STU 02-03 : Special Topic - Estimating potential growth in Belgium [ Short Term Update 02-03 - 15/06/2003]

Economic activity remained subdued in the euro area in the last quarter of 2002 and early estimates point to a stabilisation in the first quarter of the current year. International organizations are forecasting a gradual but only modest recovery in the course of 2003. In Belgium, GDP growth was higher than in the main neighbouring countries in the last quarter of 2002. This should also be the case in the first quarter of 2003. The FPB leading indicator for Belgium confirms the scenario of a recovery during the course of 2003. Annual GDP growth should nevertheless be only slightly above 1% this year.

Various risks could jeopardise the recovery in the euro zone: the continuing depreciation of the USD, and a slower recovery of confidence due to the situation in the labour market and/or the stock market. The medium-term outlook for Belgium is pointing towards a GDP growth rate of 2.4% during the 2004-08 period, which is slightly higher than potential (2.1%).This favourable development is due to both net exports and domestic demand. Private consumption should become more dynamic during the 2004-2008 period, particularly thanks to the increase in households’ disposable income (especially due to tax reform). Investment growth should attain 3% during the 2004-08 period, mainly reflecting the increase in business investment. Average export growth should be 5.3% during the same period and the contribution of net exports to GDP growth should be 0.3%. Thanks to limited wage and import cost increases and a negative output gap in the first years of the projection, the inflation rate will remain below 2% in the medium term.

The development of employment should reflect the favourable macroeconomic context, the limited increases in wage costs and various policy measures. After stagnating in 2003, about 32,000 jobs should be created every year during the 2004-2008 period (as compared with 43,000 jobs created on average during 1996-2002). Industrial employment should fall by 38,000 persons during the 2003-2008 period and the number of jobs created in market services should exceed 200,000. The unemployment rate (including long term unemployment of older workers) is still increasing in 2003 (from 13.3% to 14.0%), but will subsequently fall to 12.9% in 2008. The proportion of active job seekers within broad unemployment will increase, due to recent policy measures aimed at limiting early retirement.

The public accounts are expected to show a clear deterioration, with a net public administrations borrowing requirement appearing in 2003. Equilibrium is not expected to be reached until the end of the period covered by the forecast.

Although widely used in the policy arena, especially in EU macroeconomic surveillance procedures, potential output is, by definition, unobservable. Many different approaches and techniques have therefore been proposed, and these provide estimates which may vary significantly from each other. Here we comment briefly on the advantages and disadvantages of these various approaches and present the reference method for the estimation of output gaps adopted last year by the ECOFIN Council. We apply this methodology to the economic forecasts 2003-2008 released by the FPB.

The EU reference method for computing potential growth

The estimate of potential output is used as a synthetic indicator of the aggregate supply capacity of an economy. It therefore make is possible to assess the scope for non-inflationary growth. Given the importance of this concept and the fact that potential output is unobservable in practice, the computation of potential growth has led to intensive and contentious debates in academic and policy-making circles.

These debates led to a recent change in the method employed by the European Commission to estimate potential growth in the annual assessment of the stability and convergence programmes of the member states. In the past, a purely statistical approach based on the Hodrick- Prescott (HP) filter has been used to evaluate the cyclical position of the economy. Last year a new method based on a production function was developed by the services of the European Commission.

Unlike statistical methods, an economic approach makes it possible to identify how the various factor inputs (capital stock, labour supply etc.) and technical progress contribute towards potential growth. This breakdown in determinants highlights the constraints that weigh on the economic system and the role of economic policy and structural reforms in strengthening growth. Another advantage of this kind of approach is that it establishes a clear link between the concept of potential growth and the NAIRU, i.e. the unemployment rate compatible with steady inflation. Indeed, the computation of potential growth is based on an estimation of the non-inflationary utilisation rate of input factors, particularly for labour. Finally, the production function method offers the possibility of building different growth scenarios for the future based on hypotheses for the evolution of various demographic, institutional and technological factors.

Although the use of a production function as a tool to estimate potential growth meets most of the objections that are levelled against statistical methods, it still has one major drawback: the results obtained are not independent of the modelling and estimation strategy that is used. Furthermore, the results are sensitive to the assumptions made in constructing the data, for instance to build series for capital stock or total factor productivity.

Considering the fact that the primary use of the methodology is an operational one, i.e. as a surveillance tool in the assessment of stability programmes, the approach adopted by the European Commission had to respect some basic principles in order to guarantee full transparency and equal treatment of all member states. The methodology therefore assumes a fairly simple production function: a Cobb-Douglas technology with two factor inputs (labour and capital) and constant returns to scale. In the case of capital, an exhaustive concept is used which includes private and public spending. Labour is measured by the number of employees. Estimation of the NAIRU is based on the Kalman filter, a method that relies on both structural econometric foundations and statistical techniques to extract unobserved components. More precisely, the economic information con-tained in a Phillips curve linking changes in wage inflation to cyclical unemployment is used to identify the cyclical component. In addition, the sample mean of this ‘unemployment gap’ is limited to zero. Total factor productivity, which summarizes the overall progress in factor input efficiency, is measured as the HP filtered Solow residuals 4 . Indeed, as the Commission uses no correction for the degree of utilisation of factor inputs (labour, for instance, is measured as a head count and not in terms of hours worked), the Solow residuals are filtered to remove cyclical movements generated by the evolution in utilization rates. [More in the publication ...]

STU 2-03 was finalised on May 28th 2003.