This is a great opportunity for students, academics and professionals to expand their forecasting skills and learn how they can apply a range of techniques. All courses will teach forecasting from an applied perspective and demonstrate techniques using EViews software.



Course 1: Introduction to EViews

Level: Introductory

Learning ratio: 90% Practical; 10% Theory

The course introduces EViews most popular and useful commands and procedures to import, manipulate, transform and manage data, as well as to perform some commonly used statistical routines and econometric estimations. This session is ideal for new or beginner EViews users, and we strongly recommend any participant with no prior experience of using EViews and planning to take further courses, should attend this course.


Course 2: Univariate forecasting with EViews

Level: Introductory

Learning ratio: 70% Practical; 30% Theory

The basics of forecasting (and of using EViews) are introduced. The day contains a fair balance of theory and application, and both the ordinary regression model and ARMA models are introduced.

Session 1: Introduction

•The notions of “workfile” and “object” in EViews

•Data handling and databases in EViews

•Data description: creating, editing, freezing and exporting graphs

•Descriptive statistics and tests


Session 2: Univariate forecasting I: Structural models

•Preliminary theory for univariate regression: the classical linear regression model (CLRM), the CLRM assumptions, OLS estimation

•Creating, estimating and reading a regression in EViews


Session 3: Univariate forecasting II: Atheoretical models

•ARMA models: Box-Jenkins identification, estimation and forecasting



Session 4: Measuring forecasting performance

•Some indicators of forecasting accuracy: mean squared and mean absolute errors, Theil’s U

•The Diebold-Mariano test


Course 3: Multivariate forecasting with EViews

Level: Intermediate

Learning ratio: 50% Practical; 50% Theory

The first half of the day (more theoretical) presents a more advance look at the ordinary regression model, focusing on the issue of misspecification – i.e., from the point of view of forecasting, the issue of improving a regression model (and/or an estimation technique). In the second half of the day (more practical), the notions of ARMA models and regression are extended to a multivariate context.


Session 1: Univariate forecasting III: Misspecification tests

•More on the CLRM assumptions

•Testing for misspecification: heteroskedasticity, normality, functional form


Session 2: Univariate forecasting IV: Dynamic models

•Specification of dynamic models

•Testing for structural instability


Session 3: Multivariate forecasting I: Stationary VARs

•VAR representation and estimation

•Further testing with multivariate regression: Granger causality, lag selection


Session 4: Multivariate forecasting II: Forecasting stationary VARs

•The notion of “model”, static and dynamic prediction

•Indicators of forecasting accuracy


Course 4: Forecasting non-stationary series

Level: Intermediate

Learning ratio: 50% Practical; 50% Theory

The main focus here is on non-stationary data. After introducing the notions of stationarity and non-stationarity (and some tests), we discuss cointegration in both a single equation and multiple equation framework.


Session 1: Non-stationarity I: Unit roots

•Introduction to the notion stationarity and unit roots

•The Dickey-Fuller test


Session 2: Non-stationarity II: Cointegration

•Introduction to the notion of cointegration: preliminary theory and Engle-Granger analysis using EViews


Session 3: Multivariate forecasting II: Vector error correction models

•Cointegrated VARs in EViews: Johansen’s test for cointegration, the (vector) error correction model (VECM), estimating and interpreting a VECM in EViews


Session 4: Multivariate forecasting II: Forecasting using the VECM

•Creating and interpreting forecasts with non-stationary VARs

•Static and dynamic prediction

•Modelling and predicting UK inflation


Course 5: State space modelling in EViews

Level: Intermediate / Advanced

Learning ratio: 50% Practical; 50% Theory

The sessions are mainly practical in this case, and it could be worth revising the theory either before the session or after. A very good reading is Chapter 13 in: Hamilton, J.D., (1994). Time Series Analysis, Princeton University Press. The sessions will cover: setting up a State-space model in EViews, forecasting and getting around some typical problems (mainly, trying to achieve convergence). The main focus will be on a special case of a State-space model, namely the time-varying parameter case. It is natural to link this topic with the discussion on structural instability during Course 3.


Session 1: State space models I: Representation

•Representation and creation of a state space object in EViews

•Unobserved components


Session 2: State space models II: Estimation

•Estimation of a state space object

•Setting initial values


Session 3: State space models III: The time-varying parameter model

•Setting up and estimating a time-varying parameter model in EViews


Session 4: State-space models IV: Forecasting

•The “state space” object procedures

•Forecasting states and signals

•Kalman-Bucy filtering

•Forecasting parameters

Start Date:

End Date:

Application Deadline:

Expired help


Summer schools


London , United Kingdom

Start Date:

End Date:


United Kingdom



Summer schools

Application Deadline:

Expired help