Page Title


Dans un souci de transparence et d’information, le BFP publie régulièrement les méthodes et résultats de ses travaux. Les publications sont organisées en séries, entre autres, les perspectives, les working papers et planning papers. Certains rapports peuvent également être consultés ici, de même que les bulletins du Short Term Update publiés jusqu’en 2015. Une recherche par thématique, type de publication, auteur et année vous est proposée.

STU 04-98 : Special Topic - The accuracy of the FPB short-term economic forecasts since 1994 [Short Term Update 04-98]

The FPB is reassessing the state of the economy in 1998 and its possible evolution for 1999.

In 1998 the Belgian economy has continued to grow strongly and has moved into a “mature” phase of recovery with exports and investment no longer providing the engine for growth. Private consumption, fed mainly by employment growth, moderate real wage increases and high consumer confidence, took over their role. Employment growth remains impressive.

The outlook for the world economy for 1999 has deteriorated: the Asian crisis has widened and deepened and contagion effects have started to affect also Russia, Latin American countries and, to a lesser extent, Eastern Europe. World financial markets have shown extreme volatility. Continental European countries will be affected by the deterioration of the global economic performance and the weakening of the USD, but should nevertheless become the fastest growing area in the world.

Any forecast concerning Belgium is fragile in this context but it seems likely that the GDP-growth forecast for 1999 given in July (2.6%) is too optimistic. The Belgian economy might not be growing faster than 2.2% with significant downward risks on the domestic and international side.

Many uncertainties and downward risks regarding the international environment are linked, and, given the interdependencies in the global economy could trigger all the others and lead to a sharp deterioration in the overall economic situation.

Export growth should be significantly lower than in 1998 while private consumption should be less affected. Employment should still increase by 0.8% and the unemployment rate should further fall from 8.6% to 8.3% (Eurostat standardised definition).

In any case, consumption price inflation remains subdued at about 0.9% (1% for the “health” index). Wage increases will remain moderate, under the influence of the “wage norm”. Interest rates in Belgium drop in line with international rates. This should be a positive factor for domestic demand.

The activities of the FPB traditionally focused on the medium and long term economic developments. Short-term forecasting is a rather new discipline at the Bureau. Only since July 1993 1 has the FPB made annual short-term forecasts twice a year. As a new producer of short-term forecasts, it is useful to perform some quality con-trol and to publish the results as a kind of consumer guidance. The main “consumer” of the short-term eco-nomic forecasts of the FPB is the Federal Government. In particular, the FPB short-term economic forecasts form the macro-economic environment for establishing and following up the Federal Government’s budgets.

Three questions will be dealt with in this article. Firstly, how good are the forecasts? Secondly, do the forecasts improve as more information on the economic situation of the year concerned becomes available? Thirdly, is there a systematic bias in the forecasts?

Comparing forecasts...

Twice a year the FPB makes annual short-term forecasts, once in July and once in February. In the July publication, forecasts for the current (t) and the next year (t+1) are made. In the February publication, the July forecasts (for the past (t-1) and the current (t) year) are updated. In this way, four forecasts are available for each year, of which the first two are of particular interest, as they are used for the Government budget and for the Government budgetary control respectively. For 1994, for example, forecasts were made in July 1993, February 1994, July 1994 and February 1995. The first three exercises can be considered as “forecasts” in the proper sense of the word, while the fourth exercise can rather be called an “estimate”.

... with actual outcomes...

Forecasts are to be compared to actual outcomes. The notion of “actual outcome”, however, is not straightforward. Most of the macro-economic data that will be looked at are published in the national accounts. Revisions of the national accounts data may occur, and this for several reasons: more complete and correct data become available, methodological changes take place or a new base year for data at constant prices is introduced. As methodological modifications are difficult to foresee and in order to exclude at most the effect of a changing base year on past observations, actual outcomes are here defined as the figures published in the first version of the national accounts of the year concerned (generally dating from June of the year t+1, for 1997 dating from the autumn of 1998).

Forecasts are available from 1994 on, outcomes until 1997. For each year, four forecasts are available. As the number of years considered in this article is limited, all conclusions have to be interpreted with caution.

... a simple problem?

Assessing the accuracy of forecasts ex post may seem to be a simple problem: just measure the distance between the forecasts and the known outcome. Apart from the aforementioned problem of the definition of actual outcomes, a more fundamental problem arises when one wishes to interpret the measured errors. In fact, not all forecasting errors are due to an imperfect forecasting method. Other factors can also explain part of the forecasting errors. FPB forecasts, for instance, are deliberately based on the assumption of “unchanged government policy”, even if government intervention is very likely. Moreover, for most variables concerning the international environment (such as exchange and interest rates, world prices and export market growth) the FPB traditionally uses an exogenous scenario, delivered by international institutions such as the EC and partly based on technical assumptions. For a small, open economy like Belgium, disagreement between these assumed international scenarios and the outcomes can be an important source of forecasting errors. It follows that a sound risk analysis (assessing the sense and the size of the uncertainties) related to the underlying assumptions of the forecasts is equally important as an accurate point estimate itself.


The accuracy of the FPB forecasts has been tested for the following data series: real GDP growth and its two components, namely domestic demand (DD) and net exports (XM), consumer price inflation (CPI) and the moment of indexation for public wages, employment growth (EMP) and the general government deficit (DEF). Figures on the government account are traditionally not published in the short-term economic forecasts but were available in internal databases.

  Documents associés


  Données à consulter





  Mots clés


Please do not visit, its a trap for bots