Whether you deal with forecasting at a Central Bank, public institution, bank or consultancy firm; or you use forecasting techniques in your research, this is the perfect course to bring you up to date with the latest methods in the forecasting profession. We begin softly by reviewing some classic time series methods and standard point and density forecasting tools (fan charts), but rapidly turn to the state-of-the-art forecasting methods such as Mixed Frequency Data Sampling (MIDAS), Regime (or Markov) Switching models, and Bayesian forecasting techniques.
The focus will be more on the empirical implementation of the techniques than on their theoretical underpinnings. The techniques will be illustrated with several empirical applications, and then implemented in EViews 9. While experience in forecasting is advantageous, the course is equally suitable for professionals who have just recently began to forecast macroeconomic and financial indicators. We are flexible and the course can easily be accommodated to the level of the participants. Previous knowledge and experience in econometrics is however, essential.
This course is aimed at:
- Economists and statisticians at Central Banks, public institutions, financial institutions, consultancy firms, or firms who deal with forecasting in their daily work;
- Academics and research economists who use, or are interested in forecasting techniques for their research;
- Professionals involved in rating activities.
Day 1: Introduction to forecasting
1. Forecasting with univariate linear models
- Review of basic forecasting formulae for the linear regression model;
- ARMA models: specification, estimation, testing and forecasting;
- Properties of forecasts from ARMA models;
- Forecasting with integrated variables;
- Multistep estimation vs iterated formulae for h-step ahead forecasting;
- Rolling vs. recursive forecasting schemes;
- Example: AR vs Leading indicator forecasts for euro area GDP and inflation;
- Example: h-step ahead forecasts for US macro variables;
- Empirical example (in EViews): Forecasting the Conference Board Composite Coincident Indicator.
2. Forecast evaluation, comparison and pooling
- Point and density forecast evaluation;
- Comparing alternative forecasts;
- Pooling alternative forecasts;
- Example: Forecast pooling for short time series of macroeconomic variables;
- Empirical example (in EViews): Forecasting the Conference Board Composite Coincident Indicator (forecast evaluation and combination).
Day 2: Advanced topics
1. VARs and Bayesian VARs
- VAR models: specification, estimation and testing;
- VAR models: forecasting and forecast error variance decomposition;
- Forecasting with cointegrated variables;
- Forecasting the aggregate vs aggregating the forecasts;
- Bayesian Vector Autoregressions;
- Example: Country specific vs. euro area forecasts for euro area IP, inflation and unemployment;
- Empirical example (in EViews): Forecasting the Conference Board Composite Coincident Indicator (Effects of cointegration and aggregation).
2. Models for forecasting volatility
- Estimation and specification of GARCH models;
- E-GARCH models;
- Constructing point, interval and density forecasts;
- Example: comparison of several forecasting models for the volatility of US inflation.
Day 3: Regime switching models, mixed frequency data, nowcasting
1. Forecasting with smooth transition and threshold autoregressive models
- Estimation, specification and testing;
- Constructing point, interval and density forecasts;
- Example: comparison of several forecasting models for euro area and US macro variables;
- Example: a multivariate TAR model for forecasting GDP growth;
- Empirical example (in EViews): comparison of linear and nonlinear models for forecasting inflation and GDP growth.
2. Forecasting with Markov switching models
- Introduction to Markov chains;
- Models with intercept switches;
- Markov switching models;
- Forecasting levels, regimes, and regime duration;
- Example: a multivariate MS-ECM of the UK labour market;
- Empirical example (in EViews): comparison of linear and nonlinear models for forecasting inflation and GDP growth (continued).
3. Forecasting with mixed frequency data
- Bridge models;
- MIDAS models: specification, estimation, forecasting;
- Unrestricted MIDAS models;
- Example: a MIDAS model for forecasting US GDP growth using monthly indicators;
- Empirical example (in EViews): Forecasting quarterly GDP growth using monthly indicators.
This course has a learning ratio of approximately 70% practical to 30% theory.
The EViews Help Files (PDF manuals) provide detailed information regarding each topic, so it is advised that delegates familiarise themselves with the manual readings to be prepared for the course. Click here, to download the EViews 9 Manuals (.zip) via www.eviews.com (48mb).
Andrea Carriero will provide participants with suggested supplementary reading during the course.
- Intermediate to advanced level University training (or equivalent) in econometrics is essential.
- Intermediate level University training (or equivalent) in macroeconomics and/or finance is also essential.
- Experience in forecasting is advantageous.
- Familiarity with EViews fundamentals is required.
For full Training Courses Terms & Conditions please click here.
Payment of course fees required prior to the course start date.
Registration closes 5-calendar days prior to the start of the course.
- 100% fee returned for cancellations made over 28-calendar days prior to start of the course.
- 50% fee returned for cancellations made 14-calendar days prior to the start of the course.
- No fee returned for cancellations made less than 14-calendar days prior to the start of the course.