Artikel [ Article 2007051802 - 18/05/2007]

An accuracy assessment of FPB’s medium-term projections

The outlook is a very detailed macroeconomic projection that covers developments by industry, evolutions in the labour market, in public finances and, in recent years, even in energy consumption and associated greenhouse gas emissions. The prime objective of these projections is not to produce the forecasts that best anticipate the most likely political decisions, but rather to provide, by extending underlying trends, a benchmark scenario pointing to possible future constraints and imbalances that may never materialise if measures are taken. As such, it lends itself less to traditional accuracy evaluation than the short-term forecasts published in the economic budget, which are regularly assessed. The overall coherence of the projection and the quality of the analysis offered to policy makers and to the public in general in terms of diagnosis are probably the main features for gauging its usefulness. Nevertheless, analysing past projection errors and understanding their origins should help improve the quality of future work.

A comparison between projected growth rates and outcome reveals that GDP growth and all its components (except housing investment) were on average overestimated over the period 1987-2005, but only exports exhibit a statistically significant optimistic bias. Growth indomestic prices (deflator of private consumption and GDP) was predicted without systematic error, as was employment growth. Finally, labour force and productivity growth rates were undeniably under- and overestimated respectively. The absolute size of the errors is clearly larger for investment and exports than for consumption, while absolute errors on domestic prices are more limited.

Examination of the role played by the main exogenous variables shows the importance of potential export markets in explaining projection errors on GDP and components. If no error had been made on the future evolution of export markets, the mean absolute error on GDP growth would have been more than halved and large errors would have been absent. Interestingly, although the size of the errors is reduced considerably, the statistically significant bias for exports does not disappear. Losses in market shares, computed as the difference between growth of exports and of potential export markets, have undoubtedly been underestimated in most economic outlooks. Concerning the labour force, the origin of the projection error changed over time. While errors were almost exclusively attributable to errors in the average participation rate at the beginning of the sample, from 1997 onwards the contribution of the latter decreased, while the errors on working-age population contributed relatively more and more to the overall error.

Finally, the paper makes a qualitative assessment of the latest editions of the economic outlook using the main conclusions described above. Methodological improvements made in recent years are stressed and some key future research topics are identified, such as a better understanding of factors driving the evolution of exports and the population.