ECON 497A
Forecasting
Spring 2006

Instructor:

Professor Herman J. Bierens
Office: 510 Kern.
Email: hbierens@psu.edu
Office hour: Wednesday 4-5 PM and by appointment.

Teaching assistant:

Hosin Song
Office: 407 Kern.
Email: hus111@psu.edu
Office hours: None.

Place and time:

Place: 107 Ag Sci
Time: Tuesday and Thursday 9:45-11 AM

Goal

The goal of this course is to teach the students how to forecast time series, using econometric software, and what kind of models are useful for that purpose.

Prerequisite level

Officially there is no prerequisite level, but forecasting of time series requires the use of econometrics and econometric software. Therefore, ECON 490, Introduction to Econometrics, or an equivalent level statistics course, is the preferred level. Anyhow, the necessary econometrics will be reviewed or taught, depending on whether all the students have already done ECON 490 or not.

Course material

There is no required text book. Instead, I will use:

Moreover, I will use my free econometrics software package for demonstrations in class. EasyReg (= Easy Regression) is a Windows based software package, which can be freely downloaded from my EasyReg web site. As the name indicates, EasyReg is very easy to use (just point-and-click), and comes with a wide range of web based guided tours.

Homework

Each week I will assign homework of numerical and empirical nature, and occasionally of (mild) theoretical nature. The empirical homework assignments are designed to teach you practical modeling and forecasting skills. The theoretical homework assignments aim to improve your understanding of concepts and to prepare you for the exams. Homework assignments must be submitted in class. Homework submitted by email will not be accepted. No late homework will be accepted.

Software

Most homework assignments will require the use of econometric software. You may use whatever statistical or econometric computer software package that you are familiar with. Several are available on the network, such as Stata and E-Views. However, I recommend to use my free econometrics software package EasyReg International.

Grading

The course grade will be based on the homework (50%), a written midterm exam (25%), and a written final exam (25%). The final exam is cumulative. The large share of the homework in the final grade indicates the hands-on nature of this course. The homework has to be done individually. Turning in copied homework as your own is a breach of academic integrity (see below). Doing the homework assignments is mandatory. If you skip a homework assignment without a valid excuse (see below) then each time the missing homework will be graded zero. If you do all the homework seriously and your score on the final exam is higher than on the midterm exam then the midterm score will be ignored and the final exam will count for 50%. Otherwise, your grade will be based on a weighted average of the homework and exam scores.

There will be no make-up midterm exam. If you miss the midterm exam without valid excuse then your score on this exam will be zero, and the favorable arrangement above will not apply. If you have a valid excuse for missing the midterm exam, the final exam will count for 50%.

The main purpose of the exams is to check whether you understand the lecture notes. Consequently, all the answers to the exam questions are in the lecture notes!

Course outline

  1. Review of regression analysis, with applications to conditional forecasting [Lecture notes on bivariate and multivariate regression; paper.]
    (The dependent variables in this paper are fractions. Click here for a better way to model fractions than has been done in this paper)
  2. How to use EasyReg International.
  3. Key features of time series.
  4. Introduction to time series regression analysis, with applications.
  5. The Box-Jenkins approach to time series modeling and forecasting.
  6. Modeling and forecasting seasonal time series.
  7. Deterministic and random trends, and how to distinguish them.
  8. Modeling and forecasting volatility of financial time series.
  9. Forecasting multivariate time series.

Messages