Selected Issues


Selected Issues

Wage Moderation in the Netherlands1

A. Introduction

1. Wage growth has been subdued in the Netherlands despite tighter labor market conditions in recent years. With employment growth picking up since 2014 and labor force participation rising steadily, the unemployment rate has been falling from close to 8 percent in 2014 to 4.4 percent in last quarter of 2017, back to the pre-crisis range. However, nominal wage growth remains subdued at around one percent and real wage growth close to zero, much lower than the wage growth before the crisis. This seems to suggest a flattening of the wage Phillips curve where falling unemployment fails to generate higher wage growth.


Unemployment Rate and Nominal Wage Growth

(Year-over-year percent change)

Citation: IMF Staff Country Reports 2018, 131; 10.5089/9781484357866.002.A001

Sources: CBS, Eurostat, and IMF staff calculations.

Nominal and Real Wage Growth

(Year-over-year percent change)

Citation: IMF Staff Country Reports 2018, 131; 10.5089/9781484357866.002.A001

Sources; Eurostat, and IMF staff calculation.

2. The “wage puzzle” is shared by many other advanced economies, though the wage growth in the Netherlands is among the weakest. The latest World Economic Outlook (October 2017) finds that many advanced economies have seen a disconnect between their headline unemployment rates and nominal wage growth in recent years. As in the Netherlands, unemployment rates have been declining in these countries since 2014, and this decline has largely reflected job creation. Yet, nominal wage growth has been broadly stable and remains below pre-crisis ranges for almost all these countries. For around three-quarters of these countries, real wage growth is below its pre-crisis range. However, wage growth in the Netherlands is one of the weakest among the advanced EU countries, even including countries whose labor market conditions are more behind the curve.


Advanced EU: Nominal Wage Growth

(Year over year percent change)

Citation: IMF Staff Country Reports 2018, 131; 10.5089/9781484357866.002.A001

3. While negotiated wages are showing signs of moderate pick-up, wage drift remains subdued, weighing on the final wage growth. Since the crisis and up to 2017, negotiated wages were growing in the range of 1 to 2 percent, which is Sources: CPB Netherlands Bureau for Economic Policy Analysis. slower than the growth rates during 2000—2007 (close to 3 percent); though they had been rising faster in the recent couple of years and are expected to grow by 2.2 and 3.2 percent in 2018 and 2019 respectively, compared with 1.5 percent in 2017. In the meantime, wage drift stood out being exceptionally low in the recent years, and even turned negative in 2014 and 2016, which is quite puzzling as firm profitability is high in the Netherlands compared with other countries, and has remained solid and steadily rising after the crisis. Some anecdotal evidence suggests that this could be due to compositional changes with higher share of low-skilled workers in the labor force. Yet the puzzle of low wage drift remains, which makes it difficult to predict the final wage growth in the coming years despite rising negotiated wage growth.


Negotiated Wages and Wage Drifts


Citation: IMF Staff Country Reports 2018, 131; 10.5089/9781484357866.002.A001

Sources: CPB Netherlands Bureau for Economic Policy Analysis.

Net Return on Capital 1/


Citation: IMF Staff Country Reports 2018, 131; 10.5089/9781484357866.002.A001

Source: Eurostat, and IMF staff calculations.1/ Net return on capital – net operating surplus and mixed income / fixed assets.

4. Including self-employed income, the total labor cost share has shown a modest decline since the global financial crisis. The recent trend is not very different from other EA or EU countries. In the Netherlands, rising labor market dichotomy/duality has made self-employed income an important share of total labor income. Without accounting for self-employed income, the Netherlands’ employees’ total labor cost share would have shown a declining trend after the crisis. Over a longer-term horizon, the total labor cost share has been trending down, broadly in line with other EA and EU countries, possibly driven by factors such as automation, and rising globalization.


Total Labor Cost Share

(Employees’ total labor compensation and self employed income in share of GDP)

Citation: IMF Staff Country Reports 2018, 131; 10.5089/9781484357866.002.A001

Sources: Eurostat, IMF Staff Calculations

5. The following sections examine the fundamental factors driving the Netherlands’ “wage puzzle” as highlighted above. Section B summarizes the Netherlands’ wage bargaining system and the existing literature on the wage developments both in the Netherlands as well as in the international context. Section C highlights the key stylized facts that motivates the empirical analysis. Section D and E describe the model specifications and underlying data. Section F illustrates the regression results; and section G concludes.

B. Background of Wage Formation in the Netherlands and Literature Review

6. Wage outcomes in the Netherlands in part reflect the nature of the Netherlands’ wage bargaining system (Box 1). The collective bargaining process takes place predominantly at the sectoral level, while leaving certain scope of flexibility at the firm/local level. While the wage bargaining process is relatively independent across sectors; wages agreed in other sectors do not deviate much from the ones set in the export-oriented industry sector, which can be influenced by foreign wages, especially among the trading competitors. The bargaining process covers not only the pay, but also types of contracts (e.g. temporary or permanent), retirement and leave benefits, etc. Increasingly, there has been a trend towards more firm-level and individualized wage arrangements. The recent coalition agreement pointed out the importance of having such arrangements to allow the pay to better reflect workers’ improving performance and productivity.

7. It is well documented in the literature that wage growth has slowed down significantly both in nominal and real term since the global financial crisis (e.g. Eggelte et al. 2014). The slow wage growth has been largely driven by remaining slack in the labor market (DNB 2017) and rising labor market flexibility (DNB 2018). Peeters and den Reijer (2001) documented a shift in the bargaining power from employees to the employers since 1990s that may have also contributed to the recent low wage growth. Going forward, the tightening labor market is expected to increase negotiated wages from 1.5 percent in 2017 to 2.2 percent in 2018, the number of permanent contracts and tariffs of self-employed; though the wage development is projected to remain relatively moderate (CPB 2017/18).

8. Analysis on other advanced economies finds that subdued wage growth can also be attributed to changes to employment structure, technological advancement, etc. among other factors. IMF (2017) finds that while the bulk of the recent wage slowdown in advanced economies (AEs) can be explained by lower trend productivity, labor market slack that is not captured by the traditional unemployment measures, in many AEs, given increases in part-time employment and temporary contracts. These developments may also point to persistent changes in relationships between firms and workers in response to technological change and remaining labor market rigidities in some countries that deter employers from hiring on standard full-time contracts. Beyond these factors, it is found that automation and diminished medium-term growth expectations may have also contributed to slower wage growth.

Netherlands’ Wage Bargaining System

The collective bargaining process takes place predominantly at the sector level, but leaves certain scope of flexibility at the firm/local level in the Netherlands. These national or sectoral agreements define the broad framework of wage setting but leave large scope for bargaining at the firm/establishment level. Wage negotiations take place in their respective sector, and are independent of negotiations in other sectors. Sectors can set minimum or standard terms of employment (a “wage floor”) which employers can complement or deviate from upwards at firm level; or allow workers and employers to choose “à la carte” and trade off wages against working conditions.1 This form of wage setting is also common in the Scandinavian countries. In contrast, in countries like Germany, national or sector agreements allow and define the conditions for deviations at lower levels via the so-called opening or opt-out clauses.

The sectoral wage bargaining is led by the industry sector, followed by other sectors. There are three major unions: the Christian-Democratic Christelijk Nationaal Vakverbond (CNV), the social-democratically oriented Federatie Nederlandse Vakbeweging (FNV) and the Federation of Managerial and Professional Staff Unions (VCP). All are federations of sector-based labor unions. They bargain with employer’s organization over wage formation. The Confederation of Netherlands Industry and Employers (known as VNO-NCW) is the largest employers’ organization in the Netherlands. VNO-NCW represents the common interests of Dutch business and cover almost all sectors of the economy, including more than 80 percent of all medium-sized companies and nearly all the large corporate institutions.


Bargaining Levels

(share of employees covered by collective bargaining system)

Citation: IMF Staff Country Reports 2018, 131; 10.5089/9781484357866.002.A001

Sources; OECD Employment Outkook 2017, Chapter 4.

Collective agreements are legally binding for Dutch union members but employers are obliged to offer the same terms to non-union members, so in practice all employees are covered. In addition, the parties to a collective agreement can ask the government to make its term generally binding on all employees in an industrial sector. For this to happen, the agreement must already cover a “substantial proportion” of those employed in the industry—normally 55 percent or more. The agreed collective agreements may include not only pay but also working conditions issues, early retirement, educational leave, the organization of leave over the whole of an employee’s working life, the position of women, protecting those with disabilities, and the environment. Increasingly agreements provide for a range of benefits, from which individual employees can choose.

The system of wage formation in the Netherlands is characterized by high penetration (> 80 percent including extension) and relative stability. However, there is a recent trend of decreasing membership rates owing to fewer membership sign-up by young workers, and an increasing share of the self-employed who are not covered under a collective agreement nor are they union members). The aging union membership and decreasing unionization rates cause a widening gap between union density (currently 24 percent) and collective bargaining coverage. There is no formal mechanism in the case of failed collective bargaining and real failure is very exceptional. In this case, the content of the former agreement is still applied. In exceptional cases, such as strikes, the state may take the initiative to appoint a mediator (with the agreement of the parties involved).

Although most bargaining continues to take place at the sectoral level, the number of company agreements is on the rise. In 2013, there were 182 agreements covering normal pay and conditions issues signed at industry level and 519 company collective agreements covering together 5.9 million employees. 10 percent of all wage agreements were covered though company collective agreements which predominate in the largest companies such as Philips, DSM and Shell. There is also a tendency for industry level agreements to become framework agreements, with some of the detailed provisions being negotiated at company level. Another recent trend is towards individualization of pay setting, mainly in the form of the introduction and/or spread of flexible pay forms. The decline of collective labor agreements may furthermore be explained partly by economic trends such as globalization, technological progress and a decline in union membership, possibly leading to more flexible labour markets and lower negotiation power by unions (De Ridder and Euwals, 2016).

1 The two main wage agreement laws are: the Law on the collective agreement (Wet op de collectieve arbeidsovereenkomst, WCAO), dating from 1927; and the Law on the extension of collective agreements (Wet op het verbindend en onverbindend verklaren van collectieve arbeidsovereenkomsten, WAVV), dating from 1937.

C. Stylized Facts

9. Recent wage moderation can be largely attributed to falling labor productivity and subdued inflation expectations. Trend labor productivity declined from above 2 percent to around 1 percent after the crisis. Inflation also remains subdued. Going forward, inflation is projected to rise only gradually to the 2 percent target.


Actual and Trend of Productivity Growth

(Year-over-year percent change)

Citation: IMF Staff Country Reports 2018, 131; 10.5089/9781484357866.002.A001

Sources; Eurostat, and IMF staff calculations.1/ HP filter with lamda at 1600

10. Higher flexibility in the labor market may have also contributed a more complicated relationship between wages and unemployment. Motivated by the prospects of avoiding high pension and social security contributions, more workers are willing to become self-employed. Their work arrangements tend to be more flexible, just like workers hired under the flexible/fixed term contracts, which make them more cost competitive to hire than regular employees. As a result, the share of workers becoming self-employed and/or under temporary contracts are steadily rising, much faster than other EU countries. As suggested by DNB (2018), this may have weighed on employees’ wage bargaining power. In this context, labor unions’ prioritization in making labor contracts more permanent and employment protection more stringent during the wage bargaining process have also created additional burden for faster wage growth.


Temporary and Self Employment

(Percent of total employment)

Citation: IMF Staff Country Reports 2018, 131; 10.5089/9781484357866.002.A001

Sources: Eurostat, and IMF staff calculations.

11. In addition, despite falling unemployment, some non-traditional indicators point to remaining slack in the labor market. Although unemployment has fallen from above 8 percent to less than 5 percent, close to the pre-crisis range; other indicators such as involuntary part-time employment and long-term unemployment remain elevated. This suggest that labor market slack has not yet entirely diminished.


Labor Market Indicators


Citation: IMF Staff Country Reports 2018, 131; 10.5089/9781484357866.002.A001

Sources: Eurostat, and IMF staff calculation.1/ in share of total employment.2/ in share of total unemployed.

12. The Netherlands’ wage growth has been moving closely with or even more slowly than that in Germany and other Euro Area countries. Like other EA countries, wage growth in the Netherlands followed its trading competitors quite closely, including during the period of wage moderation in Germany (2003–05). Post-crisis wages have been growing even more slowly than most euro area countries (due to a higher unemployment gap during the recession), and much more slowly than Germany, despite similar trend productivity growth, labor market slack, and expected inflation.


Nominal Wage Growth

(Year-over-year percent change, 4-quarter moving average)

Citation: IMF Staff Country Reports 2018, 131; 10.5089/9781484357866.002.A001

Sources: Eurostat, and IMF staff calculations.

D. Model Specification

13. We use a wage curve model, based on Blanchard (1997), to analyze the drivers of Netherlands’ moderate wage growth in recent years. A wage curve relationship can be derived from a range of underlying models, including wage bargaining and efficiency wages. In equilibrium, the wage curve implies that the level of real wages is negatively related to aggregate unemployment (U). Extending this model to apply over time, it is necessary to augment this relationship with labor productivity (TLP). But there are significant costs to adjusting wages, especially to reducing wages, so in practice wages are more likely to follow a measure of the trend in labor productivity rather than actual labor productivity which is subject to cyclical swings and various other shocks.

log RW = α log TLP – βU

14. An error-correction specification (ECM) is used to estimate the wage curve.2 This specification appears similar to the more widely used Phillips curve in wages, except for the inclusion of an error correction term that depends on the lagged levels of real wages and trend labor productivity. When real wages deviate from their long run equilibrium level in relation to trend labor productivity, this error correction term impacts nominal wage growth and promotes a return to equilibrium.

15. The wage curve ECM is enriched with structural variables in addition to domestic labor market slack indicators, inflation expectations, and international labor conditions. Real wages (RW) and trend labor productivity (TLP) are in the error correction term. Nominal wages (W) are determined by domestic cyclical factors (D), including inflation expectations, unemployment gap, and additional labor market slack indicators (e.g. involuntary part-time employment); foreign cyclical factors (F), including German wage growth and foreign labor market slack (unemployment rate and involuntary part-time employment in the Euro Area), and structural factors (S) such as share of temporary, self-employed workers. We also include the interaction between the structural variables with unemployment to examine the effect of structural changes in the labor market on the Phillips curve relationship. Altogether:


Note that the difference operator covers 4 quarters, reflecting the common approach of adjusting wages annually, and the error correction term refers to real wages and trend productivity 4 quarters earlier. The specification also controls for lagged dependent variable, dlogWt-4 to reflect possible inertia or base effects.3 A statistically significant negative coefficient on lagged real wages RWt-4 would indicate cointegration between real wages and trend productivity. The parameter ø1 is expected to be approximately equal to ø2, such that real wages grow broadly at the pace of trend productivity in equilibrium.

E. Data

  • Nominal wages: Our analysis is mainly based on total labor compensation from the national accounts as a ratio to hours worked, but we also test wages and salaries from the national accounts as a ratio to hours worked for robustness check. The key difference between the two measures is that total compensation includes employers’ social security contributions. Compared with some short-term wage statistics, measures from the national accounts reflect structural changes in the composition of the labor force (by sector, occupation, and skill level). This makes the national accounts measures more consistent with the measure of labor productivity, which is an average across a changing composition of jobs. The labor cost index (LCI) wages and salary measures are based on a more stable basket of jobs, and include bonuses and benefits (e.g. car, health care, sick leave), same as the wages and salaries measure from the national accounts; but the LCI wage and salary measures only cover the business sector.

  • Real wages: the nominal wage indicator is deflated by the GDP deflator, not a consumer price measure. This ensures that real wages and real hourly labor productivity are measured consistently. It also reflects that firm’s capacity to pay depends on the price of output.

The above and other variables are summarized in the table below:

F. Estimation Results from the Netherlands’ Models

16. The estimation results indicate that wage formation has been influenced by both domestic fundamentals and international spillovers in recent decades. The above ECM specification is estimated for the period of 1995Q1 to 2017Q1 with labor compensation per hour from national accounts as the dependent in Table 1. The coefficients on the lagged level of the real wage are statistically significant in all variants of the equation, indicating cointegration between real wages and productivity, with the rate of error correction ranging from −0.5 to −0.7. Variants on the general specification are reported as follows:

  • Models 1–3 include only domestic variables.4 In addition to the ECM term (difference between real wages and trend productivity), significant variables include both the unemployment gap, expected inflation and labor productivity growth. Involuntary PT employment does not seem to be significant.

  • Models 4–5 add indicators of labor market slack in the euro area (EA). Changes in the EA unemployment gap are found to be statistically significant, while domestic labor slack indicators remain statistically significant.

  • Models 6–9 add German or EA compensation/wage growth. Foreign wages are found to have a significant impact on Dutch compensation growth, though the significance is higher for the German wages. In the more complete models (7–9), the coefficients on expected inflation, euro area unemployment changes and domestic unemployment gap remain statistically significant.

  • Models 10–14 add the structural changes in the form of employment. The rising share of temporary and self-employed employment are each found to have significant negative impact on wage growth. They are also found to reduce the responsiveness of the wage growth to unemployment gap. But when temporary and self-employment are both included in the regression at the same time, the effect of self-employed share becomes less important.

Definition of Key Variables

article image

17. The long run regression suggests that compensation has been growing more slowly than trend productivity. The ECM term from the regression 7 show that 1 percent increase in productivity is associated with 0.84 percent increase in the compensation, and the coefficient is significantly less than 1.


Nominal Compensation Growth

(Year-over-year percent change)

Citation: IMF Staff Country Reports 2018, 131; 10.5089/9781484357866.002.A001

Sources: Eurostat, and IMF staff calculations.

log RW = 0.84logTLP −0.10U

18. This can be partly explained by the rising flexibility in the labor market. If the change in labor market structure, e.g. the rising share of temporary employment is included, labor compensation growth becomes more aligned with the trend productivity; suggesting that the rising share of temporary and self-employed workers may have lowered the responsiveness of the real wage to trend productivity and unemployment over the long run perhaps due to reduced bargaining power of the more flexible employees.


Real Product Wage

(Log level)

Citation: IMF Staff Country Reports 2018, 131; 10.5089/9781484357866.002.A001

Sources: Eurostat, and IMF staff calculations.

19. The error correction model shows that wage growth reacts to both domestic and foreign factors and explains more than 80 percent of the total variation of the actual wage development. The domestic wage curve model (red line in text chart), which includes unemployment, productivity and expected inflation, has an R-square of 80 percent. The spillover effects from the euro area labor market conditions and German wage growth and structural changes of employment are significant, and contributes an additional 10 percent to the model fit; though neither of their contribution is sizable.

20. The decomposition exercise suggests that the recent moderation is associated with sluggish productivity growth, moderate wage growth in the EA, and changes in the form of employment over time. Decomposing the nominal compensation growth into contributions from variables included in the most complete short-run regression (model 14), it is found that lower productivity growth and expected inflation are the main factors inhibiting the recovery of labor compensation growth that traditional wage Phillips curve would imply. Recent slow wage growth in the EA, including in Germany since the Hartz reform, also weighed on wage growth as maintaining external competitiveness is an important factor in the wage formation process in the Netherlands. In addition, the negative contribution from the error correction term in recent years reflects structurally lower real wages driven by a rising share of temporary workers.


Decomposition of Nominal Wage Growth

(Year-over-year percent change)

Citation: IMF Staff Country Reports 2018, 131; 10.5089/9781484357866.002.A001

Sources: Eurostat and IMF staff calculations.

Demeaned Long Run Error Correction Terms

Citation: IMF Staff Country Reports 2018, 131; 10.5089/9781484357866.002.A001

Sources: Eurostat, and IMF staff calculations.

G. Conclusions

21. Besides various cyclical factors, rising labor market flexibility may have contributed to the wage moderation in the Netherlands. Like other advanced economies, slower productivity growth and lower expected inflation are important drivers to the wage moderation in the recent years. In addition to that, remaining slack in the labor market also weighed on wage growth. Like many other EA or EU countries, foreign wage growth has been showing strong spillovers to domestic wage development, especially for small open economies with strong trade exposures that strive to safeguard competitiveness. But more specifically to the Netherlands, rising labor market duality/flexibility with higher share of temporary and self-employed workers, may have also contributed to stagnant wage growth. Reforms to harmonize labor market employment contracts in a manner that increases flexibility but also allows greater bargaining power for the more “flexible” employees might allow both greater flexibility and higher wages.

22. Going forward, wages are expected to growth faster given higher expected inflation, foreign wage spillovers, and tightening labor market. With labor market slack diminishing further, inflation edging up, and foreign wages growing faster (e.g. the recent round of German wage negotiation), wage growth in the Netherlands might slowly pick up. Negotiated wages are projected to increase from 1.5 percent last year to 2.2 and 3.2 percent in 2018 and 2019 respectively; though the projection of wage drift remain unclear given possibilities of compositional changes, etc. It is also hard to predict whether the proposed policy measures in the coalition agreement that strive to balance the employment protection between temporary and permanent workers would have a positive or negative implication on wages. More research can be done in this area once the outcome of discussions with social partners produces a clearer outcome.

Table 1.

Netherlands’ Wage Equations

article image
Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1

lags range between 4 to 6 based on the highest significance level


  • Blanchard, Olivier (1998). “The Wage Equation,” April, MIT Press.

  • CPB Netherlands Bureau of Economic Policy and Analysis (2017). “Macroeconomic Outlook 2018.”

  • De Nederlandsche Bank (2017). “Economic Developments and Outlook.”

  • De Nederlandsche Bank (2018). “Labour market flexibilization linked to fall in labor income share”, DN Bulletin, February 2018.

  • De Ridder, Maarten and Rob Euwals (2016). “What are the Wage Effects of Extending Collective Labor Agreements? Evidence from the Netherlands,” CPB Background Document.

    • Search Google Scholar
    • Export Citation
  • Eggelte, Jurriaan, Jos Jansen, Guido Schotten and Diederik Dicou, (2014). “Wage Development Considered,” DNB Occasional Studies 1201, Netherlands Central Bank, Research Department.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund (2017). “World Economic Outlook,” October.

  • Graaf-Zijl, Marloes, Albert Horst, Daniel van Vuuren, Hugo Erken and Rob Luginbuhl, (2015). “Long-Term Unemployment and the Great Recession in the Netherlands: Economic Mechanisms and Policy Implications,” De Economist, 163, (4), 415434.

    • Search Google Scholar
    • Export Citation
  • Peeters, H.M.M. & Den Reijer, A.H.J. (2008). “On Wage Formation, Wage Development, and Flexibility: A comparison between European countries and the United States,” Applied Econometrics and International Development, Euro-American Association of Economic Development, Vol. 8(1), pages 5974.

    • Search Google Scholar
    • Export Citation
  • Peeters, H.M.M. & Den Reijer, A.H.J. (2001). “On Wage Formation, Wage Development and Unemployment,” WO Research Memoranda (discontinued) 677, Netherlands Central Bank, Research Department.

    • Search Google Scholar
    • Export Citation
  • Sargan, J. D. (1964). “Wages and Prices in the United Kingdom: A Study in Econometric Methodology,” 16, 2554. in Econometric Analysis for National Economic Planning, ed. by P. E. Hart, G. Mills, and J. N. Whittaker. London: Butterworths.

    • Search Google Scholar
    • Export Citation
  • Zhang, Yuanyan Sophia (2017). “Recent Wage Moderation in Sweden,” IMF


Prepared by Yuanyan Sophia Zhang and Dilyana Dimova. We are grateful for helpful comments from the authorities.


Sargan (1964) originated the estimation of an error correction model for wages. Zhang (2017) elaborate the model to include more labor slack indicators and external spillovers.


A base effect could arise in case of a temporary increase in the level of wages four quarters ago, which would tend to raise the y/y wage growth that quarter, while tending to lower the y/y growth rate four quarters later.


Model 1 is similar to the basic wage equation in Blanchard (1997).

Kingdom of the Netherlands - Netherlands: Selected Issues
Author: International Monetary Fund. European Dept.