The natural interest rate in semi-structural unobserved components models

我们的项目沿着两条不同的道路前进，在估计自然利率(NRI)时推进科学知识的前沿。我们将证明，在结构多指标未观测分量模型中利用横截面数据可以显著减少估计的不确定性，并显著提高估计NRI的精度。第二个调查途径涉及决定净资产收益率的因素的选择和适当的计量模型。我们开发了一个贝叶斯模型选择程序来经验地确定NRI的时间序列属性。我们还建议结合贝弗里奇-纳尔逊趋势周期分解和sVaR方法来识别NRI及其潜在的结构性决定因素。

Our project moves along two different avenues of advancing the frontier of scientific knowledge in estimating the natural rate of interest (NRI). We are going to demonstrate that utilizing cross-sectional data within a structural multiple-indicator unobserved components model leads to a substantial reduction in estimation uncertainty and a dramatic improvement in the precision of estimating the NRI. The second avenue of investigation is concerned with the selection and appropriate econometric modeling of the factors determining the NRI. We develop a Bayesian model selection procedure to determine empirically the time series properties of the NRI. We also propose to combine a Beveridge-Nelson trend-cycle decomposition with a SVAR approach to identify the NRI and its underlying structural determinants.