The Course - The objective of the course is to serve as a refresher for economists and analysts on energy modeling and forecasting issues and introduce them to “recent” developments in the time series econometrics field. The course is designed to be hands-on. Hence, it is assumed that the students have already had a least one semester of econometrics or its equivalent.

Model building is an attempt to characterize the properties of observed energy-economic data using simple parametric relationships which remain reasonably constant over time, account for the findings of previous models, and are interpretable in an energy-economic sense. The lectures will approach applied econometric issues from the LSE or Hendry tradition. The course focuses on three key aspects of empirical model building: data properties, including integration and cointegration; dynamic specification, including the use of error correction models; and model evaluation and design.

The course is an intensive course providing an overview of the OxMetrics system and the econometric features of PcGive. Special attention is given to Autometrics, which implements automatic model selection. The objective is to allow the user to make effective use of the many facilities provided. Some hands-on experience forms part of the course. No experience with OxMetrics is required, but a reasonable level of econometrics will be assumed.

Who should attend ?
The course is aimed at forecasters and researchers in

• Economic Research/ Model Building
• Energy Load Forecasting
• Sales and Inventory Forecasting
• Traffic Modellers
• Financial Modelling/ Arbitrage Trading
• Academic lecturers and researchers
• and more

The course is, therefore, suitable for anyone who works with econometric modelling data in business, industry and government or for those who teach or research time series at universities. Only a basic knowledge in regression and time series analysis is assumed. Familiarity with the software is not required. If you want to get some familiarity with the software prior to the course, please request a demo copy

Day 1: Registration starts at 9H00.

Introduction to OxMetrics and PcGive

• Overview of the OxMetrics system
  • PcGive
  • Ox Console and Ox Professional
  • Other components of OxMetrics
• Exploratory data analysis with OxMetrics
  • loading saving data
  • graphical analysis of time-series properties
  • data transformations* Practical session

Session II Introduction to Dynamic Econometric Models

• Practical Session
  • Univariate econometric models
  • Diagnostic testing and model evaluation
  • Testing for stationarity

Session III Typology of Single Equation Dynamic Econometric Models

• Practical Session
  • Serial Correlation: to Correct or Not?
  • The Wald, Likelihood Ratio, and Lagrange Multiplier Tests
  • Forecasting with a Model
  • PcGive Batch

Session IV Multivariate Modelling with PcGive

• Practical Session
• Formulating the System or VAR
  • Estimation with a VAR
  • Diagnostic Testing and Evaluation

DAY 2: Econometric Modelling with PcGive

Session I Multivariate Modelling with PcGive

• Practical Session
• Cointegration analysis with a VAR
  • Estimation with a VAR
  • Diagnostic Testing and Evaluation
  • Testing for Cointegration
  • Testing the Cointegrating Relation(s)
  • Forming the VECM
  • Forecasting

Session II Autometrics: General-to-Specific Model Building

• Properties of automatic model selection
  • the principles behind Autometrics
  • why never to use stepwise regression
  • the experimental properties of Autometrics

Session III Autometrics: General-to-Specific Model Building

• Practical session
• Getting started with Autometrics
  • automated general-to-specific modelling
  • overview of the facilities provided by PcGive
  • using Autometrics in practice
  • interpreting the output of Autometrics

Session IV Econometric Modelling with PcGive Further Topics

• Practical Session
  • Dynamic Panel Data Modeling 
  • GARCH

Copyright of Timberlake Consultants Limited
Last Revised:9/24/2009