讲座题目:Forecasting Three-Pass Regression Filter Model with Time-varying Coeffcients: A Rolling Window Selection(时变系数的三重回归滤波模型预测:滚动窗口的选择)
面向对象:面向全院师生
主讲嘉宾:周前坤副教授(单位:路易斯安那州立大学(美国))
学科方向:计量经济学,应用计量经济学
讲座时间:2024年6月14日(周五)下午:10:00-12:00
讲座地点:我学院407
主办单位:44118太阳成城集团我校现代经济学研究中心我校创新发展研究中心
摘要:In this paper, we introduce smooth time-varying coefficients to the three-pass regression filter (3PRF) forecasting method proposed by Kelly and Pruitt (2015). Following Inoue et al. (2017), we employ rolling window selection to estimate the parameters and to generate forecasts. The rolling window selection uses only the most recent observations, which resolves the trade-off between forecast bias and variance. We establish the optimal rate for selecting the most recent observations. Monte Carlo simulations demonstrate that, in general, the rolling window selection method for three-pass regression filter forecasting with time-varying coefficients produces relatively smaller forecasting mean square errors compared to the original three-pass regression filter forecasting method or the method that uses the full sample. An empirical application of our proposed method is considered to forecast the real GDP growth, which highlights the necessity of using our approach.
翻译:在这篇文章中,我们为Kelly和Pruitt(2015)提出的三重回归滤波预测模型引入了平滑的时变系数。我们采用了Inoue等(2017)的滚动窗口选择方法来估计参数和生成预测。这种方法只使用最近的观测数据,从而解决了预测偏差和预测方差之间的权衡问题。我们确定了选择最新观测数据的最佳比率。蒙特卡洛模拟显示,在大多数情况下,结合了时变系数的三重回归滤波预测模型,通过滚动窗口选择法,能够产生比原始模型或使用全部样本的方法更小的预测均方误差。我们还通过实证分析展示了我们方法在预测实际GDP增长方面的应用,进一步验证了采用我们方法的必要性。
作者简介
周前坤,南加州大学经济学博士,现任美国路易斯安那州立大学巴吞鲁日奥尔斯商学院终身副教授,研究主要集中在面板数据模型的估计、检验以及应用。目前已有30多篇论文发表于Journal of Econometrics、Journal of Business & Economic Statistics、Econometric Theory、Journal of Applied Econometrics等国际权威期刊杂志上,并担任计量经济前沿进展期刊(Advances in Econometrics)的联合主编。