An Introduction to Time Series Forecasting in Economics (Doctoral & Master Course)

Syllabus “An Introduction to Time Series Forecasting in Economics”

Doctoral & Master Course

University of Münster, School of Business and Economics

December 6-9, 2022

Lecturers

Dr. Alexander Pütz: alexander.puetz@wiwi.uni-muenster.de

Elissa Iorgulescu, M.Sc.: elissa.iorgulescu@wiwi.uni-muenster.de

Course description

The primary goal of this course is to provide a useful overview to PhD students in business and economics as well as master students seeking a point of entry into the field of time series forecasting from a practitioner’s point of view. The course will cover a selection of theory and methods underlying forecasting as currently practiced in economics and finance such as traditional linear models, forecasts combinations, forecasting in data-rich environments etc. Students will learn the importance of forecasting in economics/finance, how to construct, compare and evaluate a forecast, and finally, how to choose the best forecasting model for some underlying data.

Throughout the course we will use real time series data from the fields of economics and finance and the statistical software R to implement the discussed forecasting methods. Students may use other statistical software for the class (Python or Stata etc.).

Exam

There will be no written exam at the end of the semester.

Grading will be based on an assignment which consists of a forecasting exercise with real data and a short term paper (max. 4-5 pages) – briefly explaining the data, methods and forecasting results. Deadline for the assignment is the 31st of January.

Prerequisites

Basic knowledge in R and econometric methods.

Registration

The course is limited to 15 participants. Course slots are assigned according to a “first-come-first-served” principle. For registration, please send an email to Elissa (elissa.iorgulescu@wiwi.uni-muenster.de) by no later than 15th of September. Please indicate your matriculation number and whether you attend as a master* or PhD** student.

*Master students: please do not forget to register for the course at the examination office (Ausgewählte Themen der Volkswirtschaftslehre 1-4) during the regular registration period. Your grade will be based on your active participation throughout the course (20%) and assignment (80%).

**PhD students: in order to obtain a certificate (neue PO: Methodenkurs or alte PO: A-Schein) you need to attend all the lectures.

Important: Course material will be provided entirely via Learnweb. Only those who register will have access to the course material and receive the enrolment key.

Dates & Times

Date

Time

Topic

Room

Tue. 6.12

9:30-12:00

13:30-15:30

1+2+

Lab

LEO 11.9 (Leonardo-Campus 11)

Wed. 7.12

9:30-12:00

13:30-15:30

3+

Lab

LEO 11.9 (Leonardo-Campus 11)

Thur. 8.12

9:30-12:00

13:30-15:30

4+

Lab

LEO 11.9 (Leonardo-Campus 11)

Fri. 9.12

9:30-12:00

13:30-15:30

5+

Lab

LEO 11.9 (Leonardo-Campus 11)

Outline

1. Introduction

     1.1 What is forecasting and for whom?

     1.2 Loss function

2. Time series models

    2.1 Univariate linear prediction models: AR (MA) X

    2.2 Vector autoregressions

3. Forecast evaluation

    3.1 MSE

    3.2 MAE

    3.3 MD

4. Forecasts comparisons and combinations

     4.1 Test of equivalent expected loss: Diebold-Marino test

     4.2 Optimal forecast combination

     4.3 Model combination

     4.4 Cross-validation

     4.5 Out-of-sample forecast

5. Forecasting in a data-rich environment

     5.1 Ridge regression

     5.2 LASSO

     5.3 Elastic net

     5.4 Principal Components