课程内容简介 这是一门应用学科, 主要学习如何分析时间序列的数据. 时间序列指的是一组按照时间顺序观测到的数字序列. 时间序列分析是宏观经济学和金融学研究中重要的分析工具. 本课程将介绍时间序列模型的建立和应用, 使用时间序列模型来预测经济和金融变量未来的值, 及识别已获得数据的模型结构. 时间序列的建模方法主要有两种类型. 其一为时域方法. 该方法将经济和金融变量现在或未来的值理解为过去观测值的线性函数(加上一个误差项). 该方法得到的模型被称为ARIMA模型, 也叫做Box-Jenkins模型. 另一时间序列的建模方法为频域方法. 该方法利用了付立叶变换将平稳过程的时间变量变换为角频率, 因而得到平稳过程的谱表示, 主要的分析工具是谱分析, 该方法将使经济与金融的理论研究更简便更容易. 本课程的基本内容包括: 白噪声, 鞅差分序列, 混合鞅, 随机漫步, 自回归过程, 移动平均过程, ARMA模型, 渐进收敛理论, 极小平方估计方法, 最大似然估计方法, ARMA模型的预测, ARMA模型的识别, ARCH模型, GARCH模型, 单位根过程, 及协整理论等时间序列分析的核心内容.通过本课程的学习,学生将能够理解基本的时间序列模型,并能够利用这些模型对实际的经济与金融问题进行实证研究. | 课程内容简介(英文) This is an applied course, meaning that we will focus on how to analyze time series data. A time series is a sequence of observations of one or more variables collected repeatedly over time. Although there is no official definition of how many observed times constitute a time series, many (most) time series methods work best for series of at least 50-60 observations. We will consider the methods for building and using models for time series data, use time series models to forecast future values of the variable(s), and learn about the structure of the pattern in past observations. The course will cover the two fundamental approaches to time series modeling. The first approach obtains time domain models in which a present or future observation is considered to be a function of past observations (and errors). Models used in this approach are from the class of models called ARIMA models, and they are also called Box-Jenkins models. The second approach leads to frequency domain models in which the time series is modeled as a combination of systematic sinusoidal patterns (sine and cosine waves). The principal tool in this approach is extensively known as the spectral analysis. Basic contents of this course include: white noise, martingale difference sequence, mixingale, random walk, autoregressive model, moving average model, ARMA model, asymptotical convergence theory, least square estimation, maximum likelihood estimation, forecast of ARMA model, identification of ARMA model, ARCH model, GARCH model, unit root process, and cointegration theory.Students can use the methods introduced in this course to do empirical research in finance and macroeconomics by learning this course. | 教学大纲 第一讲 随机过程的定义内容:随机过程, 平稳过程, 及平稳过程的遍历性. 第二讲 ARMA模型内容:白噪声, 鞅差分序列, 鞅, 随机漫步, 混合鞅; 自回归过程, 移动平均过程, ARMA过程.第三讲 ARMA过程的性质内容: 自相关函数, 偏自相关函数, 及脉冲相应; 滞后(Lag)算子.第四讲 谱分析 内容; 平稳过程的谱表示, 及谱密度与ARMA过程第五讲 渐进收敛定理 内容: 关于鞅差分序列, 及混合鞅的大数定律和中心极限定理.第六讲 ARMA模型的预测 内容: 条件数学期望预测, 及线性预测.第七讲 ARMA过程的估计内容: 极小平方估计第八讲 ARMA过程的估计内容: 极小平方估计, 与极大似然估计第九讲 ARMA过程的识别 内容: ARIMA模型的Box-Jenkins标准方法第十讲 GARCH模型 内容: 条件异方差, ARCH模型, 及GARCH模型.第十一讲: 单位根过程内容: 经济和金融数据的趋势, 及单位根检验第十二讲: 协整理论内容: 协整的概念, Granger定理, 及协整检验. | 课程进度计划 (无) | 课程考核要求 上课出席, 讨论, 及作业40%; 期末完成一篇论文60%. | 参 考 文 献 - 1. 1. James D. Hamilton, Time Series Analysis, Princeton University Press, 1994.2. 2. Ruey S. Tsay, Analysis of Financial Time Series, John Wiley & Sons Inc., 2005
|
|