Abstract
Selected Issues
Understanding Productivity Growth in Belgium: Sectoral and Firm-Level Analysis1
A. Introduction
1. Productivity growth in Belgium has declined and fallen short of many peer countries. Average labor productivity growth declined from 2 percent in the 1990s to 1.3 percent in the 2000s before the crisis to only 0.7 in 2010–16. While many advanced economies have experienced a similar trend, the productivity slowdown has been more pronounced in Belgium. The shortfall in labor productivity growth relative to the average of Germany, France, and the Netherlands has averaged 0.4 percentage points over the last two decades. The employment growth, however, was 0.1 percentage points higher on average and may have generated some downward pressures on the level of labor productivity. The difference in productivity slowdown between Belgium and neighboring countries may also partially reflect how the crisis affected each economy and which policy responses were pursued.
Evolution of Real GVA per Hour Worked
(Index, 1996=100)
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
1/ Difference between productivity growth in Belgium and the average of Germany, France and the Netherlands.Sources: Eurostat and IMF staff calculations.Evolution of Real GVA per Hour Worked
(Index, 1996=100)
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
1/ Difference between productivity growth in Belgium and the average of Germany, France and the Netherlands.Sources: Eurostat and IMF staff calculations.Evolution of Real GVA per Hour Worked
(Index, 1996=100)
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
1/ Difference between productivity growth in Belgium and the average of Germany, France and the Netherlands.Sources: Eurostat and IMF staff calculations.2. Slow productivity growth is a particular concern in Belgium in light of population aging and the need to preserve external competitiveness. Belgium’s economic dependency ratio is projected to reach 65 percent by 2030 (from current 55 percent). This means that maintaining a positive medium-term GDP per capita growth will require improvements in labor productivity growth. In addition, productivity, together with nominal wages, determine unit labor costs—a key measure of competitiveness for small and very open economies such as Belgium. If productivity growth continues to diverge from peer countries, wage moderation alone will not be sufficient to preserve competitiveness.
Labor Force and Dependency Ratio
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: United Nations and IMF staff calculations.Labor Force and Dependency Ratio
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: United Nations and IMF staff calculations.Labor Force and Dependency Ratio
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: United Nations and IMF staff calculations.3. A variety of factors could be responsible for weak productivity growth in Belgium. The broader literature on the productivity slowdown in advanced economies identifies a number of structural reasons, several of which are relevant to Belgium, including: i) sectoral shifts, with a growing share of less productive service sectors in total employment (Dabla-Norris et al., 2015); ii) lack of public investment and deteriorating quality of infrastructure; iii) slowing productivity in some ICT related activities (Adler et al., 2017); iii) aging populations and underinvestment in human capital (Feyrer (2007), Aiyar et al. (2016)); and iv) lack of competition, particularly in service sectors, and regulatory distortions (Duval et al., 2015; Gal et Hijzen, 2016). A number of papers have emphasized the importance of these factors for Belgium, e.g. Dhyne et Fuss, 2014; Biatour et Kegels, 2017; Bourles et al., 2010; Ariu et Vandenberghe, 2014. In particular, Cette et al. (2016) have estimated a long-run increase in productivity of 6 percentage points if labor and product markets regulations in Belgium were reduced to the best practices in the OECD; and Andrews et al. (2015) have linked the weak dynamics of the firms’ population in Belgium to high entry and exit barriers and increased misallocation of resources.
4. The purpose of this paper is to provide new empirical evidence that assesses a broad range of factors that have contributed to the slowdown of productivity growth in Belgium. In contrast to the existing literature, our analysis combines regulatory parameters with other factors that have played an important role in Belgium, such as infrastructure quality and aging. We also look separately at the role of secular shifts in sectoral employment.
Sectoral shifts in employment. Continued deindustrialization of advanced economies has implied a reallocation of resources to service sectors, where productivity growth is generally slower. In Belgium, the share of manufacturing in total employment has declined and the share of services, both tradable and non-tradable, has increased considerably. We examine whether this sectoral shift was more pronounced in Belgium and resulted in a larger drag on aggregate productivity, thus explaining part of the productivity gap with the neighbor countries.
Sector-specific factors: barriers to competition, public infrastructure, and workforce aging. The analysis focuses on the following questions: i) What is the impact of barriers to competition on firms’ performance in regulated sectors? ii) What are the spillover effects on the productivity of companies in downstream sectors (e.g., sectors that rely on inputs from the regulated sectors)? iii) What is the sector-specific role of infrastructure quality? iv) What is the impact of the increasing share of older workers. The use of firm-level data allows us to control for additional firm-specific characteristics such as size and access to finance.
5. The paper is structured as follows. Section B describes sectoral developments in Belgium’s productivity and assesses the relative role of sectoral shifts in the productivity slowdown. Section C describes key explanatory variables and their expected effect. Section D outlines the empirical strategy and firm-level data. Section E presents the econometric estimation results and counterfactual simulations. Section F concludes.
B. Sectoral Productivity and Sectoral Shifts
6. Productivity developments in Belgium have been heterogeneous across economic sectors. Manufacturing, construction, and finance have been performing at or above the regional average in terms of value-added per hour worked. For instance, the average annual growth in Belgian industry was 4 percent versus an average 2.7 percent in Germany, Netherlands and France. At the same time, service subsectors with large employment shares—including trade, travel, accommodation, public administration, education, and science—have seen subdued productivity growth. Labor productivity growth in ICT services has been particularly poor compared to peer countries (almost two times slower on average) and might have negatively affected total factor productivity (TFP) in the rest of the economy (Bart van Ark, 2014) (Figure 1, charts 1–3).
Belgium: Sectoral Productivity
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: Eurostat and IMF staff calculations.Belgium: Sectoral Productivity
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: Eurostat and IMF staff calculations.Belgium: Sectoral Productivity
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: Eurostat and IMF staff calculations.7. Sectoral compositional effects have amplified the adverse productivity trend. Over 1996–2016 the employment share of industry has declined by more than a third, while the share of professional services has doubled and the share of other non-tradable services has also increased. Figure 1, chart 5 shows that sectors with stagnating productivity growth have increased their employment the most, while those with high productivity growth have seen shrinking employment shares.
8. Sectoral reallocation effects explain about half of the cumulative productivity growth gap relative to peers since 1996. To estimate the role of sectoral shifts in the slowdown of aggregate productivity growth and the widening of the gap with peer countries, we calculate a hypothetical total labor productivity growth rate that assumes unchanged sectoral shares of employment at 1996 levels combined with actual labor productivity growth by sector (at one digit NACE). Figure 1, chart 6 shows the difference between the counterfactual and actual aggregate productivity growth. The results show that if the structure of the economy in Belgium had remained the same as in 1996, aggregate annual productivity growth would have been 0.4 percentage points higher. This difference is significantly higher in Belgium than in France, Germany, and the Netherlands and accounts for half of the productivity growth gap with these countries.
9. In the next section, we explore whether sector-specific factors could be responsible for the remaining gap in productivity. The next section focuses on other factors that may have contributed to the gap, namely underinvestment in public infrastructure, an aging population, and policy distortions, including regulations that limit competition in services.
C. Potential Explanatory Variables: Regulation, Infrastructure and Aging
10. Belgium has excessive regulation of network industries and professional services. While Belgium’s overall product market regulation score is not worse than the OECD average, several service sectors, including telecommunications, retail, legal and accounting, and land transportation face comparatively high barriers to entry and competition according to the OECD PMR sectoral indicators for 1996–2013. Moreover, the new OECD dataset for 2014–16, which collects information on trade restrictions across 19 major services sectors (STRI), shows that virtually all service sectors in Belgium are more closed to foreign competition than the service sectors in France, Germany or Netherlands.
Sectoral Market Regulation, 2013
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: OECD and IMF staff calculations.Sectoral Market Regulation, 2013
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: OECD and IMF staff calculations.Sectoral Market Regulation, 2013
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: OECD and IMF staff calculations.11. Obstacles to competition in services can hurt productivity and push up prices in the regulated sectors. A number of studies have looked at the impact of product and labor market imperfections on productivity (e.g., Aghion and Howitt, 2009). In theory the impact of regulations causing or supporting these imperfections can go both ways: on the one hand, incumbent firms, protected by barriers to entry or other regulations limiting competition, have less incentive to be efficient and innovate; on the other hand, they might have more resources to invest in innovation. Most empirical studies, however, show that the first mechanism dominates, and therefore higher anti-competitive regulatory protection has a negative impact on firms’ productivity (Duval et al 2015, Gal et Hijzen, 2016). In addition, barriers to entry consistently enable incumbent firms that are shielded from competition to raise their prices, while a lack of regulatory transparency and complex administrative procedures tend to add to firms’ operating expenses. An empirical OECD study (Rouzet et Spinelli, 2016) explored the relationship between services trade policies and mark-ups at the firm level. The authors found that restrictive regulations enable firms to charge higher mark-ups in a majority of services sectors. In addition, high entry and exit barriers may result in misallocation of resources to less productive firms and reduce aggregate productivity.
12. Higher prices and lack of competition in regulated sectors can have significant negative spillovers to downstream sectors and the rest of the economy. Higher rents and market power in regulated sectors can result in higher costs of inputs in downstream industries, thus lowering productivity in the connected industries.2 A number of empirical studies find evidence that anticompetitive upstream regulations have significantly curbed productivity growth in downstream industries (Bourles et al. 2010, WEO 2016, Chapter 3, Gal et Hijzen, 2016). Biatour and Kegels (2017) show that market services in Belgium have seen higher price increases than in neighboring countries. Annual report by the Belgian Price Observatory also points to high market concentration in a number of service sectors.3
Indirect Regulation Exposure, 2013
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: OECD and IMF staff calculations.Indirect Regulation Exposure, 2013
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: OECD and IMF staff calculations.Indirect Regulation Exposure, 2013
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: OECD and IMF staff calculations.13. Our estimates show that most Belgian sectors have high intensity of indirect regulation. Following Lanau, Topalova (2016), we construct a measure of indirect exposure to regulation for each sector:
where PMR_jct is the OECD sectoral product market regulation index for country c sector j in year t; Intk is the share of intermediate inputs provided by each sector k to sector j. To calculate the share of intermediate inputs, and to avoid endogeneity issues, we use the US Input Output matrix in European 2-digit NACE sector classification.
14. Productivity in some sectors might be more affected by the declining quality of infrastructure. A long period of low public investment in Belgium has brought the net stock of public capital well below peers, which has negatively affected the quality of infrastructure.4 According to the Global Competitiveness Report (GCR), the quality of Belgian infrastructure in general, and roads especially, has been declining for years and is well below comparator countries. As some industries are more dependent than others on transportation for their inputs, the quality of infrastructure will have varying effects on sectoral productivity. To take this into account, we construct a sector-specific indicator by interacting a country-level measure of infrastructure quality with the intensity of infrastructure use by each sector. The intensity of use is measured by the share of transportation in total intermediary inputs of the sector using the US Input Output matrix at 2-digit NACE sector classification. The index shows that the transportation, trade, and electricity sectors in Belgium are most affected by infrastructure quality and have the largest gap.
Infrastructure Quality Gap by Sector
(Index)
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Source: Global Competitiveness Report and IMF staff calculations.Infrastructure Quality Gap by Sector
(Index)
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Source: Global Competitiveness Report and IMF staff calculations.Infrastructure Quality Gap by Sector
(Index)
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Source: Global Competitiveness Report and IMF staff calculations.15. Belgium has a rapidly aging workforce. The share of 55+ workers in Belgium has increased twofold from 6.5 percent in 2000 to 15 percent in 2017. While this is a positive development in general, it may have had an impact on labor productivity growth, with some studies linking workforce aging to declining labor productivity in Belgium (e.g. Ariu et Vandenberghe, 2014).
Share of 55+ Workers in Employment by Sector
(Percent)
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: Eurostat and IMF staff calculations.Share of 55+ Workers in Employment by Sector
(Percent)
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: Eurostat and IMF staff calculations.Share of 55+ Workers in Employment by Sector
(Percent)
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: Eurostat and IMF staff calculations.D. Firm-Level Data and Empirical Strategy
16. The empirical strategy follows Lanau, Topalova (2016) and expands the analysis by including public infrastructure and aging. To assess the impact on productivity of the factors discussed in the previous section we estimate several empirical specifications. We start by replicating specifications of the previous studies.
Equation 1 estimates the direct impact of regulation on the sample of companies in service and network sectors subject to regulatory barriers:
where Y_it is a performance measure of firm i in year t (these include TFP growth, labor productivity growth, value added growth, markup); PMR_jct is the OECD sectoral product market regulation index for country c sector j in year t; X_it are additional firm-level controls, including proxies for financial constraints (debt-to-asset ratio), firm size (using lagged value added) and age; α_t and α_c are time and country dummies that are used to directly control for macroeconomic fluctuations in country c, and all other factors that may affect productivity equally across firms. Negative sign of coefficient βt would indicate direct positive impact on firms’ productivity from reduced regulatory obstacles in these sectors.
Equation 2 estimates the spillover impact of regulation on a sample of companies in downstream industries:
where Intk is the intensity of sector k input use by sector j. Negative sign of coefficient βt would indicate positive spillover impact on productivity of companies in downstream sectors.
Equation 3 estimates both effects on a pooled regression of both samples and with added controls of infrastructure quality and aging:
where Infrjct -is the GCR index of infrastructure quality weighted by the industry specific intensity of infrastructure use; e55jct is an employment share of 55+ employees in total employment of sector j.
17. We use data extracted from the ORBIS database of Bureau van Dijk, compiled by the IMF’s research department (Gal and Hijzen, 2016). We use data on balance sheets, income statements, and sectoral classifications of around 650,000 active companies from 14 advanced economies over 1996–2013. For Belgium the number of companies varies from 5,000 to 8,500 depending on the year (Table 1). Table 2 shows the sample breakdown in terms of NACE 1-digit sectors for the whole sample and for Belgium. We use the OECD sectoral regulation measures (PMR) that cover electricity, gas, telecom, post, air, roads, rail, accountants, lawyers, architects, engineers, and retail. We use the Global Competitiveness Report measure for infrastructure quality and Eurostat data on employment by NACE2 2-digit industry and age. To measure the indirect exposure to regulation by different industries we use the US Input Output table in European classification.
Sample Size by Country
Sample Size by Country
| Country | Average Number of Firms Year 1996–2013 |
|---|---|
| BE | 6,727 |
| CZ | 11,779 |
| DE | 10,313 |
| ES | 205,285 |
| FI | 9,168 |
| FR | 126,245 |
| GB | 17,819 |
| IT | 125,631 |
| JP | 11,998 |
| KR | 9,532 |
| PT | 49,351 |
| SE | 60,637 |
| SI | 1,387 |
| SK | 6,559 |
| Total | 646,424 |
Sample Size by Country
| Country | Average Number of Firms Year 1996–2013 |
|---|---|
| BE | 6,727 |
| CZ | 11,779 |
| DE | 10,313 |
| ES | 205,285 |
| FI | 9,168 |
| FR | 126,245 |
| GB | 17,819 |
| IT | 125,631 |
| JP | 11,998 |
| KR | 9,532 |
| PT | 49,351 |
| SE | 60,637 |
| SI | 1,387 |
| SK | 6,559 |
| Total | 646,424 |
Belgium: Description by Sector
Belgium: Description by Sector
| Number of Observations | Average number of Firms | |||
|---|---|---|---|---|
| Total | Belgium | Total | Belgium | |
| Agriculture | 260,821 | 364 | 14,490 | 20 |
| Mining | 48,950 | 638 | 2,719 | 35 |
| Manufacturing | 2,603,180 | 31,153 | 144,621 | 1731 |
| Electricity | 34,479 | 298 | 1,916 | 17 |
| Water supply | 75,881 | 1,495 | 4,216 | 83 |
| Construction | 1,806,319 | 10,204 | 100,351 | 567 |
| Trade | 3,405,777 | 43,556 | 189,210 | 2,420 |
| Transportation | 637,931 | 9,527 | 35,441 | 529 |
| Accom. and food | 624,125 | 909 | 34,674 | 51 |
| ICT | 376,294 | 5,128 | 20,905 | 285 |
| Financial and insura | 85,096 | 1,089 | 4,728 | 61 |
| Real estate activitie | 440,860 | 2,356 | 24,492 | 131 |
| Prof. services | 908,278 | 9,365 | 50,460 | 520 |
| Administrative | 435,749 | 5,009 | 24,208 | 278 |
Belgium: Description by Sector
| Number of Observations | Average number of Firms | |||
|---|---|---|---|---|
| Total | Belgium | Total | Belgium | |
| Agriculture | 260,821 | 364 | 14,490 | 20 |
| Mining | 48,950 | 638 | 2,719 | 35 |
| Manufacturing | 2,603,180 | 31,153 | 144,621 | 1731 |
| Electricity | 34,479 | 298 | 1,916 | 17 |
| Water supply | 75,881 | 1,495 | 4,216 | 83 |
| Construction | 1,806,319 | 10,204 | 100,351 | 567 |
| Trade | 3,405,777 | 43,556 | 189,210 | 2,420 |
| Transportation | 637,931 | 9,527 | 35,441 | 529 |
| Accom. and food | 624,125 | 909 | 34,674 | 51 |
| ICT | 376,294 | 5,128 | 20,905 | 285 |
| Financial and insura | 85,096 | 1,089 | 4,728 | 61 |
| Real estate activitie | 440,860 | 2,356 | 24,492 | 131 |
| Prof. services | 908,278 | 9,365 | 50,460 | 520 |
| Administrative | 435,749 | 5,009 | 24,208 | 278 |
E. Empirical Results and Counterfactual Simulations
18. We find evidence that higher regulatory barriers are associated with lower productivity and higher mark-ups. Product market restrictions have a significant and negative impact on firms’ value-added growth, as well as labor productivity and TFP growth (Table 3). Figure 2 chart 1 shows standardized coefficients β1 (blue bars) for the Equation 1, that estimates the direct impact of regulatory restrictions on the productivity indicators for the sample of companies in the regulated service sectors. The estimations show that an increase in regulations by one standard deviation will hurt TFP growth of these companies by 1.5 percentage points, labor productivity growth by 0.5 to 1 percentage points depending on the measurement, and value-added growth by more the 1 percentage point. These results are in line with previous literature findings. Moreover, and as expected from theoretical literature, we find evidence that regulatory barriers increase markups in the regulated sectors.
Belgium: Regressions with Direct Regulatory Effect
Belgium: Regressions with Direct Regulatory Effect
| (1) | (2) | (3) | (4) | (5) | |
|---|---|---|---|---|---|
| TFP growth | Labor Productivity Growth (VA) | Labor Productivity Growth (Output) | VA growth | Markup | |
| Size (lag value added) | −0.06*** | −0.07*** | −0.06*** | −0.08*** | 0.00*** |
| Debt to assets | −0.10*** | −0.05*** | −0.03*** | −0.07*** | −0.02*** |
| Direct Sectoral Regulation | −0.01*** | −0.01*** | −0.00*** | −0.01*** | 0.01*** |
| Time Dummies | Yes | Yes | Yes | Yes | Yes |
| Country Dummies | Yes | Yes | Yes | Yes | Yes |
| _cons | 1.08*** | 1.32*** | 1.02*** | 1.49*** | 0.17*** |
| Number of obs. | 1,443,501 | 1,443,501 | 1,443,501 | 1,443,501 | 1,443,501 |
| Number of firms | 399,401 | 399,401 | 399,401 | 399,401 | 399,401 |
| R-sq between | 0.02 | 0.02 | 0.01 | 0.02 | 0.09 |
Belgium: Regressions with Direct Regulatory Effect
| (1) | (2) | (3) | (4) | (5) | |
|---|---|---|---|---|---|
| TFP growth | Labor Productivity Growth (VA) | Labor Productivity Growth (Output) | VA growth | Markup | |
| Size (lag value added) | −0.06*** | −0.07*** | −0.06*** | −0.08*** | 0.00*** |
| Debt to assets | −0.10*** | −0.05*** | −0.03*** | −0.07*** | −0.02*** |
| Direct Sectoral Regulation | −0.01*** | −0.01*** | −0.00*** | −0.01*** | 0.01*** |
| Time Dummies | Yes | Yes | Yes | Yes | Yes |
| Country Dummies | Yes | Yes | Yes | Yes | Yes |
| _cons | 1.08*** | 1.32*** | 1.02*** | 1.49*** | 0.17*** |
| Number of obs. | 1,443,501 | 1,443,501 | 1,443,501 | 1,443,501 | 1,443,501 |
| Number of firms | 399,401 | 399,401 | 399,401 | 399,401 | 399,401 |
| R-sq between | 0.02 | 0.02 | 0.01 | 0.02 | 0.09 |
Belgium: Estimation Results
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: ORBIS and IMF staff calculations.Belgium: Estimation Results
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: ORBIS and IMF staff calculations.Belgium: Estimation Results
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: ORBIS and IMF staff calculations.19. Indirect exposure to regulation is associated with lower productivity and low markups in downstream sectors. The estimations show a significant negative impact on the productivity growth of firms in connected industries, ranging from 0.3 to 0.5 percentage points. The sign of the impact on the mark-ups, however, is reversed: higher regulation reduces mark-ups in the downstream industries (Table 4). Figure 2 chart 1 shows standardized coefficients β2 (grey bars). The above results hold in a pooled regression of both samples and with added controls of infrastructure quality and aging (equation 3, Table 5). Figure 2 chart 2 shows standardized coefficients for equation 3.
Belgium: Regressions with Indirect Regulatory Effect
Belgium: Regressions with Indirect Regulatory Effect
| (1) | (2) | (3) | (4) | (5) | |
|---|---|---|---|---|---|
| TFP growth | Labor Productivity Growth (VA) | Labor Productivity Growth (Output) | VA growth | Markup | |
| Size (lag value added) | −0.05*** | −0.06*** | −0.05*** | −0.06*** | 0.00*** |
| Debt to assets | −0.08*** | −0.02*** | −0.01*** | −0.05*** | −0.02*** |
| Indirect regulatory effect | −0.03*** | −0.03*** | −0.03*** | −0.01*** | −0.01*** |
| Time Dummies | Yes | Yes | Yes | Yes | Yes |
| Country Dummies | Yes | Yes | Yes | Yes | Yes |
| _cons | 0.81*** | 0.81*** | 0.81*** | 1.12*** | 0.08*** |
| Number of obs. | 6,779,865 | 6,779,865 | 6,779,865 | 6,779,865 | 6,779,865 |
| Number of firms | 1,644,206 | 1,644,206 | 1,644,206 | 1,644,206 | 1,644,206 |
| R-sq between | 0.02 | 0.02 | 0.02 | 0.02 | 0.01 |
Belgium: Regressions with Indirect Regulatory Effect
| (1) | (2) | (3) | (4) | (5) | |
|---|---|---|---|---|---|
| TFP growth | Labor Productivity Growth (VA) | Labor Productivity Growth (Output) | VA growth | Markup | |
| Size (lag value added) | −0.05*** | −0.06*** | −0.05*** | −0.06*** | 0.00*** |
| Debt to assets | −0.08*** | −0.02*** | −0.01*** | −0.05*** | −0.02*** |
| Indirect regulatory effect | −0.03*** | −0.03*** | −0.03*** | −0.01*** | −0.01*** |
| Time Dummies | Yes | Yes | Yes | Yes | Yes |
| Country Dummies | Yes | Yes | Yes | Yes | Yes |
| _cons | 0.81*** | 0.81*** | 0.81*** | 1.12*** | 0.08*** |
| Number of obs. | 6,779,865 | 6,779,865 | 6,779,865 | 6,779,865 | 6,779,865 |
| Number of firms | 1,644,206 | 1,644,206 | 1,644,206 | 1,644,206 | 1,644,206 |
| R-sq between | 0.02 | 0.02 | 0.02 | 0.02 | 0.01 |
Belgium: Regressions with TFP growth as Dependent Variable
Belgium: Regressions with TFP growth as Dependent Variable
| Labor Productivity Growth (Output) | |
|---|---|
| Size (lag value added) | −0.05*** |
| Debt to assets | −0.10*** |
| Direct Sectoral Regulation | −0.01*** |
| Downstream Regulation Spillover | −0.06*** |
| Infrastructure Quality | 0.03*** |
| Ageing Workforce | −0.00*** |
| Constant | 0.75*** |
| Number of obs. | 2,977,397 |
| Number of firms | 1,043,826 |
| R-sq | 0.06 |
Belgium: Regressions with TFP growth as Dependent Variable
| Labor Productivity Growth (Output) | |
|---|---|
| Size (lag value added) | −0.05*** |
| Debt to assets | −0.10*** |
| Direct Sectoral Regulation | −0.01*** |
| Downstream Regulation Spillover | −0.06*** |
| Infrastructure Quality | 0.03*** |
| Ageing Workforce | −0.00*** |
| Constant | 0.75*** |
| Number of obs. | 2,977,397 |
| Number of firms | 1,043,826 |
| R-sq | 0.06 |
20. The quality of infrastructure has a significant positive impact on the productivity of industries that have high transportation costs. An increase in the quality of infrastructure by one standard deviation results in an increase in labor productivity growth by 1.4 percentage points.
21. A higher share of older workers in overall employment has a negative impact on productivity. An increase in the share of 55+ workers in the sector by one standard deviation reduces labor productivity growth by 0.7 percentage points. This result should be interpreted with caution as the aging structure at sectoral level may reflect the structure of large companies in a given sector. Firm-level data on worker’s age would have provided better estimate of the aging effect, however, was not available for our sample.
22. Bringing regulatory practices to the average level in the OECD could have a very significant positive impact on productivity in Belgium. Under a hypothetical policy scenario, where all sector-specific regulation is reduced to the OECD average, firms total factor productivity could increase by 0.2 to 1 percentage points depending on the sector (Figure 3 chart 2). The potential gains are the highest for construction, ICT, accommodations, food, and retail industries. Most of the total gains for the economy are coming from the indirect effect of deregulation (Figure 3 chart 1). Compared to neighboring countries, Belgium stands to gain more than Italy but less than France from reducing barriers to competition. This counterfactual exercise should be interpreted with caution. While the productivity level increases in the counterfactual scenario, some inefficient firms might be forced to exit the market. The magnitude of potential transition costs from eliminating incumbent protection will depend on the cyclical position and whether macroeconomic policy support is provided.
Belgium: Total Factor Productivity Gains
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: ORBIS and IMF staff calculations.Belgium: Total Factor Productivity Gains
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: ORBIS and IMF staff calculations.Belgium: Total Factor Productivity Gains
Citation: IMF Staff Country Reports 2018, 072; 10.5089/9781484346068.002.A001
Sources: ORBIS and IMF staff calculations.23. Policymakers could consider a variety of measures that might boost productivity over the medium term, including promoting competition in service sectors and investing in public infrastructure and human capital. The degree of competition in different markets depends on many factors, including technology and product market specifics (e.g., high entry costs due to the fixed investments required), obstacles imposed by self-regulatory associations (e.g., additional licenses, bans on advertisement), and regulatory restrictions that raise barriers to entry and protect the incumbents. A review of the regulations that may limit competition in different sectors could help policymakers develop productivity-enhancing reform options. This could be usefully complemented by providing additional support to the institutions that are in charge of addressing anti-competitive behavior. Increasing the efficiency of bankruptcy procedures could also help raise firm dynamism and reduce resource misallocation. Improving the quality of infrastructure is equally important for raising productivity growth, especially for sectors that have high transportation costs.5 Addressing the productivity challenges of an aging workforce will require a comprehensive set of measures that target the demand and supply sides of training and lifelong learning.
F. Conclusions
24. Belgium’s subdued productivity growth can be explained by a combination of sectoral employment shifts, barriers to competition, the declining quality of infrastructure, and an aging workforce. The shift of employment toward lower productivity service sectors, common to many advanced economies, has been more pronounced in Belgium and explains half of the productivity gap with neighboring countries. Population aging is another secular factor that has contributed to the productivity slowdown. In addition, barriers to competition in some service sectors have lowered productivity growth and raised rents in these sectors. Higher prices and lack of competition in upstream sectors can cause significant negative spillovers, as higher indirect exposure to regulatory barriers is associated with lower productivity in downstream sectors. Bringing regulatory practices to the average level in the OECD could have a significant positive impact on productivity in Belgium. Another important factor is the declining quality of transport infrastructure, which has adversely affected productivity in the sectors using it intensively.
References
Adler, G., Duval, M.R.A., Furceri, D., Sinem, K., Koloskova, K. and Poplawski-Ribeiro, M., 2017. Gone with the Headwinds: Global Productivity. International Monetary Fund.
Aiyar, M.S. and Ebeke, M.C.H., 2017. The Impact of Workforce Aging on European Productivity. International Monetary Fund.
Andrews, Dan, Chiara Crisculo and Peter N. Gal (2015). “Frontier Firms, Technology Diffussion and Public Policy: Micro Evidence from OECD Countries”, OECD
Ariu, A. / Vandenberghe, V. (2014), “Assessing the role of aging, feminising and better-educated workforces on TFP growth” Working Paper N° 265
Biatour, B. and Kegels, C., 2017. Working Paper 11–17-Growth and Productivity in Belgium (No. 1711).
Blanchard, O. and Giavazzi, F., 2003. Macroeconomic effects of regulation and deregulation in goods and labor markets. The Quarterly Journal of Economics, 118(3), pp.879–907.
Bourlès, R., Cette, G., Lopez, J., Mairesse, J. and Nicoletti, G., 2010. Do product market regulations in upstream sectors curb productivity growth? Panel data evidence for OECD countries (No. w16520). National Bureau of Economic Research.
Cette, G., Lopez, J. and Mairesse, J., 2016. Market regulations, prices, and productivity. American Economic Review, 106(5), pp.104–08.
Dabla-Norris, M.E., Guo, M.S., Haksar, M.V., Kim, M., Kochhar, M.K., Wiseman, K. and Zdzienicka, A., 2015. The new normal: A sector-level perspective on productivity trends in advanced economies. International Monetary Fund.
Dhyne, E. and Fuss, C., 2014. Main lessons of the NBB’s 2014 conference “Total factor productivity: measurement, determinants and effects”. Economic Review, (iii), pp.63–76.
Feyrer, J., 2007. Demographics and productivity. The Review of Economics and Statistics, 89(1), pp.100–109.
Van Ark, B., 2014. Total factor productivity: Lessons from the past and directions for the future (No. 271). Working Paper Research.
Rouzet, D. and Spinelli, F., 2016. Services trade restrictiveness, mark-ups and competition.
Prepared by Anna Shabunina (EUR).
Some studies focusing on the network sectors regulation estimate forward linkages of regulation, e.g. impact of higher output of the regulated sectors on the demand for intermediate inputs from upstream sectors. This was not the focus of our paper.
“Market Functioning in Belgium: Horizontal Screening of the Sectors (2016),” http://economie.fgov.be/en/modules/publications/statistics/economy/marche_en_belgique_screening_2016.jsp
Please see accompanying SIP “Simulating an Increase in Public Investment”
Please see Selected Issues Paper “Simulating an Increase in Public Investment” for the discussion of policy options and recommendations.
References
Baxter, M. and King, R.G., 1993. “Fiscal policy in general equilibrium.”, The American Economic Review, pp.315–334.
Bom, Pedro RD, and Jenny E. Ligthart. “What have we learned from three decades of research on the productivity of public capital?”, Journal of economic surveys 28, no. 5 (2014): 889–916.
Broyer, S. and J. Gareis, 2013, “How large is the infrastructure multiplier in the euro area?”, Flash Economics by Natixis, No. 227.
Elekdag, S. and Muir, D., 2014. “Das Public Kapital: How Much Would Higher German Public Investment Help Germany and the Euro Area?”, IMF Working Paper No. 14/227.
European Central Bank, 2017, “The effect of public investment in Europe: a model-based assessment”, Working paper No 2021.
European Commission, 2017, “Country Report Belgium 2017”, Commission Staff Working Document SWD (2017) 67.
European Commission, 2014, “Infrastructure in the EU: Developments and Impact on Growth”, Occasional Papers 203.
Federal Planning Bureau, 2017, “Public Investment in Belgium Current State and Economic Impact”, FPB Working Paper 1–17.
Feenstra, R. C., Obstfeld, M. Russ, K. N., 2012. “In search of the Armington elasticity”, mimeo.
International Monetary Fund, 2016, “Belgium—Making Public Expenditure more Efficient”, IMF Country Report No. 16/78 (Selected Issues Paper).
International Monetary Fund, 2014, “Is it time for an Infrastructure Push? The macroeconomic Effects of Public Investment”, World Economic Outlook, October 2014, Chapter 3.
Jong-A-Pin, R. and J. De Haan, 2008, “Time-varying impact of public capital on output: new evidence based on VARs for OECD countries”, EIB Papers, Vol. 13 (1), 57–81.
Kamps, C., 2005, “The dynamic effects of public capital: VAR evidence for 22 OECD countries”, International Tax and Public Finance, Vol. 12 (4), 533–558.
Laxton, M.D., Mursula, S., Kumhof, M.M. and Muir, D., 2010. “The Global Integrated Monetary and Fiscal Model (GIMF): Theoretical Structure”, IMF Working Paper No. 10/34.
Leeper, E.M., Walker, T.B. and Yang, S.C.S., 2010. “Government investment and fiscal stimulus.” Journal of monetary Economics, 57(8), pp.1000–1012.
National Bank of Belgium, 2016, “Should government investment be promoted?”, NBB Economic Review September.
Prepared by Simon Voigts (EUR).
Data shown in Figures 1 and 3 is compiled in accordance with European System of National and Regional Accounts (ESA 2010) rules, in which the definition of general government excludes entities controlled by the government but considered “market producers”. The amount of fixed assets held by these entities may differ across countries. Hence, the picture presented here may not reflect the evolution of fixed assets of the broader public sector. There is no directly available data based on a more comprehensive definition of the government.
See European Commission (2014) for a recent survey.
See Kamps (2005), Jong-A-Pin and De Haan (2008), Broyer and Gareis (2013), IMF (2014), and ECB (2017) for examples.
FPB (2017) considers a budget-neutral expenditure shift as an alternative funding scenario but does not discuss it in detail.
While GIMF is more detailed than QUEST III (GIMF features, for example, overlapping-generation households and a financial accelerator mechanism), the basic structure of both models is very similar.
Transport infrastructure is a main constraint holding back productivity growth (see for example EC, 2017). However, there are also large investment needs in other areas, as for example education and energy infrastructure.
While the model is comparatively detailed, it remains stylized and does not capture all relevant country-specific factors, including the very high import content of exports in Belgium (and the importance of transit trade). As discussed below, this can have significant implications regarding the magnitude of model-implied effects.
Simulating the standardized experiment of a two-year, 1 percent of GDP per year debt-financed increase in public investment allows for a comparison with other studies. Under this calibration, the experiment raises output by 0.72 percent after one year (the peak response), and by 0.64 percent after five years (the adjustment is not shown). This is a weaker adjustment then Elekdag and Muir (2014) report for Germany (about 1.1 and 0.72 percent after one and five years respectively, under monetary accommodation). The weaker short-run adjustment can be explained by Germany’s significantly smaller import share, which implies weaker import leakages of the stimulus. The medium- term adjustment is broadly in line as it is strongly influenced by the Elasticity of TFP in the production of final goods w.r.t. public capital stock, which is the same as in the calibration at hand.
Higher long-run investment (3.2 instead of 2.2 percent of GDP) is required to offset increased depreciation in absolute terms.
While IMF (2016) argues that there is significant room for efficiency improvements in the public sector, government purchases on goods and services stand at 4 percent of GDP in Belgium, which would have to be halved.
The model is linear in its approximation around the steady state.
The graph shows rising government spending because it is constant as a share of current GDP and thus increasing as a share of initial GDP.
The public capital stock can be thought of as an input into the production process that is free of charge. An increase thus allows to produce more output for a given amount of labor and capital, or, equivalently, to produce a given output at lower production costs.
Figure 6 shows the average terms of trade of intermediate goods and final goods, weighted by their steady-state share in total exports. The deterioration is particularly strong for final goods, reflecting the model assumption that a higher stock of infrastructure capital increases total factor productivity in the production of final goods (which takes imported intermediate goods as input), implying lower production costs and export prices. The decline in consumer prices is less pronounced because of imported consumer goods.
This implies a long-run elasticity of output with respect to public capital of 0.14 (the public capital stock increases by about 41 percent relative to its initial value). Throughout the exercise, additional tax revenues stemming from the growing tax base are paid to households as non-distortionary transfers.
Belgium has a comparatively high share of re-exports and transit trade in total trade.
The baseline policy is preferable if the decline in government consumption is achieved via efficiency improvements that do not reduce household utility.
The evolution of the public capital stock is determined by the change in public investment, which is identical across all considered policy scenarios.
It is assumed that government bond yields remain constant, which abstracts from possible adverse feedback from higher debt to higher interest rates. However, during the sovereign-debt crisis, government bond spreads proved to be very resilient for a highly indebted country.
As simplified numerical example, consider public debt and the public per-capita capital stock to stand at 106 and 36 percent of GDP respectively, and abstract from interest costs, population growth and capital depreciation. Holding GDP constant, debt-financed public investment of one percent of GDP would add one percentage point to the long-run debt level, which increases the numerator of the debt-to-GDP ratio by 1%/106%=0.94%. To take GDP growth into account, note that the additional investment adds one percentage point to the long-run per-capita capital stock, which is an increase of 1%/36%=2.8%. Assuming a long-run output elasticity w.r.t. public capital of 0.2 (which is above the average estimate reported in Bom and Ligthart, 2014), GDP increases by 2.8%*0.2=0.56%. The debt-to-GDP ratio rises as the numerator grows by more than the denominator. If we accounted for interest costs, population growth and depreciation, debt-to-GDP would increase by more because debt-financed investment would expand long-run debt by more than 1-to-1 while it would increase long-run per-capita capital by less than 1-to-1.
