Trending Economic Time Series and Panels

经济数据的一个主要特征是它们有随时间变化的趋势。这种行为包括利率、汇率和通胀的游荡性质及其漫长而不稳定的周期，以及GDP等系列的长期漂移。重要的社会经济问题，如国家经济增长、增长的决定因素以及各国之间的增长差距，都与趋势问题有关。许多现代时间序列计量经济学以及最近关于面板数据的工作都与这种趋势行为的统计分析有关，包括不同序列之间趋势可能存在的相互关联。经济学理论提供了对趋势和联系的潜在来源的洞察，但它几乎没有就实证研究的合适公式提供实际指导。这项研究将模拟这种情况下经济数据的趋势，并开发一种统计推断理论，允许对应用工作中的趋势进行错误的说明。该项目将开发一种研究常用拟合趋势函数的正式仪器，并开发在坐标系的背景下有效解释这些拟合趋势的方法，该坐标系旨在根据更一般的维持假设来捕捉数据中的趋势和长期周期行为。使用坐标函数，可以分析单个系列的趋势和几个系列的共同运动，而只对生成机构的类别施加较弱的限制。这项研究将为这一方法开发新的计量经济学方法，并探索它们与为真实生成过程已知的情况下开发的传统方法的关系。这项研究将涉及用于推断和趋势预测的渐近理论，用于评估方法性能特征的模拟实验，以及使用宏观经济和金融数据集的实证应用。这些方法将有助于我们理解与经济增长相关的更广泛的社会经济问题，它们将被用于评估关于世界经济在开始收敛之前的过渡时期出现分歧的假设的经验证据。因此，这项研究的结果将使社会科学家和统计学家能够建立模型并正确分析数据，这些数据的趋势到目前为止还知之甚少。

A dominant characteristic of economic data is their tendency to trend over time. This behavior includes the wandering nature of interest rates, exchange rates, and inflation with their long and erratic cycles, as well as the secular drift of series like GDP. Important socioeconomic issues such as national economic growth, the determinants of growth, and disparities in growth across nations all relate to matters of trend. Much of modern time series econometrics, as well as recent work on panel data, is concerned with the statistical analysis of such trend behavior, including the possible interconnectedness of the trends across different series. Economic theory provides insights about potential sources of trend and linkages but it gives little practical guidance concerning suitable formulations for empirical research. This research will model trends in economic data in such situations and develop a theory of statistical inference that allows for incorrect specification of the trends in applied work. The project will develop a formal apparatus for studying commonly fitted trend functions and develop methods for validly interpreting these fitted trends within the context of a coordinate system that is designed to capture trends and long cyclical behavior in the data under a much more general maintained hypothesis. Using coordinate functions, it is possible to analyze trends in a single series and co-movement in several series while placing only weak restrictions on the class of generating mechanisms. The research will develop new econometric methods for this approach and explore their relationship to conventional methods developed for cases where the true generating process is known. The research will involve an asymptotic theory for inference and for trend forecasting, simulation experiments to evaluate performance characteristics of the methods, and empirical applications with both macro-economic and financial data sets. The methods will contribute to our understanding of wider socioeconomic issues related to economic growth and they will be used to assess empirical evidence on the hypothesis that the world economy diverges during a period of transition before it starts to converge. Thus, the results of this research will allow social scientists and statistician to model and correctly analyze data whose trends had hitherto been poorly understood.