An IMF Course on Macroeconometric Forecasting and Analysis

A course on Macroeconometric Forecasting and Analysis was delivered at the Joint Vienna Institute during 1-12 April 2024 by a team from the IMF Institute for Capacity Development. The program for this two-week offering focused on popular forecasting methods, including Autoregressive Integrated Moving Average (ARIMA) Models, Non-stationarity and Unit Roots, Vector Autoregressive Models (VAR), Structural VAR (SVAR), Bayesian VAR (BVAR), Cointegration, Forecast Evaluation/Combination, Nowcasting, State Space Models, and the Kalman Filter. Each lecture was accompanied by a practical workshop using macroeconomic data with EViews.   

A total of 24 participants from 16 countries in the region, representing ministries of finance, central banks, and other government agencies, attended the course. These government officials received a rigorous foundation in the estimation of macro-econometric models and their application for nowcasting, forecasting, and policy analysis in central banks and various government ministries. Participants appreciated the mixture of theory and practice presented during the course. They noted that the toolkit that was developed will be helpful in their daily work.

In the second week of the course, participants worked in groups on the estimation and forecasting of chosen macroeconomic variables of countries in the region using extensive actual data sets, and discussed and defended their models and conclusions in plenary presentations. This exercise allowed participants to strengthen their understanding of the various concepts presented in the course, apply them to real and monetary sector variables (i.e., real GDP growth, inflation), and test them allowing for current macroeconomic issues (e.g., COVID-19, political instability, oil price shocks).  In doing so, participants experienced many of the difficulties encountered in practice and the need for judgment and strong economic intuition during any forecasting exercise. One of the key lessons learnt is that the best model is often not the most complex model.  Simplicity often provides better results. This message was reinforced by Mr. Karel Musil, Head of the Macroeconomic Forecasting Division at the Czech National Bank, who gave a very enlightening guest lecture on the development and evolution of the Forecasting and Policy Analysis System (FPAS) of the Czech National Bank.  

The course is scheduled to be delivered again during 28 April – 9 May 2025.

Céline Rochon, Senior Economist, IMF