Page Title


To promote transparency and provide information, the Federal Planning Bureau regularly publishes the methods and results of its works. The publications are organised in different series, such as Outlooks, Working Papers and Planning Papers. Some reports can be consulted here, along with the Short Term Update newsletters that were published until 2015. You can search our publications by theme, publication type, author and year.

STU 01-07 : Special Topic : An accuracy assessment of FPB’s medium-term projections [Short Term Update 01-07]

In the course of 2006, quarterly economic growth in Belgium slowed down in line with the international business cycle from 0.9% in the first quarter to 0.6% in the last quarter. This year, qoq GDP growth should stabilise around 0.5%. On a yearly basis, economic growth should slow down from 3% in 2006 to 2.2% in 2007.

In 2006, economic growth was only supported by domestic demand while net exports contributed negatively to it. In 2007, however, both domestic demand and net exports should support GDP growth. Despite the deceleration in quarterly Belgian export growth due to the slowdown of the euro area and the US economy, annual average export growth should rise to 6.5% in 2007 as it benefits from a considerable carry-over from 2006. After several years of decrease, the current account surplus should rise by 0.4% of GDP in 2007, mainly as a result of the decline in oil prices leading to an improvement in the terms of trade. Domestic demand growth should weaken this year, which is essentially due to the evolution of private consumption and public investment. Private consumption growth should be less buoyant than in 2006 as the personal income tax reform then gave its final boost to real disposable income. Public investment rose markedly last year in view of the local elections in October 2006, but should fall by the same extent in 2007.

After a net gain of about 44,000 persons in 2006, employment is expected to record an average annual rise of 45,600 persons this year. As the number of jobs grows faster than the labour force, the broad administrative unemployment rate is expected to decline from 13.9% in 2006 to 13.5% in 2007. The harmonised Eurostat unemployment rate (based on labour force surveys) should fall from 8.3% in 2006 to 7.9% next year.

This year, the increase in the national index of consumer prices (NICP) should amount to 1.8%, just as in 2006. It should be noted that the inflation picture in 2006 was blurred by the introduction of a new NICP-basket. The rise of the private consumption deflator, which is not affected by this factor, should decline from 2.3% in 2006 to 1.8% in 2007, mainly due to the decrease in oil prices.

An accuracy assessment of FPB’s medium-term projections

The Federal Planning Bureau has been publishing medium-term macroeconomic projections for the Belgian economy since the beginning of the eighties. The prime objective of these projections is not to produce the forecasts that best anticipate the most likely policy decisions, but rather to provide, by extending underlying trends, a benchmark scenario pointing to possible future constraints and imbalances that may never materialise if prompt measures are taken. The quality of the analysis offered to policy makers and to the public in general in terms of diagnosis is probably the main criterion to apply to gauge its usefulness. However, in this special topic past projection errors are scrutinised to give to users a broad idea of the uncertainties surrounding the figures and also to identify possible methodological weaknesses that should be improved.

The economic outlook is a very detailed macroeconomic projection that covers developments in industry, the labour market, public finances and, in recent years, even energy consumption and associated greenhouse gas emissions. The tradition at the FPB of producing a baseline simulation with a medium-term horizon was inherited from the (failed) indicative planning experiments in the seventies. This baseline is a no-policy-change scenario, notably with regard to fiscal and social policies, that is based upon an international environment founded on projections prepared by international institutions such as the European Commission or the OECD. Scenario analysis is sometimes performed to illustrate the potential impact of risks included in the baseline or to analyse the effects of changes in economic policy. The baseline and variants for the Belgian economy are produced using the HERMES model. The economic outlook has been released in April since the mid-nineties, with an update in October in recent years used as the macroeconomic framework for the Belgian stability programme. Previously, a first version was published in February-March and an updated version in July.

Comparison between projected growth rates and outcome

The April releases (before 1996, the July issues) were used in the following analysis as we chose to select only one projection per year in our data sample, which starts, for technical reasons, with the outlook of July 1986 (covering the period 1987-1990) and ends with the April 2001 release (covering the period 2002-2006). Subsequent editions cannot be fully evaluated yet due to the lack of hard data beyond 2005. To analyse the projection errors we decided to work with growth rate averages covering the projection period. Until the 1997 release, the projection period covered four years beyond the year of the projection; from 1998 one extra year was added. In the analysis discussed below we examine four-year geometric growth rate averages beyond the current year for all releases.

A traditional problem faced when measuring forecasting errors is the choice of the data vintage to be defined as the outcome, as most macroeconomic data, notably national accounts, are regularly revised when new information sets become available or when methodological changes are introduced. As these factors can hardly be anticipated, we chose to compare the projection for year t with the estimate available in the database of year t+2. For instance, we defined as an outcome for the year 1990 the data available in spring 1992. This choice is based on the calendar of the national accounts as that data vintage will entail the first release of the national accounts for 1990 published in autumn 1991.

The key descriptive statistics used to assess the average growth rate errors, constructed as described above, are given in Table 1 for a selection of macroeconomic variables. Fiscal variables are not presented here, as their projections under the “no-policy-change” assumption require a different approach for investigation. The mean error in the first column indicates by how much the projected growth rates were on average overestimated (negative sign) or underestimated (positive sign) over the period 1987-2005. A desirable property of forecasts is unbiasedness, meaning that negative and positive errors
should cancel out: this property has been examined by testing if the mean error was not statistically significantly different from zero (see next column No bias). Potential export market growth assumptions show a tendency to be too optimistic as all components of GDP (except housing investment) and disposable income, but only the projections for exports exhibit a statistically significant bias (imports being a borderline case). Growth of international prices expressed in dollars was also on average overestimated, while future growth in the exchange rate and domestic prices was predicted without systematic error, as was employment growth. Finally, the two supply-factor variables, namely labour force and productivity, were undeniably under- and overestimated respectively.

  Related documents


  Consult data








  PDF & Download


Collective publication, , , , ,
A : Author, C : Contributor


  Publication type

Short Term Update

Please do not visit, its a trap for bots