Program

Program

Time Series

Bachelor’s Degree in Economics - Pontificia Universidad Católica Argentina

Course Style

I wholeheartedly believe the best way to guide the learning experience of this subject is a four steps process:

Getting the Motivation

Grasping the historical context in which the idea was born and the state of the art at said point in time, to fully understand the driving forces behind them and why they where received the way they where.

Replicating

Live statistical programming replications of papers from mainstream literature, to empirically validate their relevance regardless of their recognition in mainstream circles.

Repeated Experience

Applying the idea / algorithm to many different cases (datasets) to understand their real life value today.

Knowledge Sharing

Finally, teaching the subject to others in order to consolidate the learning experience and finding elusive blindspots.

1. Objectives

1.1 General

Provide students with an introduction to econometric analysis of time series, reinforcing technical knowledge acquired through understanding the historical context in which they took place, and dimensioning their value through replication and simulation exercises in Python.

1.2 Specific

Enable students to:

  1. Analyze time series, detecting their relevant characteristics;
  2. Fit univariate and multivariate models using datasets with time series realizations in tabular format;
  3. Properly interpret statistical tests and best practices for each case;
  4. Present technical content adequately in oral and written form to both technical and non-technical audiences.

2. Contents

2.1 Module I

Historical approach to linear regression. Review of OLS fundamentals. Introduction to Time Series. Motivation. Basic notions. Stationary and non-stationary series. Strong and weak stationarity. Theoretical and sample functions of autocovariance, autocorrelation, and partial autocorrelation in stationary series. Autocovariance and autocorrelation functions. Typical transformations. Historical context. Box-Jenkins methodology. Model selection based on information criteria. Particular case ARIMA (0,1,1) in connection with simple ETS models. Autoregressive integrated moving average model of order p, d, and q (ARIMA(p,d,q)).

2.2 Module II

Unit Root Tests. Structural Breaks. Lucas Critique. Vector Autoregressive Models (VAR). Historical context. Cointegration and Granger Causality.

3. Evaluation System

To pass the course, students must pass a midterm exam and a final integrative project, along with the completion of weekly assigned activities. The exam will be graded on a scale of 0 to 10 points. An exam will be considered passed when the grade is 4 (four) points or more. To access the final exam, students must pass the midterm and practical work and presentations delivered on time, with only one recovery instance available after taking the two midterm exams. The course grade will be a weighted average of the midterm exam grade (25%), the integrative project (50%), and class participation and completion of activities (25%). Once the course is passed, students must take a final exam. The final exam grade will consist of 80% of the course grade and 20% of the result of the exam taken on the final date. The final exam will only be passed if the exam taken on the day is passed, regardless of the average.

4. Schedule

Classes
  • Twesdays: Puerto Madero from 18.20 to 19.50 pm
  • Fridays: Online Lessons from 18.20 to 19.50 pm
Exams
  • Midterm: Apr 30
  • Project : TBD
  • Final Exam: TBD First call, TBD Second call

5. Bibliography

Year Author Name Link
2018 J. D. Levendis Time Series Econometrics: Learning Through Replication get
1994 J. D. Hamilton Time Series Analysis get
2003 W. Enders Applied Econometric Time Series get
2013 Rob. Hyndman Forecasting: Principles and Practice get