本文为美国密歇根大学（作者：Zhen Wang）的博士论文，共101页。

目前，时间序列在统计学、计量经济学和数理金融等许多领域得到了广泛应用。

Nowadays, time series is widely used inmany real-world applications, such as statistics, econometrics and mathematicalfinance.

时间序列是数据点的序列，通常在连续的时间点上测量，时间点之间的间隔均匀。

Time series is sequence of data points,measured typically at successive time instants spaced at uniform timeintervals.

时间序列的例子如道琼斯指数的日收盘价或美国的年度GDP。

Examples of time series are the dailyclosing value of the Dow Jones index or the annual GDP in US.

时间序列分析涉及分析时间序列数据的技术，以便提取有意义的统计数据和数据的其他特征。

Time series analysis involves techniques ofanalyzing time series data in order to extract meaningful statistics and othercharacteristics of the data.

在本论文中，我们将重点讨论经济学中的面板模型（第二章）和状态空间模型（第三章和第四章）。

In this thesis, we will focus on thescenarios of panel model (Chapter II) and state-space model (Chapter III &IV) in economics.

面板数据（又称纵向或横截面时间序列数据）是指多维数据。

Panel data (also known as longitudinal orcross-sectional time-series data), refers to multi-dimensional data.

面板数据包含对同一对象在多个时间段内的多个现象的观测。

Panel data contains observations onmultiple phenomena observed over multiple time periods for the sameobservation.

观测对象可能是家庭、企业、地理区域，或者任何一组随时间变化而保留其身份的实体。

The observation may be households, orenterprises, or geographical areas, or indeed any set of entities that retaintheir identities over time.

面板数据允许您控制单个未观测到的异质性，如国家政策、联邦法规、国际协议等。

Panel data allows you to control forindividual unobserved heterogeneity, eg., national policies, federalregulations, international agreements, etc.

使用面板数据，您可以在不同分析级别（即个人、州、国家等）包括适合多级或分层建模的变量。

With panel data you can include variablesat different levels of analysis (i.e., individual, states, countries, etc.)suitable for multilevel or hierarchical modeling.

缺点是在微面板的情况下没有响应，或者在宏面板的情况下存在跨状态依赖（即状态之间的相关性）。

Some drawbacks are non-response in the caseof micro panels or cross states dependency in the case of macro panels (i.e.,correlation between states).

1 引言

2 非线性趋势面板分析揭示宏观经济对死亡率的影响

3 MIF和PMCMC在DSGE模型估计中的比较

4 p-MIF

5 未来工作展望

更多精彩文章请关注公众号：