New Methods and Theory for the Comparison of Nonparametric Trend Curves

比较非参数趋势曲线的新方法和理论

• 批准号: 430668955
• 负责人: Professor Dr. Michael Vogt
• 金额: --
• 依托单位: Institut für Statistik
• 依托单位国家: 德国
• 项目类别: Research Grants
• 财政年份: 2019
• 资助国家: 德国
• 起止时间: 2018-12-31 至 2021-12-31
• 项目状态: 已结题
• 来源: https://gepris.dfg.de/gepris/projekt/430668955?language=en
• 关键词: New; Methods; Theory; Comparison; Nonparametric

The main purpose of the project is to develop new methods and theory for the analysis of nonparametric time trend curves. Recently, there has been a growing interest in econometric models with non- and semiparametric time trends. Non- and semiparametric trend modelling has attracted particular interest in a panel data context. Important questions are whether the observed time series in the panel all have the same trend or whether they can be clustered into groups with the same trend. A number of test and clustering methods have been developed in the literature to approach these questions, which are relevant in a variety of economic and financial applications. Most of the proposed methods, however, depend on a number of bandwidth or smoothing parameters whose optimal choice is a notoriously difficult problem. In our project, we tackle the challenge of developing new test and clustering methods which are free of classic bandwidth parameters and thus avoid the issue of bandwidth selection. To achieve this, we will build on techniques from statistical multiscale testing which have recently been introduced into the literature. The methodological and theoretical analysis of the project will be complemented by simulations and empirical applications. In particular, we intend to apply the developed methods to an empirical question of interest in macroeconomics, that is, the question of whether real GDP growth has been faster in some countries than in others.

该项目的主要目的是为非参数时间趋势曲线的分析开发新的方法和理论。近年来，人们对具有非参数和半参数时间趋势的计量经济模型越来越感兴趣。非参数和半参数趋势建模在面板数据背景下引起了特别的兴趣。重要的问题是面板中观察到的时间序列是否都具有相同的趋势，或者它们是否可以被聚类到具有相同趋势的组中。一些测试和聚类方法已在文献中开发，以接近这些问题，这是相关的各种经济和金融应用。然而，大多数提出的方法，取决于一些带宽或平滑参数的最佳选择是一个众所周知的困难的问题。在我们的项目中，我们解决了开发新的测试和聚类方法的挑战，这些方法没有经典的带宽参数，从而避免了带宽选择的问题。为了实现这一点，我们将建立在统计多尺度测试，最近被引入到文献中的技术。该项目的方法和理论分析将辅之以模拟和实证应用。特别是，我们打算将发达的方法应用于宏观经济学中感兴趣的经验问题，即某些国家的真实的GDP增长是否比其他国家快的问题。