17277 - Forecasting Techniques

Academic Year 2012/2013

  • Teaching Mode: Traditional lectures
  • Campus: Bologna
  • Corso: First cycle degree programme (L) in STATISTICAL SCIENCES (cod. 8054)

Course contents

General concepts.Statistical and theoretical information. Quantitative and qualitative methods. Forecast horizon, loss function, information set and predictability. Exogeneity.

Econometric models for short-term forecast. Dynamic models: estimation, testing, specification analysis and forecasting. Multivariate models: vector autoregressive models (VAR), impulse-response functions, variance decomposition.

Econometric models for long-term forecast. Stationarity and non-stationarity. Deterministic and stochastic trends. Modeling trends.

Choosing the predictor. Ex-post assessment of forecast ability. Choosing among many predictors: the Diebold-Mariano test. Forecast combination.

Intercept correction. Forecasting under structural breaks and time-varying parameters. Forecasting using intercept corrections.

Forecasting in practice. Defining the economic problem of interest. Collecting statistical data. Choosing the econometric model. Empirical analysis and economic interpretation of the results.

Readings/Bibliography

F.X. Diebold (2001) "Elements of forecasting", South Western/Thomson Learning.

Teaching methods

Computer lab

Assessment methods

Oral and written exam

Teaching tools

All topics are illustrated with economic data using some econometric packages

Links to further information

http://www2.stat.unibo.it/cavaliere

Office hours

See the website of Giuseppe Cavaliere