Course Description
An introduction to the statistical analysis of time series data. Focus is on fundamental models of time series processes and how these models can be used for forecasting and influence. Although some statistical theory is necessary and will be developed, the main thrust involves applying models to the data. Because data analysis will rely on the R statistical programming language, the basics of that language will also be covered.
Additional Requirements for Graduate Students:
In addition to submitting all work required of undergraduate
students, graduate students in the course must submit a completed
research project on a topic related to the course material that
demonstrates a broader and deeper understanding of that material
than is required of undergraduate students. The research paper
will incorporate a thorough review of the relevant literature,
will rely on extensive knowledge of the quantitative tools
developed in the course, and will be assessed in terms of
demonstrated competence in synthesizing, criticizing, and
extending knowledge in the field. Overall, graduate students will
be held to the high standards of scholarship that guide the
Graduate School and will be expected to exhibit a mastery of
skills that goes beyond the learning outcomes for undergraduate
students.
Athena Title
Time Series Analysis
Undergraduate Prerequisite
ECON(MARK) 4750/6750
Graduate Prerequisite
ECON(MARK) 4750/6750
Semester Course Offered
Offered every year.
Grading System
A - F (Traditional)
Course Objectives
After successfully completing the course, students will 1. understand the fundamental difference between time series processes and cross-sectional data; 2. know the basic concepts of time series processes and how to apply these concepts to time series data; 3. know how to specify and estimate autoregressive-moving average (ARMA) models and how to use such models for forecasting; 4. know how to specify and estimate models of time-varying conditional variance; 5. understand the implications for modeling and forecasting of non-stationary processes; 6. know how to estimate and specify multivariate time series models of stationary and non-stationary processes; 7. know when to consider and apply simple non-linear time series models; 8. have a working knowledge of the R programming language.
Topical Outline
1. Preliminaries a. Introduction to time series i. time-series versus cross-section samples ii. stochastic processes iii. time series models iv. forecasting and inference b. Fundamental concepts in time series analysis i. sequences and convergence ii. stationarity and ergodicity iii. auto-correlation, serial independence, and white noise iv. review of linear regression with time series samples 2. Univariate time series models a. ARMA models i. AR models and stationarity conditions ii. MA models and impulse response functions iii. specification and estimation iv. forecasting b. Time-varying volatility (GARCH) models c. Univariate models of non-stationary processes i. deterministic trends ii. stochastic trends, unit roots, and ARIMA models 3. Multivariate time series models a. VAR models i. notation and the nature of multivariate extensions ii. specification and estimation iii. innovation accounting iv. forecasting b. Multivariate models of non-stationary processes i. cointegration ii. vector-error-correction models 4. Non-linear time series a. implications of non-linearity b. the bilinear model c. threshold AR models
Syllabus