A. Introduction and Summary

1. The improved performance of the Belgian economy since 1996 has raised the possibility that sustainable growth may have risen. The assessment of potential growth is of considerable relevance for current economic policy in Belgium. The government’s budgetary plans for the next five years are predicated on a trend growth rate of 2.5 percent, compared to an average over the past two decades of 2.2 percent. Higher trend growth is also the starting point of current projections of the cost of population aging, and thus of planned strategies to prepare for the demographic shock. Finally, assessment of demand pressures is important for judging inflation risks and the stance of fiscal policy.

2. This chapter gauges the cyclical position and potential growth rate of the Belgian economy. Most of the analysis focuses on separating cyclical developments from structural trends. As estimates of the cyclical position are crude, a number of methods are used, in order to provide an appreciation of the margin of uncertainty inherent in such estimates. A final section discusses the potential growth rate for the next five years.

3. Potential output is also measured for the three regions—Brussels, Flanders and Wallonia—to the extent possible given the availability of data. Cyclical developments are diverging among the different parts of the country and, as a result, aggregation may hide notable developments at the regional level. In particular, demand pressures in one region may not show up in nation-wide measures, but, given largely centralized wage-setting, repercussions on wages could spill over to other regions. Another reason to review regional cyclical developments is to assess the need to tailor policies to regional needs.

4. The main findings of this chapter are that the Belgian economy is currently operating at a level close to potential and that annual potential growth appears to be between 2¼ percent and 2½ percent. The NAIRU has fallen substantially in Flanders, whereas structural unemployment has remained very high in Wallonia and Brussels. Both Flanders and Wallonia were close to potential at end-2000, in contrast to positions of substantial excess demand at the peak of the previous business cycle. While trend growth has risen slightly above 2½ percent in 2000, this is unlikely to be sustainable over the medium-term, given adverse demographic shifts and in the absence of evidence of a productivity growth increase due to ‘new economy’ effects.

B. Estimates of Capacity and Demand Pressure

5. This section provides a number of complementary estimates of potential output and the output gap, variables that cannot be observed directly.² Four approaches are used: trend-based estimation, a production function approach, estimates derived from indicators of spare capacity, and a joint estimation technique for potential output and the NAIRU.³ The NAIRU is a valuable concept in the analysis of cyclical pressure, as equilibrium unemployment links potential output with the available labor input. For example, in the production function approach, potential output is modeled as a function of the NAIRU, among other variables. An overview of the results is presented in Table 1, which shows trend growth, the output gap and the unemployment gap at end-2000.

Table 1.

Belgium: Estimates of the Output and Unemployment Gap in 1990 and 2000

Source: Fund staff calculations.

¹Data for the second quarter of 2000.

Table 1.

Belgium: Estimates of the Output and Unemployment Gap in 1990 and 2000

			Trend Growth	Output gap		Unemployment gap
Estimation procedure:			2000	1990	2000	1991	2000
Hodrick Prescott filter
	100 (national labor data)		2.5	2.5	0.5	-2.4	-0.4
	100 (harmonized data)					-2.7	0.1
	200 (national labor data)		2.4	2.5	0.6	-2.6	-0.4
Elmeskov						-1.8	-0.5
OECD			2.6	3.4	0.3	-2.0	0.0
Production function, with unemployment based on:
	Hodrick Prescott filter (100)		2.6	2.5	0.8
		(and Central bank data for the capital stock)	2.6	2.6	0.7
	Hodrick Prescott filter (200)		2.6	2.6	0.8
	Elmeskov		2.8	2.0	0.9
	OECD NAWRU		2.7	2.3	0.7
Spare Capacity indicators1
	Unemployment gap		2.3	2.3	1.1
	Capacity utilization		2.6	0.3	0.4
Joint estimate1			2.5	1.7	-0.7	-2.6	1.0

Source: Fund staff calculations.

¹Data for the second quarter of 2000.

View Table

Table 1.

Belgium: Estimates of the Output and Unemployment Gap in 1990 and 2000

			Trend Growth	Output gap		Unemployment gap
Estimation procedure:			2000	1990	2000	1991	2000
Hodrick Prescott filter
	100 (national labor data)		2.5	2.5	0.5	-2.4	-0.4
	100 (harmonized data)					-2.7	0.1
	200 (national labor data)		2.4	2.5	0.6	-2.6	-0.4
Elmeskov						-1.8	-0.5
OECD			2.6	3.4	0.3	-2.0	0.0
Production function, with unemployment based on:
	Hodrick Prescott filter (100)		2.6	2.5	0.8
		(and Central bank data for the capital stock)	2.6	2.6	0.7
	Hodrick Prescott filter (200)		2.6	2.6	0.8
	Elmeskov		2.8	2.0	0.9
	OECD NAWRU		2.7	2.3	0.7
Spare Capacity indicators1
	Unemployment gap		2.3	2.3	1.1
	Capacity utilization		2.6	0.3	0.4
Joint estimate1			2.5	1.7	-0.7	-2.6	1.0

Source: Fund staff calculations.

¹Data for the second quarter of 2000.

6. Openness and wage interventions complicate assessment of the NAIRU, although they do not fundamentally undermine the concepts of potential growth and the output gap. For a small open economy within EMU, with tradables prices determined externally, it may be more appropriate to focus on the link between unemployment and nontradables prices or wages (resulting in a non-accelerating wage rate of unemployment, or NAWRU, estimate). Moreover, in Belgium the translation of cyclical pressures into wages has been affected much by government intervention aimed at wage moderation and the restoration of competitiveness.⁴ Estimation is also hampered by hysteresis in unemployment developments. In light of these problems economists have tried to estimate equilibrium or natural unemployment for many European countries without making use of the price relation.⁵

Time-series and production-function estimates

7. Time-series and production-function estimates of potential output are closely related. The easiest approach to estimating the NAIRU and potential output is to smooth the actual unemployment and real output series. The production function links potential output to the trend evolution of the inputs, labor and capital, and assumed technological progress. While adding to the understanding of the factors underlying potential output, this basically shifts the issue of smoothing from overall output to the individual arguments of the production function.

8. Time-series estimates of potential output and the NAIRU were derived using the Hodrick-Prescott filter,⁶ A Cobb-Douglas production function was used to decompose output, with labor input modeled as the product of working-age population, the participation rate, one minus the unemployment rate, and average hours worked per person. To determine potential output, the unemployment rate was replaced by the NAIRU, and the other arguments—including total factor productivity, but excluding the working-age population and the capital stock—by their trend levels. In addition to the HP filter and the OECD NAWRU, Elmeskov’s (1993) approach was also applied to determine trend unemployment. For this method, wage inflation is assumed to be proportional to the difference between actual unemployment and the NAIRU (or, more accurately, the NAWRU), while also assuming that the NAWRU does not change significantly from year to year.

9. From the trend and production-function based estimates it appears that both cyclical and structural changes underpinned the improved growth performance of 1996-2000. From a cyclical perspective, the higher growth entailed a recovery from the large unemployment and output gaps that had opened up during the 1991-93 downturn (Table 1, and Figures 1 and 2). Both gaps were closed in 1999 or 2000. This result agrees with the OECD’s estimates, which are also reported. The estimates also indicate that potential growth edged up to between 2.4 percent and 2.8 percent by 2000, while the NAIRU declined from 1214 percent in 1995 to about 11 percent.⁷

Belgium: Unemployment and the Unemployment Gap, 1970-2000

(In percent)

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source: Data provided by the authorities, OECD Analytical Database, and Fund staff calculations.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Belgium: Unemployment and the Unemployment Gap, 1970-2000

(In percent)

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source: Data provided by the authorities, OECD Analytical Database, and Fund staff calculations.

• Download Figure
• Download figure as PowerPoint slide

Belgium: Unemployment and the Unemployment Gap, 1970-2000

(In percent)

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source: Data provided by the authorities, OECD Analytical Database, and Fund staff calculations.

• Download Figure
• Download figure as PowerPoint slide

Belgium: GDP Growth and Output Gap—based on Trends and a Production Function, 1970-2000

(In percent)

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source; Data provided by the authorities, OECD Analytical Database, and Fund staff calculations.1/Cobb-Douglas production function, with NAIRU based on Elmeskov approach.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Belgium: GDP Growth and Output Gap—based on Trends and a Production Function, 1970-2000

(In percent)

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source; Data provided by the authorities, OECD Analytical Database, and Fund staff calculations.1/Cobb-Douglas production function, with NAIRU based on Elmeskov approach.

• Download Figure
• Download figure as PowerPoint slide

Belgium: GDP Growth and Output Gap—based on Trends and a Production Function, 1970-2000

(In percent)

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source; Data provided by the authorities, OECD Analytical Database, and Fund staff calculations.1/Cobb-Douglas production function, with NAIRU based on Elmeskov approach.

• Download Figure
• Download figure as PowerPoint slide

10. The main source of the recent pick up of actual and potential growth (so estimated) has been the decline in unemployment. A decomposition of actual and potential growth is presented in the Tables 2 and 3, which also indicates that smaller contributions resulted from higher total factor productivity growth and a slower decline of average working hours per person. The most important structural trends over the past three decades are the gradual end of the growth of the working-age population, which has been broadly offset by an increasing participation rate and lower unemployment. Overall trends in participation reflect the balance of rising female participation, especially in part-time jobs, and a declining male participation rate. The latter development has reflected the introduction of extensive early retirement schemes and a special unemployment regime for older workers that does not require job search. There has also been a striking decline of participation rates of those below 25 years of age, with youth leaving school later than in other EU countries. Increases in part-time employment and a gradual reduction in the workweek have had a consistently negative, though diminishing, effect on labor supply.

Table 2.

Belgium: Contribution to GDP Growth

(In percent)

Source: Fund staff estimates.

Table 2.

Belgium: Contribution to GDP Growth

(In percent)

		Decomposition of growth
Year	Actual	POP	LF/POP	UN	HRS/PERS	CAP	TFP
1995	2.6	0.0	0.5	0.0	-0.1	0.8	1.3
1996	1.2	0.0	0.1	0.2	0.0	0.9	0.0
1997	3.4	0.0	0.2	0.3	-0.2	0.9	2.2
1998	2.4	0.0	0.3	0.5	-0.2	1.0	0.9
1999	2.7	0.0	0.2	0.6	-0.1	1.0	1.0
2000	3.8	0.1	0.4	0.6	0.0	1.1	1.7
1970-80	3.6	0.4	0.2	-0.4	-0.9	1.3	3.1
1981-90	2.0	0.2	0.0	-0.1	-0.2	1.0	1.2
1991-95	1.5	0.1	0.5	-0.6	-0.2	1.0	0.8
1996-2000	2.7	0.0	0.2	0.4	-0.1	1.0	1.1

Source: Fund staff estimates.

View Table

Table 2.

Belgium: Contribution to GDP Growth

(In percent)

		Decomposition of growth
Year	Actual	POP	LF/POP	UN	HRS/PERS	CAP	TFP
1995	2.6	0.0	0.5	0.0	-0.1	0.8	1.3
1996	1.2	0.0	0.1	0.2	0.0	0.9	0.0
1997	3.4	0.0	0.2	0.3	-0.2	0.9	2.2
1998	2.4	0.0	0.3	0.5	-0.2	1.0	0.9
1999	2.7	0.0	0.2	0.6	-0.1	1.0	1.0
2000	3.8	0.1	0.4	0.6	0.0	1.1	1.7
1970-80	3.6	0.4	0.2	-0.4	-0.9	1.3	3.1
1981-90	2.0	0.2	0.0	-0.1	-0.2	1.0	1.2
1991-95	1.5	0.1	0.5	-0.6	-0.2	1.0	0.8
1996-2000	2.7	0.0	0.2	0.4	-0.1	1.0	1.1

Source: Fund staff estimates.

Table 3.

Belgium: Contribution to Potential Growth

(In percent)

Source: Fund staff calculations.

Table 3.

Belgium: Contribution to Potential Growth

(In percent)

	GDP growth			Decomposition of potential growth
Year	Actual	Smoothed	Potential	POP	LF/POP	UN	HRS/PERS	CAP	TFP
1995	2.6	2.2	2.0	0.0	0.3	0.0	-0.1	0.8	1.0
1996	1.2	2.3	2.1	0.0	0.3	0.1	-0.1	0.9	1.0
1997	3.4	2.4	2.3	0.0	0.3	0.1	-0.1	0.9	1.1
1998	2.4	2.5	2.4	0.0	0.3	0.2	-0.1	1.0	1.1
1999	2.7	2.5	2.5	0.0	0.3	0.2	-0.1	1.0	1.1
2000	3.8	2.5	2.6	0.1	0.2	0.3	-0.1	1.1	1.1
1970-80	3.6	3.2	3.1	0.4	0.1	-0.4	-0.8	1.3	2.6
1981-90	2.0	2.1	2.1	0.2	0.0	-0.2	-0.3	1.0	1.3
1991-95	1.5	2.2	2.1	0.1	0.3	0.0	-0.1	1.0	1.0
1996-2000	2.7	2.4	2.4	0.0	0.3	0.2	-0.1	1.0	1.1

Source: Fund staff calculations.

View Table

Table 3.

Belgium: Contribution to Potential Growth

(In percent)

	GDP growth			Decomposition of potential growth
Year	Actual	Smoothed	Potential	POP	LF/POP	UN	HRS/PERS	CAP	TFP
1995	2.6	2.2	2.0	0.0	0.3	0.0	-0.1	0.8	1.0
1996	1.2	2.3	2.1	0.0	0.3	0.1	-0.1	0.9	1.0
1997	3.4	2.4	2.3	0.0	0.3	0.1	-0.1	0.9	1.1
1998	2.4	2.5	2.4	0.0	0.3	0.2	-0.1	1.0	1.1
1999	2.7	2.5	2.5	0.0	0.3	0.2	-0.1	1.0	1.1
2000	3.8	2.5	2.6	0.1	0.2	0.3	-0.1	1.1	1.1
1970-80	3.6	3.2	3.1	0.4	0.1	-0.4	-0.8	1.3	2.6
1981-90	2.0	2.1	2.1	0.2	0.0	-0.2	-0.3	1.0	1.3
1991-95	1.5	2.2	2.1	0.1	0.3	0.0	-0.1	1.0	1.0
1996-2000	2.7	2.4	2.4	0.0	0.3	0.2	-0.1	1.0	1.1

Source: Fund staff calculations.

Spare capacity indicators

11. To derive potential output from indicators of spare capacity, following Bayoumi (2000), output, Y_t, is modeled as a function of slack, X_t, and a polynomial time trend:

$\begin{array}{cc}\log (Y_t)=\alpha +\beta X_t+{\gamma }_{1}t+{\gamma }_2{t}^{(1/2)}+{\gamma }_3{t}^{(1/3)}+{ϵ}_t& \left(1\right)\end{array}$

Capacity utilization and the unemployment gap—with the NAIRU computed using an HP-filter—are used as indicators of slack.⁸ Alternative power sequences of time trends were tried, but did not improve the fit. On the assumption that the error term, ϵ_t, reflects changes in potential output (Y^pot) rather than cyclical disturbances, it follows that

$\begin{array}{cc}\log (Y_t^{\mathrm{pot}})=\log (Y_t)-\beta X_t& \left(2\right)\end{array}$

The annualized rate of potential growth was calculated as a moving average over the previous four years, in view of the volatility of the derived estimate of potential output.

12. The trend growth and output gap estimates based on the unemployment gap are broadly similar to those derived using a production function, although the potential growth series is more volatile. The outcome of the exercise is illustrated in Figure 3, and the full regression results are shown in Appendix B. By contrast, output gaps derived using capacity utilization are much smaller than those from the production function. This reflects the fact that this measure of slack does not explain much of the fluctuation in actual output. Given that the approach attributes the remaining part of output variation (the estimated error term) to changes in potential output, this part of output variation does not show up in the output gap.

Belgium: GDP Growth and the Output Gap—Estimates Based on Spare Capacity

(In percent)

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source: Fund staff calculations

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Belgium: GDP Growth and the Output Gap—Estimates Based on Spare Capacity

(In percent)

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source: Fund staff calculations

• Download Figure
• Download figure as PowerPoint slide

Belgium: GDP Growth and the Output Gap—Estimates Based on Spare Capacity

(In percent)

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source: Fund staff calculations

• Download Figure
• Download figure as PowerPoint slide

Joint estimates of potential output and the NAIRU

13. A drawback of the above calculations is that the mutual dependence of the output and unemployment gaps is not taken into account. Both the production function and the spare-capacity approach are based on an exogenously determined measure of the unemployment gap. A useful alternative that provides a joint estimate of potential output and the NAIRU, based on Okun’s Law and a Phillips Curve relationship, is the unobserved components model of Apel and Jansson (1997) (Box 1).

14. Joint estimates of potential output and the NAIRU, presented in Figure 4, suggest that at mid-2000 the cyclical improvement had not yet resulted in the complete elimination of the output and unemployment gaps. Notably, the Apel-Jansson model indicates an earlier decline of the NAIRU than the trend-based or Elmeskov methods. At a technical level, this reflects the use of the Phillips relation: there has been little inflation during the current upswing, implying that unemployment had not fallen below the NAIRU. The interpretation is that either the NAIRU is lower than its trend suggests, or that inflationary pressures have been tempered by an exogenous factor other than oil or import prices (which were controlled for in the estimation).⁹ Still, like the other estimates, the output gap for 2000 is not significantly different from zero.

Belgium: Output and Unemployment Gap—based on joint estimation, 1980-2000

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source: Fund staff calculations.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Belgium: Output and Unemployment Gap—based on joint estimation, 1980-2000

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source: Fund staff calculations.

• Download Figure
• Download figure as PowerPoint slide

Belgium: Output and Unemployment Gap—based on joint estimation, 1980-2000

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source: Fund staff calculations.

• Download Figure
• Download figure as PowerPoint slide

15. A comparison with developments during the previous cyclical upturn, at the end of the 1980s, is illuminating. After the unemployment rate started to decline in the course of 1987, inflation picked up in the second quarter of 1988. Accordingly, by mid-1988 the estimated unemployment gap was closed, with a NAIRU of 8.9 percent (harmonized unemployment). Wages remained stable until the 1990-91 wage settlement, which featured annual increases by more than 7 percent. Turning to recent years, while the unemployment rate has declined steadily since 1995, there was no significant pick up of inflation until the beginning of 2000—a price hike fully explained by sharply rising import prices.

Box 1.

The Joint Estimation of Potential Output and the NAIRU

Apel and Jansson (1999) propose a procedure for joint estimation of the unobserved potential output and NAIRU variables, based on the Phillips curve for identifying capacity limits. Formally, the model reads (with the equations presented directly in their empirical format):

$\begin{array}{cc}\Delta {\pi }_t=\sum _{i=1}^I{\rho }_i\Delta {\pi }_{t-i}+\sum _{j=0}^J(u_{t-j}-{ū}_{t-j})+\sum _{k=0}^K{\omega }_k{z}_{t-k}+{\epsilon }_{1t}& \left(1\right)\end{array}$

$\begin{array}{cc}{y}_t-y_t^p=\sum _{l=0}^L{\phi }_l(u_{t-l}-{ū}_{t-l})+{\epsilon }_{2t}& \left(2\right)\end{array}$

$\begin{array}{cc}{ū}_t={ū}_{t-l}+{\epsilon }_{3t}& \left(3\right)\end{array}$

$\begin{array}{cc}{y}_t^p=\alpha +y_{t-l}^p+{\epsilon }_{4t}& \left(4\right)\end{array}$

$u_{t-{\overline{u}}_t}=\underset{m-1}{\overset{M}{\mathrm{\Sigma }}}{\delta }_m\left(u_{t-m-{\overline{u}}_{t-m}}\right)+{ϵ}_{5t}\left(5\right)$

where π_t is the log difference of the CPI, u_t the unemployment rate, ū_t the NAIRU, z_t exogenous (supply-shock) variables, y_t the log of real output, and y_t^p the log of potential output. The error terms ε_1t, ε_2t, ε_3t, ε_4t, and ε_5t are assumed to be IID distributed.

The model incorporates the Phillips curve (equation (1)) and Okun’s Law (equation (2)), and assumes that the NAIRU and potential output are characterized by stochastic trends (equations (3) and (4)). Equation (5) specifies the evolution of cyclical employment to close the model. Following Apel and Jansson, z_t includes relative import prices and the relative oil price (both deflated by the CPI), the exchange rate, and productivity. The model is estimated using the Kaiman filter.

Regional estimates

16. Diverging trends among the regions within Belgium reflect the very low degree of interregional labor mobility and the distinct sectoral characteristics and economic history of the different parts of the country.¹⁰ Wallonia still has not recovered from the downturn of its large iron, steel, and metal products industries. These declining industries featured less prominently in Flanders, which, moreover, harbors a flourishing service sector. Limited data availability restricts the scope for replicating at the regional level the exercises presented for the national level. In particular, there are no regional data on the capital stock, and output data are available only at annual frequency. In the absence of regional capital stock data, for the production-function estimates the contributions to production growth of capital and total factor productivity have been combined, and derived as a residual. In determining potential growth, this combined term was replaced by its trend using an HP filter.

17. The trend unemployment rate, measured with a Hodrick-Prescott filter on the actual rate, has fallen sharply in Flanders, whereas in Wallonia trend unemployment has stabilized at a very high level (Figure 5). This divergence is mirrored by diverging patterns of potential growth among the regions (Figure 6). In Flanders, potential growth rose to 2.6 percent during 1996-2000, whereas in Wallonia it remained at about 2 percent. The causes of this disparity are presented in terms of the arguments of the production function in Table 4. In addition to different unemployment performance, the combined growth contribution of capital and productivity has also added to the divergence in potential growth rates. In other words, trend labor productivity increased faster in Flanders than in Wallonia.

Belgium: Measures of Regional Unemployment, 1980-2000 ^1/

(In percent)

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source: Data provided by the authorities, Eurostat, and Fund staff calculations.^1/ With the NAIRU based on an HP filter.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Belgium: Measures of Regional Unemployment, 1980-2000 ^1/

(In percent)

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source: Data provided by the authorities, Eurostat, and Fund staff calculations.^1/ With the NAIRU based on an HP filter.

• Download Figure
• Download figure as PowerPoint slide

Belgium: Measures of Regional Unemployment, 1980-2000 ^1/

(In percent)

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source: Data provided by the authorities, Eurostat, and Fund staff calculations.^1/ With the NAIRU based on an HP filter.

• Download Figure
• Download figure as PowerPoint slide

Belgium: Potential Growth and the Output Gap by Region, 1980-2000 ^1/

(In percent)

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source: Data provided by the authorities, Eurostat, and Fund staff calculations.^1/ With potential output based on regional labor input (with the NAIRU derived using an HP filter), and trend labor productivity.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Belgium: Potential Growth and the Output Gap by Region, 1980-2000 ^1/

(In percent)

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source: Data provided by the authorities, Eurostat, and Fund staff calculations.^1/ With potential output based on regional labor input (with the NAIRU derived using an HP filter), and trend labor productivity.

• Download Figure
• Download figure as PowerPoint slide

Belgium: Potential Growth and the Output Gap by Region, 1980-2000 ^1/

(In percent)

Citation: IMF Staff Country Reports 2001, 045; 10.5089/9781451803167.002.A001

Source: Data provided by the authorities, Eurostat, and Fund staff calculations.^1/ With potential output based on regional labor input (with the NAIRU derived using an HP filter), and trend labor productivity.

• Download Figure
• Download figure as PowerPoint slide

Table 4.

Belgium: Contribution to Regional Potential Growth

(In percent)

Source: Fund staff calculations.

Table 4.

Belgium: Contribution to Regional Potential Growth

(In percent)

		GDP Growth			Decomposition of Potential Growth
		Actual	Smoothed	Potential	POP	LF/POP	UN	HRS/PERS	CAP and TFP
Belgium
	1990-1995	1.6	2.1	2.0	0.1	0.3	0.0	-0.2	1.8
	1996-2000	2.6	2.4	2.3	0.0	0.3	0.2	-0.1	2.0
Flanders
	1990-1995	2.1	2.5	2.4	0.1	0.3	0.1	-0.2	2.1
	1996-2000	2.8	2.6	2.6	0.0	0.2	0.3	-0.1	2.2
Wallonia
	1990-1995	1.4	1.7	1.6	0.0	0.4	-0.3	-0.2	1.6
	1996-2000	2.3	2.1	1.9	0.0	0.4	0.0	-0.1	1.7
Brussels
	1990-1995	0.4	1.1	0.9	-0.3	0.3	-0.2	-0.2	1.2
	1996-2000	2.4	1.9	1.9	0.2	0.2	0.0	-0.1	1.7

Source: Fund staff calculations.

View Table

Table 4.

Belgium: Contribution to Regional Potential Growth

(In percent)

		GDP Growth			Decomposition of Potential Growth
		Actual	Smoothed	Potential	POP	LF/POP	UN	HRS/PERS	CAP and TFP
Belgium
	1990-1995	1.6	2.1	2.0	0.1	0.3	0.0	-0.2	1.8
	1996-2000	2.6	2.4	2.3	0.0	0.3	0.2	-0.1	2.0
Flanders
	1990-1995	2.1	2.5	2.4	0.1	0.3	0.1	-0.2	2.1
	1996-2000	2.8	2.6	2.6	0.0	0.2	0.3	-0.1	2.2
Wallonia
	1990-1995	1.4	1.7	1.6	0.0	0.4	-0.3	-0.2	1.6
	1996-2000	2.3	2.1	1.9	0.0	0.4	0.0	-0.1	1.7
Brussels
	1990-1995	0.4	1.1	0.9	-0.3	0.3	-0.2	-0.2	1.2
	1996-2000	2.4	1.9	1.9	0.2	0.2	0.0	-0.1	1.7

Source: Fund staff calculations.

18. The estimated unemployment and output gaps do not indicate a serious overheating threat; even for Flanders the gap estimates for mid-2000 are not significantly different from zero. A similar pattern was found for Wallonia, reflecting a small drop in unemployment relative to a stable trend. The elimination of the labor-market gap in Flanders is consistent with recent reports of increasing labor shortages there, which are corroborated by a sharp increase in vacancies.

C. Assessing the Scope for Future Growth

19. A tentative assessment suggests that potential growth during the 2001-05 period is in the 2¼ to 2½ percent range. Potential growth rose to 2.4 percent on average during 1995-2000, and reached 2.6 percent in 2000. The evolution of the components of the production function provide a useful starting point for evaluating the scope for future economic growth. Unless a fairly optimistic assessment is made of the likely evolution of TFP, capital, and labor input, there will be a slowdown in potential growth. In short, the future course of the demographics and, to a lesser extent, the likely changes in the NAIRU, workweek, and capital accumulation, indicate that for the medium-term, 2½ percent trend growth should be seen as an upper estimate. Such evaluations are, of course, subject to important caveats, for example regarding the effectiveness of labor market policies and ongoing social changes, such as increasing female labor participation.

20. Starting with the growth contribution of TFP, it is hard to predict whether the small increase in productivity growth recorded in recent years indicates the advent of a “new economy.” While trend TFP growth accelerated slightly in recent years, this has been a common feature of cyclical upswings.¹¹ In addition, there appears to be little evidence that a “new economy” is boosting productivity growth in Western Europe more generally.¹²

21. The growth contribution of capital accumulation is likely to return to its trend. This contribution has steadily increased since 1994, as the investment ratio has risen to about 16 percent—its peak level during the previous boom, recorded in 1990.¹³ High capacity utilization and profitability and low long-term real interest rates are projected to result in a further increase in the investment ratio in 2001. Beyond that year, however, a return to historical levels could be expected, which would imply an average growth contribution during 2000-05 of about 1.0 percentage point—in line with the 1995-2000.

22. Some slowdown in the growth of effective labor input, the third argument of the production function, also appears likely. The evolution of labor input can be analyzed in four steps.

• Pure demographic changes are a relatively predictable factor. There are two major factors to consider: the total working-age population, and its changing composition by age group. The net effect is to reduce annual potential output growth by 0.1 percentage points (Table 5).¹⁴ Total working-age population is expected to increase slightly over the 2001-05 period, by 0.2 percent a year on average, following five years of virtually no growth. For a given participation rate, this would boost annual GDP growth by 0.1 percentage point. However, this will be more than offset by the effects of aging within the working-age population. The share of the total working-age population of those between 20-55 years old is expected to decline from 75.4 percent in 2000 to 73.8 percent by 2005. Given the very low employment rates of those below 20 years or above 55 years of age, this would lower the participation rate by 1 percentage point from 2000 to 2005 and reduce GDP growth—relative to the previous five-year period—by 0.2 percentage points a year on average.

• Although participation rates per age cohort may well continue increasing, this is unlikely to compensate fully for the negative net demographic shock. Over the previous five-year period, higher participation within each cohort (defined by age-group and sex) added 1 percentage point to the participation rate and 0.2 percentage point to average annual output growth. It is hard to project to what extent this contribution will be repeated in the medium-term future. On the one hand, in the recent past, participation may have benefited from an encouraged worker effect. On the other hand, several new policies aim to encourage participation, including a plan for starter jobs (introduced in 2000), restrictions on the special unemployment regime for older workers (to be introduced in 2001), and a workers tax credit to reduce unemployment trap problems (starting in 2002). Of particular importance is the effect of the increase in the pensionable age for women. The Federal Planning Bureau (2000) has projected that between 1999 and 2005 this measure alone will account for almost 0.1 percentage point annual increase in potential labor supply.¹⁵ Overall, and in spite of the negative demographic effect, the Bureau projected an average increase in the (actual) labor force by almost 0.2 percentage points a year during this period, only slightly less than during 1995-99.

• It is unlikely that the NAIRU will decline more during the next five years than it has during the 1996-2000. The improvement recorded during the latter period amounted to between 0.5 (OECD) and 1.2 (HP filter) percentage points. This resulted almost exclusively from a 2 percentage point (HP filter) fall in the NAIRU in Flanders to about 6¾ percent (5½ percent on a harmonized basis). The latter figure is not far from unemployment rates of booming countries like the Netherlands (with a 4½ percent harmonized NAWRU according to the OECD). Although there is ample scope for the NAIRU in Wallonia and Brussels to go down, such trend is not evident yet.

• Finally, a gradual shortening of working hours is likely to continue to dampen effective labor supply. The 2001-2002 interprofessional agreement proposed a reduction of the maximum workweek from 39 to 38 hours by 2002 and accepted government proposals for more extensive individual leave schemes. These initiatives suggest a further reduction in average working hours. The labor supply effect of new leave schemes for older workers is more ambiguous, as workers could be motivated to remain in the workforce longer.

Table 5.

Belgium: Demographics, Participation, and Growth ¹

(In percent)

Source: Data provided by the authorities and Fund staff calculations. Labor force data include unemployment as defined by the Ministry of Employment, exlaining a small difference with participation effects mentioned in Table 1, based on the national definition.

^1/GDP growth effects are derived using a Cobb-Douglas production function with a labor input coefficient of 0.67.

^2/The ratio of the 20-55 age group over the 15-65 group serves as a proxy for relevant population aging.

^3/The data reflect changes in the distribution of the working-age population over cohorts (defined by age group and sex).

As older cohorts have lower participation rates, population aging reduces overall participation.

Table 5.

Belgium: Demographics, Participation, and Growth ¹

(In percent)

	Working-age population		Population aging		Participation rate by cohort	Overall participation rate		Participation level
	Growth	GDP growth Effect	Ratio 20-55 to total2/	GDP growth Effect3/	GDP growth effect	Level	GDP growth effect	GDP growth effect
		(1)		(2)	(3)		(4)=(2)+(3)	(5)=((1)+(4)
1995	0.0	0.0	74.4	0.1	0.3	64.4	0.4	0.4
1996	0.0	0.0	74.7	0.1	0.1	64.6	0.2	0.2
1997	0.0	0.0	75.1	0.1	0.2	64.9	0.3	0.3
1998	0.0	0.0	75.3	0.0	0.1	65.0	0.1	0.1
1999	0.1	0.0	75.3	0.0	0.2	65.0	0.2	0.2
2000	0.1	0.1	75.4	0.0	0.4	65.0	0.4	0.4
2001	0.4	0.2	75.3	-0.1		64.9	-0.1	0.2
2002	0.2	0.2	75.0	-0.2		64.7	-0.2	0.0
2003	0.1	0.1	74.6	-0.2		64.5	-0.2	-0.1
2004	0.1	0.1	74.2	-0.3		64.3	-0.3	-0.2
2005	0.2	0.2	73.8	-0.3		64.0	-0.3	-0.2
Averages
1991-1995	0.1	0.1	74.0	0.2	0.2	63.8	0.4	0.4
1996-2000	0.0	0.0	75.1	0.0	0.2	64.9	0.2	0.3
2001-2005	0.2	0.1	74.6	-0.2		64.5	-0.2	-0.1

Source: Data provided by the authorities and Fund staff calculations. Labor force data include unemployment as defined by the Ministry of Employment, exlaining a small difference with participation effects mentioned in Table 1, based on the national definition.

^1/GDP growth effects are derived using a Cobb-Douglas production function with a labor input coefficient of 0.67.

^2/The ratio of the 20-55 age group over the 15-65 group serves as a proxy for relevant population aging.

^3/The data reflect changes in the distribution of the working-age population over cohorts (defined by age group and sex).

As older cohorts have lower participation rates, population aging reduces overall participation.

View Table

Table 5.

Belgium: Demographics, Participation, and Growth ¹

(In percent)

	Working-age population		Population aging		Participation rate by cohort	Overall participation rate		Participation level
	Growth	GDP growth Effect	Ratio 20-55 to total2/	GDP growth Effect3/	GDP growth effect	Level	GDP growth effect	GDP growth effect
		(1)		(2)	(3)		(4)=(2)+(3)	(5)=((1)+(4)
1995	0.0	0.0	74.4	0.1	0.3	64.4	0.4	0.4
1996	0.0	0.0	74.7	0.1	0.1	64.6	0.2	0.2
1997	0.0	0.0	75.1	0.1	0.2	64.9	0.3	0.3
1998	0.0	0.0	75.3	0.0	0.1	65.0	0.1	0.1
1999	0.1	0.0	75.3	0.0	0.2	65.0	0.2	0.2
2000	0.1	0.1	75.4	0.0	0.4	65.0	0.4	0.4
2001	0.4	0.2	75.3	-0.1		64.9	-0.1	0.2
2002	0.2	0.2	75.0	-0.2		64.7	-0.2	0.0
2003	0.1	0.1	74.6	-0.2		64.5	-0.2	-0.1
2004	0.1	0.1	74.2	-0.3		64.3	-0.3	-0.2
2005	0.2	0.2	73.8	-0.3		64.0	-0.3	-0.2
Averages
1991-1995	0.1	0.1	74.0	0.2	0.2	63.8	0.4	0.4
1996-2000	0.0	0.0	75.1	0.0	0.2	64.9	0.2	0.3
2001-2005	0.2	0.1	74.6	-0.2		64.5	-0.2	-0.1

Source: Data provided by the authorities and Fund staff calculations. Labor force data include unemployment as defined by the Ministry of Employment, exlaining a small difference with participation effects mentioned in Table 1, based on the national definition.

^1/GDP growth effects are derived using a Cobb-Douglas production function with a labor input coefficient of 0.67.

^2/The ratio of the 20-55 age group over the 15-65 group serves as a proxy for relevant population aging.

^3/The data reflect changes in the distribution of the working-age population over cohorts (defined by age group and sex).

As older cohorts have lower participation rates, population aging reduces overall participation.

23. Beyond 2005, demographic projections suggest a further slowdown in potential growth. The working-age population is expected to contract at an accelerating pace, on average by 0.2 percent between 2006 and 2020, and by 0.6 percent between 2020 and 2030. The share of those between 20 and 55 years of age relative to the total working-age population would decrease from 74 percent in 2005 to 69½ percent by 2020, and recover slightly thereafter. Together, these developments would imply an average slowdown in economic growth by 0.5 percent during 2006-30 relative to 1991-2000. Increasing employment by age cohort could offset them, however, to this negative shock. For example, the Lisbon agreement aims to raise the employment rate to 70 percent by 2010. Applied to Belgium, this would imply an increase of 11 percentage points from the rate in 2000. Attaining this very ambitious target would contribute more than a percentage point to the potential growth rate, more than offsetting the ageing effect.