在网上广泛传播所有研究成果，接触代表性不足的群体。第三，它将通过建立各种合作来加强研究和教育的基础设施：学科之间（通过加深我们对计量经济学/宏观经济学界面的理解），研究人员与国家之间（通过利用国家和国际合作作者和联合项目），以及学术界与包括政府和政策组织在内的其他社区之间（通过促进和加速学术界的知识转移）。这种知识转移是至关重要的，因为所提出的研究重点正是那些阻碍了现代宏观经济模型在政府和政策组织中广泛应用的问题。因此，如果一项研究的智力价值很高，那么它的价值就比纯粹与智力价值有关的价值大得多。事实上，这项研究将大大推动现代宏观经济建模、预测和政策分析走向日常应用，产生更好的政策，并最终改善宏观经济的总体表现。