In recent years, the IMF has released a growing number of reports and other documents covering economic and financial developments and trends in member countries. Each report, prepared by a staff team after discussions with government officials, is published at the option of the member country.


In recent years, the IMF has released a growing number of reports and other documents covering economic and financial developments and trends in member countries. Each report, prepared by a staff team after discussions with government officials, is published at the option of the member country.

The Swedish Labor Market—Responding to Global Challenges1

Over the past decades, Sweden has undertaken a broad range of structural reforms, including in the labor market. Yet, unemployment has remained stubbornly high and increasingly concentrated among the young and low-skilled. High unemployment despite a strong growth outlook suggests that structural factors play a role. Tackling these factors requires strengthening Sweden’s unique labor market model, including through lower employment protection, more focused active labor market policies, and greater firm-level wage flexibility. In addition, removing housing market inefficiencies and reforming the educational system would address increasing signs of labor market mismatch in a number of areas.

A. Structural Change in the Labor Market

1. Swedish unemployment is high despite otherwise strong economic dynamics. Since the onset of the crisis, unemployment averaged 7.7 percent, up from an average of about 6.7 percent during 2001–07 and a long-time low of about 5½ percent in 2000. At currently over 8 percent, the Swedish unemployment rate exceeds that of most other advanced European economies, despite having had the largest cumulative growth since 2009 among EU members. And strong economic growth, along with robust growth in credit, housing prices and employment, indicates a quickly closing output gap, projected to turn positive by end-2014.2


Unemployment Rate


Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: Statistics Sweden and Fund staff calculations.

Unemployment Rate, 2013


Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: OECD and Fund staff calculations.

2. Unemployment is largely structural. Okun’s Law—the empirical correlation between cyclical unemployment fluctuations and economic conditions—would have predicted a decline in unemployment to between 6.9–7½ percent in 2013, depending on whether output gap or growth dynamics are used. These calculations suggest that since 2007 the structural component of unemployment has increased by about ½–1 percentage points compared to pre-crisis levels. Alternative approaches indicate that structural increases could be even larger (Box 1).


Okun’s Law—Actual and Predicted Unemployment


Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: IMF World Economic Outlook and Fund staff calculations.Note: Based on a regression of the change in (level of) unemployment on GDP growth (output gap) during 1998-2013. The third series is based on the baseline regression (column 1) in Table 1 and includes both inflation and the output gap during 1960-2013.

3. The structural deterioration in unemployment has been asymmetric across labor market participants. The average employment rate for all workers reached nearly 76 percent in 2013:Q3, the highest in over 15 years. By contrast, employment rates among the low skilled, the young, and the immigrants have trended down over time, with the employment rate for young individuals (aged 15–24) at about 47 percent, young foreigners at about 26 percent, and the young low-skilled at 23 percent. Aggregate employment growth, while positive at about ¾ percent in 2013, was moderate in cross-country comparison and insufficient to bring down the unemployment rate in the context of a strong rise in the labor force.


Employment Rates, 2002Q1:2013Q3


Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: Eurostat and Fund staff calculations.

4. Labor force participation increased for several reasons. (Net) immigration flows increased sharply around 2005 and over the past decade accounted for an annual average of about 1 percent of the labor force. As immigrants are overrepresented in the young and low-skilled segments, they have contributed to the elevated unemployment rate. Labor force participation also increased domestically, including especially among the Swedish-born aged 15–24, likely reflecting improved employment expectations in the context of labor market reforms in recent years which focused on facilitating labor market entry of the young and low-skilled (Box 2), but which for many have not (yet) translated into employment. Labor force participation was likely also helped by robust wage gains, which averaged about 3 percent annually since 2005. Reflecting these dynamics, a divide has emerged in the Swedish labor market, with healthy, and declining, unemployment among high-skilled, Swedish-born, prime-age workers, and increasing unemployment rates among other, more vulnerable groups.


Labor Force and Immigration


Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: World Bank WDI and Fund staff calculations.

Labor Force Participation Rates, 2002Q1:2013Q3

(Percent of population subgroups)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: Eurostat and Fund staff calculations.

B. The Swedish Labor Market Model

5. Recent developments are linked to the uniqueness of Swedish labor market institutions. Social insurance and redistribution are important pillars of Sweden’s labor market model, distinguishing it from the Anglo-Saxon model with relatively lower social protection. But Swedish institutions also differ in important ways from the flexicurity model, the approach often associated with Nordic labor markets. In its essence, flexicurity (“protect workers, not jobs”) aims to combine flexible labor reallocation (via low employment protection (EPL) and active labor market measures (ALMP)) with generous insurance of workers against labor market risk (e.g., via unemployment insurance (UI) systems). In practice, none of the Nordic countries fully adhere to this model, but Sweden has diverged from it somewhat more so than others.


Labor Market Institutions

(Rate, X-axis; Index:0-6 (6 indicates stricter regulation), Y-axis)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: OECD and Fund staff calculations.

6. Key aspects of the Swedish labor market policies and institutions are as follows:

  • Unemployment insurance (UI) is generous in Sweden, especially over longer unemployment spells. Based on OECD data, the net unemployment benefit replacement rate for a representative household was about 60 percent in 2012, somewhat below the median of about 70 percent in most other European economies. However, Swedish UI is substantially more generous when viewed over longer horizons—averaged over 60 months of unemployment, the net replacement in Sweden is 44 percent, compared to the EU average of 34 percent. Generous UI systems cushion the financial impact of a job loss and can help lead to a more productive next job by providing the unemployed with the resources to search more exhaustively. But more generous UI systems also reduce individual’s incentives to accept job offers at wages that might otherwise be acceptable, thereby lengthening unemployment spells. A relative slow UI benefit decline as in Sweden matters more the higher the fraction of long-term unemployment. The increase in the duration of unemployment in recent years is especially concerning in this context.

  • Active Labor Market Programs (ALMPs) have been an integral part of Sweden’s labor market set-up and encompass a large numbers of programs, targeted at various subgroups. That said, in 2011, ALMP expenditure in Sweden accounted for 1.7 percent of GDP, slightly above the OECD average, but substantially below the levels in some of its Nordic peers (e.g., 2½ percent of GDP in Finland and nearly 4 percent in Denmark). At less than 1/10 percent of GDP, spending on job training has been particularly low, amounting to about ½ the expenditure in Norway and 1/5 of that in Denmark and Finland.

  • Employment protection of regular contracts in Sweden is among the highest in Europe, and exceeds the levels in all Nordic peers, while lay-off restrictions on temporary contracts are among the lowest. Thus, Sweden has one of the most asymmetric employment protection legislations in Europe, a feature that tends to be associated with increased labor market duality (defined as the share of employees in non-permanent contracts) in cross-country comparison. High job protection often gives rise to insider-outsider dynamics, as the protected have little reason to adjust wages.


Labor Market Program Expenditures

(Percent of GDP)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: OECD and Fund staff calculations.

Average duration of unemployment

(Number of weeks, SA)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: Riksbank and Fund staff calculations.

Temporary Employment Contracts, 2013

(Percent of total employment)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: OECD and Fund staff calculations.

7. Wage adjustment is shaped by broad union coverage. Collective bargaining is a key pillar of wage determination in Sweden. Both membership (union density) and coverage (the share of individuals covered by union agreements independent of membership) are high: union density was about 70 percent of workers in 2012, on par with Nordic peers Denmark and Finland, but exceeding unionization levels in the rest the EU. Union coverage was even higher at over 90 percent (in 2011), surpassed only by Austria and Belgium.

8. The collective bargaining process is highly coordinated. Coming out of the crisis of the early 1990s, unions, employers and the government agreed on the need for increased cooperation in incomes policies (Thomas, 1998). Partly as a result, while collective bargaining in Sweden today takes place at the sectoral level, it is highly coordinated across industries, and with the national industry agreement setting the example for the remaining labor market. The national agreements regularly set the floor on wage increases at the firm/local level. Pay agreements typically take place every three years, with the latest agreement in 2013 specifying a cumulative 6.8 percent salary increase over a 3-year period for industrial workers. While the Swedish system shares similarities with the German bargaining framework (Box 3), it provides for less scope at the firm-level to adjust wages to local conditions, with implications for wage outcomes.


Wage and Productivity Growth

(Percent change)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: Haver Analytics and Fund staff calculations.

Union Coverage

(Percent of employees)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: OECD and Fund staff calculations.

9. On average, the Swedish bargaining framework has led to wage growth broadly in line with productivity. Following double-digit nominal wage increases during the 1980s, reforms to Sweden’s collective bargaining framework have helped moderate (manufacturing) wage growth to about 3 percent annually on average since 2000. Strong average labor productivity growth of nearly 4 percent annually has resulted in declining unit labor costs, supporting also Sweden’s external competitiveness.


Wage Dispersion

(Ratio of 5th to 1st gross earnings decile, 2011)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: OECD and Fund staff calculations.

Collectively Agreed Wage Floors, 2011

(Percent of average wage)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: Swedish Trade Union Confederation (LO) and Fund staff calculations.

10. But wage stability may have come at the cost of more limited wage flexibility. Stable and moderate wage growth has come along with a highly compressed wage distribution: the difference between the median wage and the bottom 10th percentile is the second-lowest in the OECD.3 The high degree of wage compression reflects at least in part high collectively bargained wage floors—these amounted to over 80 percent of actual average wages, suggesting that wage floors were in many cases binding. As a point of comparison, collectively agreed wage floors in Germany were in most sectors closer to about ½ of average hourly earnings.4,5

11. Overall, Swedish labor market institutions differ in important aspects from those in other countries. They are substantially different from the Anglo-Saxon model, as represented by the USA, but also differ in important ways from an idealized flexicurity model, perhaps best represented by the current Danish set-up. The question is to which degree this uniqueness comes at a price. To contribute to this discussion, the following section aims to quantify the changes in labor market outcomes that could be expected from adjustments to the the Swedish labor market setup along some, or all, dimensions discussed above—keeping in mind that there is no absolute benchmark, and that the “optimal” labor market model will depend on societal characteristics and preferences.


Labor Market Institutions Across Countries

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: European Commission, OECD and Fund staff calculations.Notes: Larger distances from the center are associated with higher unemployment rates. The upper limit is scaled by the maximum values observed in the OECD in 2011 (inverted scale for ALMP per unemployed).

C. A Quantitative Perspective

12. Which labor market institutions matter most for labor market outcomes? A call for policy actions requires a deeper understanding of which factors actually impact labor market outcomes, and in what magnitudes. To examine this, a panel, fixed effects model is estimated, with the baseline regression taking the following form:


where the dependent variable is the (total) unemployment rate in country i at time t, regressed on a set of labor market institutions (indexed by j), inflation, the output gap, and a country-fixed effect. The specification is intentionally kept parsimonious, focusing on steady-state correlations between institutions and outcomes, with inflation and the output gap included to control for cyclical factors and country fixed effects to capture country-invariant characteristics.

13. The empirical analysis reveals the complexities of designing labor market institutions (Table 1). Column (1) shows a basic regression controling for the roles that inflation (along the lines of the Philips curve concept) and cyclical dynamics (Okun’s Law) play in determining the unemployment rate. As the discussion above indicated, while the estimation highlights that cyclical factors matter in general, they explain only some of the increase in Swedish unemployment since the crisis. This points to the importance of structural factors in the Swedish labor market in recent years. The remaining specifications explore the effects of key labor market institutions in addition to cyclical factors.6 In particular:

  • Column (2) indicates that the impact of ALMP7 is statistically significant and in the expected direction, with greater ALMP expenditure per unemployed associated with lower unemployment. At the same time, column (5) suggests that the impact of ALMP expenditure is not evenly distributed across programs, with programs providing training showing a relatively higher impact than others.

  • While the coefficient for EPL is negative and not statistically significant;8 column (3) shows that it is EPL on regular contracts (EPLR) that has appears to matter most: higher EPLR has a large impact on increasing unemployment, while EPL on temporary contracts (EPLT) appears to matter less (for aggregage unemployment). However, column (4) shows that while the level of EPLT might not matter, large differences between EPLR and EPLT do.

  • The degree of unionization affects unemployment: as predicted by theory, higher unionization rates are associated with higher unemployment, as shown in columns (6) and (7). This association can best be understood in the context of the wage channel—as Table 3 indicates, more unionized labor markets are associated with both higher wage growth and stronger wage compression, both of which can impede job creation especially for low-skill, low-productive jobs.

  • The UI replacement rate is positively and significantly associated with higher unemployment, reflecting the well-known fact that more generous UI systems provide incentives for individuals to be more selective about job offers and search for longer.

Table 1.

Total Unemployment and Labor Market Institutions

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

14. Youth unemployment has similar determinants, with subtle differences (Table 2). First, EPL does not have a statistically significant impact. A possible explanation is that EPLT may be more relevant for youth, who are more likely to be in temporary work arrangements, where EPL appears to generally be less effective (Table 1). Similarly, UI benefits are not statistically significant, which is plausible given that many youth may not be eligible for UI benefits. By contrast, the impact of ALMP remains important, and the difference between general expenditure and those on training is even more pronounced. So is the impact of unionization, which can be understood in the context of the insider-outsider conflict, and the costs of unions disproportionately falling on the youth (who, on average, are “outsiders”).9

Table 2.

Youth Unemployment and Labor Market Institutions

article image
Standard errors in parentheses*** p<0.01, ** p<;0.05, * p<;0.1
Table 3.

Wages and Collective Bargaining

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

15. The impact of union coverage on labor market outcomes seems to be linked to their role in shifting and compressing the wage distribution (Table 3).10 Column (1) in Table 3 confirms the expected determinants of wage growth, including productivity growth and inflation as well as the cyclical pressures.11 However, unions impart an additional wage premium, in line with the theoretical predictions of standard union models. But beyond raising the average wage, unions are also associated with more compressed wage distributions: the spread between the median wage and the bottom 10th percentile is significantly narrower the higher unionization (column (3) shows the analogous result when considering the 90th percentile rather than the median). Both effects are important: higher average wages imply a lower overall labor demand, and a more compressed wage affects those groups where wage flexibility is most important—at the bottom end of the wage distribution, where most of the unemployment (low-skilled, young, immigrants) is concentrated.

D. Policy Implications

16. Overcoming Sweden’s labor market challenges will require comprehensive reform. The Swedish labor market has performed well after significant reforms were implemented in the early 1990s, with the unemployment rate declining from nearly 10 percent in the mid-1990s to about 5½ percent in 2000. However, since then, unemployment has trended upwards again despite additional reforms since 2000. The structural deterioration in labor market outcomes reflects in part important changes in the labor market environment: significant immigrant inflows (often with initially insufficient skills) and (globally) increased need for skilled labor has left current labor market institutions unable to integrate the added labor supply into productive employment.12 While the Swedish system continues to do well in terms of achieving social equity objectives, it has coped less well with the need to match a changing labor force with the need for increasingly high-skilled employment. This points to reform needs along a number of dimensions.

17. On the labor demand side, a number of measures will be key, including securing wage flexibility at the firm level and for the most vulnerable, a reformed EPL, and an improved business environment.

  • Wage flexibility. Limited wage adjustment at the low end of the wage distribution holds back employment growth among the most vulnerable groups, including the young and the low-skilled. It is important that the collective bargaining framework allows firms and workers to find wage agreements that reflect overall macro and (un)employment conditions as well as firm-specific factors. Increased wage flexibility in collective agreements (proxied by union coverage in the regressions), for example, by introducing additional scope for firms to deviate from national agreements under certain conditions, could lower unemployment by more than ½ percentage point if the USA model was adopted (Table 4). In this context, while UI generosity is high in Sweden, which likely contributes to elevated reservation wages, wage flexibility seems to be the more crucial factor—with collectively bargained wage floors at 80 percent of average wages, they are likely the binding constraints on job creation.

  • Employment protection. Overall, EPL is high in Sweden. Although it has come down substantially since the mid-1980s, this was driven primarily by the deregulation of temporary contracts. EPL on permanent contracts has remained high and associated with a lack of labor demand.13 Reduced EPL would also facilitate labor reallocation and enhance productivity growth which has slowed down to pre-1990 levels, from near double-digits during much of the 1990s and 2000s. Reducing employment protection on regular contracts, especially via shorter advance notice requirements, could lower unemployment by between ½ (Denmark, OECD average) to 2½ (USA) percentage points, depending on the benchmark.

Table 4.

Potential Gains from Adjusting the Swedish Labor Market Model

article image
Source: Fund staff calculations.

Sweden: Employment Protection Legislation (EPL)


Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: OECD and Fund staff calculations.

18. Measures to strengthen labor market matching would also help. Key among them are ALMPs, which are generally viewed as helpful for improving labor market matching—a result reflected in the empirical results discussed above. Consequently, an increase in ALMP expenditures closer to, for example, the Danish level could lower unemployment by ½ percent. But the composition of ALMP spending also matters. As shown earlier, Sweden lags behind Denmark especially in terms of training expenditures and job matching services (public employment services), while having an elevated level of expenditures on employment incentives. Wage subsidies are costly and have ambiguous effectiveness, as short-term subsidies often do not lead into long-term employment, and because they often do not generate additional employment.14 By contrast, training programs can have a longer-lasting impact by providing the unemployed with basic work skills. A refocusing of ALMPs on those areas most cost effective can help both labor market efficiency and rationalize expenditures. But adjusting the scheme of UI benefits should also be considered: while initial UI levels are in line with other countries, the long benefit eligibility exacerbates the decline in matching efficiency and the recent increase in unemployment duration. A steeper decline in UI benefits would provide additional incentives for accepting jobs—however, this is unlikely to be helpful as long as low-wage flexibility is not improved.

19. Importantly, reforming Sweden’s labor market does not mean compromising on its basic tenets. Moving Sweden’s labor market institutions to the “best frontier” level in each of its dimensions would have an estimated unemployment-reducing effect of over 6 percent, but this may not be desirable.15 Swedish labor market institutions are built on a careful balance of efficiency and equity. A reform package appropriate for Sweden should not abolish its basic tenets, including fostering social dialogue, helping those in need, and ensuring fair incomes. But fine-tuning institutions along key dimensions can have a large impact and help Sweden’s labor market meet the challenges of the future. In particular, adding more scope for wage flexibility into Sweden’s well-functioning collective bargaining framework, tilting ALMP expenditures to more effective and efficient programs, and allowing for easier labor reallocation through lower EPL could lower structural unemployment substantially while maintaining Sweden’s model.

20. Measures outside the labor market are also important to improve (un)employment outcomes.

  • Housing supply restrictions contribute to rising house prices, which have created financial stability risks but also inhibit labor reallocation by making it costly to move into metropolitan areas with higher labor demand. Addressing housing market inefficiencies would therefore help address both financial sector and labor market challenges.

  • Business entry requirements. New business formation is an important driver of labor

    demand. While Sweden still performs satisfactorily overall, its ranking on the World Bank’s 2014 Doing Business analysis dropped to 61 from 55 in 2013 (driven in part by the subcategories “starting a business” and “getting (corporate) credit”). The deterioration reflects largely an improvement in comparator countries rather than an absolute decline in Sweden, however, in an increasingly globalized world, falling behind can mean losing out on, for example, FDI inflows.16

  • Vocational training. The usefulness of ALMP-training measures notwithstanding, they cannot substitute for an educational system, including vocational training. Ongoing efforts by the authorities are to be commended, but a fundamental vocational system, for example, along the lines of the German system, would help provide the Swedish labor force with needed skills and support needed productivity growth.17


Sweden: Time to Start a Business


Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: World Bank Doing Business 2014 and Fund staff calculations.

Cyclical versus Structural Factors in the Swedish Labor Market

Changes in the Swedish labor market reflect both cyclical and structural factors. Starting around 2006, the Swedish authorities initiated a broad set of labor market reforms (see Box2). These measures increased effective take-home pay for many groups, contributing to an expansion of the labor force as the rewards for job search increased and the administrative incentives for labor force participation were strengthened. In the context of a simple demand-supply framework, this corresponds to a shift in the labor supply curve. While a robust recovery from the crisis brought about an expansion of labor demand, overall employment growth ultimately fell short of that in the labor force and led to a substantial increase in unemployment. As the model illustrates, a lack of wage adjustment likely played an important role in this outcome: with supply growth exceeding that in demand, reaching the new equilibrium (E2) requires a downward adjustment in the wage (to w1). However, if the wage level remains close to the prior wage (w), employment growth will be limited (at E1). Low inflation in Sweden has meant that real wages did not adjust either. Over time, if real wages adjust—through nominal wage adjustment, inflation, or both—and if labor demand recovers further, employment increase. Whether the improvement would translate into a reduction of unemployment to pre-crisis levels depends on the wage setting process and, more broadly, on the evolution of structural employment.


Average duration of unemployment

(Number of weeks, SA, left; Percent, right)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Other models suggest that structural unemployment has increased in recent years. Search-theoretic models view unemployment as a function of inflows (job destruction) and outflows (job creation), with outflow rates inversely related to the average length of unemployment. A marked increase in the average unemployment duration since about 2008 suggests lower job finding rates in Sweden. A search-based model parameterized in part based on unemployment duration data suggests that unemployment has increased permanently, that is, structural has increased from about 6½ percent during the pre-crisis years to over 7½ percent since the crisis.


Swedish Unemployment: Actual and Simulated


Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A003

Sources: Statistics Sweden and Fund staff calculations.Note: The simulations are based on a simple flow-model of unemployment.. The model with an unchanged job finding rate is parameterized based on the pre-crisis average, while the alternative simulation adjust the job finding rate based on average post-crisis unemployment duration

Major Labor Market Reforms over the Past Two Decades

Sweden has undertaken substantial structural reforms over the past two decades. A first wave of reforms took place after the banking crisis in the early 1990s and included both product and labor market reforms.1 Major milestones related to labor markets included2

  • Tax reforms to shift the tax burden from labor income to consumption and capital income through lower marginal tax rates on earned income and other measures;

  • Wage restraint through an agreement by industrial unions to restrain wage increases; more coordination in wage bargaining; and re-establishing the leadership role of those sectors most exposed to international competition; and

  • Reform of labor market institutions, including a reorientation of active labor market policies toward training programs; reduced employment protection; and reforms to social insurance (including lower UI replacement rates).

A second wave of labor market reforms started with the incoming government in 2006. A strategy for labor market reforms aimed at addressing unemployment and poverty traps was implemented in steps over the following years and often targeted at non-prime age workers (the very young and the very old). Key measures involved reducing the tax burden for low-wage jobs through income tax deductions, a greater focus of ALMPs on training and job matching, and reductions in unemployment and sickness benefits. Others included providing wage subsidies to employers hiring vulnerable groups (New Start Jobs), immigrants (Step-In Jobs), or the previously sick (Well Again Jobs). More recently, as part of the Budget Bill for 2014, the authorities proposed measures to further reduce the low-wage tax burden, through an increase in the earned income tax credit. The budget bill also included plans to improve vocational training in Sweden (e.g., through wage subsidies to support “vocational introduction agreements” between social partners in a number of occupational sectors and for employers who offer apprenticeship training) and to reduce youth unemployment. The bill also included measures to provide resources for a short-time work program, intended for use in a particularly deep economic crisis.

1 Major product market reforms in Sweden included deregulation in a number of sectors, including especially the telecommunications sector, and earlier (in the 1980s) measures to promote competition. These reforms are seen to have fostered rapid restructuring and created large productivity gains in the export sector.2 See also Box 3.3 in the IMF’s EUR REO October 2011.

Collective Bargaining and Economic Outcomes

The relationship between collective bargaining and economic outcomes is theoretically complex. Early models, such as Dunlop’s (1944) “monopoly union” model or Nickell and Andrews’ (1983) “right-to-manage” model, found that greater union bargaining power results in higher wages and lower employment, with unions effectively choosing a point on the downward-sloping labor demand curve. Insider-outsider conflicts (e.g., Lindbeck and Snower, 1988) often also lead to higher wages and lower employment, as insiders exploit their position at the expense of outsiders. Richer models, where unions negotiate over variables other than just the wage (such as the level of employment), can change these findings. In addition to affecting the average wage, models with worker heterogeneity (e.g., differences in skill levels) typically find that collective bargaining reduces the spread of wages compared to the case where workers are paid based on marginal productivity (e.g., Cahuc and Zylberberg, 2004, Alvarez and Shimer, 2011). The empirical literature tends to supports the main findings in the theoretical literature. A large empirical literature has found substantial union wage premia, with estimates ranging from 8 percent in the UK (Booth, 1995) to 15 percent in the US (Lewis, 1986; Blanchflower and Bryson, 2002). More unionized labor markets also appear to be associated with more compressed wage distributions, with, for example, Kahn (1998, 2000) finding lower wage inequality in OECD economies with larger collective bargaining coverage.

The degree of centralization and coordination of bargaining also matters. Collective bargaining can take place at many levels: the firm, the sectoral (or industry), or the national level. Calmfors and Driffill (1988) famously hypothesized a hump-shaped relation between the degree of centralization of bargaining and the unemployment rate, arguing that in centralized regimes, trade unions internalize the adverse impact of higher wages in inflation and unemployment, while in decentralized regimes, limited market power by firms constrains wage increases. However, at intermediate levels of bargaining (industry level), both market forces and wage internalization effects may be too weak to restrain wage increases, thus leading to relatively higher wage outcomes and lower production and employment. Empirical evidence on this relationship is mixed. While some authors have subsequently found some empirical support (e.g., Scarpetta, 1996, and Elmeskov et al., 1998), the majority of empirical studies have not (Flanagan, 1999). Measurement problems are likely one of the main reasons for the mixed empirical findings. Similarly classified countries may differ in crucial details. For example, while collective bargaining in both Germany and Sweden takes place mostly at the sectoral level and is highly coordinated, with certain sectors acting as “pacesetters” for other sectors, firm-level agreements and opening clauses in Germany impart a large degree of flexibility that is absent in Sweden.


  • Alvarez, Fernando, and Robert Shimer, 2011, “Search and Rest Unemployment,Econometrica Vol. 79, pp. 75122.

  • Anderson, Derek, Bergljot Barkbu, Lusine Lusinyan, and Dirk Muir Andersen, 2014, “Assessing the Gains from Structural Reforms for Jobs and Growth,” in Jobs and Growth: Supporting the European Recovery, Chapter 7, Schinder, Bakker, Berger, Spilimbergo (eds) Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Blanchflower, David, and Alex Bryson, 2002, “Changes over Time in Union Relative Wage Effects in the UK and the US Revisited,NBER Working Paper 9395.

    • Search Google Scholar
    • Export Citation
  • Booth, A, 1995, The Economics of the Trade Union (Cambridge, UK: Cambridge University Press).

  • Cahuc, Pierre, and André Zylberberg, 2004, Labor Economics (Cambridge: MIT).

  • Calmfors, Lars, and John Driffill, 1988, “Bargaining Structure, Corporatism and Macroeconomic Performance,Economic Policy, Vol. 3(6), pp. 1461.

    • Search Google Scholar
    • Export Citation
  • Dunlop, J., 1944, Wage Determination Under Trade Unions (New York: Macmillan).

  • Elmeskov, J., J. Martin, and Stefano Scarpetta, 1998, “Key Lessons for Labour Market Reforms: Evidence from OECD Countries’ Experience,Economic Council of Sweden, Stockholm.

    • Search Google Scholar
    • Export Citation
  • Espinoza, Raphael, and Esther Pérez Ruiz (forthcoming), “Tax Wedge Cuts and Employment: A General Equilibrium Approach for France,IMF Working Paper (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • European Commission, 2013, “Labour Market Developments in Europe 2013,European Economy Vol. 6, 2013.

  • Flanagan, R.J. (1999), “Macroeconomic Performance and Collective Bargaining: An International Perspective, Journal of Economic Literature 37(3), 115075.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund (IMF), 2014, Euro Area Policies: Selected Issues, IMF Country Report 14/199 (Washington).

  • Jimeno, Juan F., and Carlos Thomas, 2013, “Collective Bargaining, Firm Heterogeneity and Unemployment,European Economic Review, Vol. 59, pp. 6379.

    • Search Google Scholar
    • Export Citation
  • Kahn, Lawrence, 1998, “Collective Bargaining and the Interindustry Wage Structure,Economica, Vol. 65, pp. 507534.

  • Kahn, Lawrence, 2000, “ Wage Inequality, Collective Bargaining and Relative Employment from 1985 to 1994: Evidence from Fifteen OECD Countries,Review of Economics and Statistics, Vol. 82, pp. 564579.

    • Search Google Scholar
    • Export Citation
  • Lewis, H., 1986, “Union Relative Wage Effects,” in Handbook of Labor Economics, Ashenfelter, O., and Layard, R. (eds.), Vol. 2 (Amsterdam: Elsevier Science/North-Holland).

    • Search Google Scholar
    • Export Citation
  • Lindbeck, Assar, and Dennis J. Snower, 1988, The Insider-Outsider Theory of Employment and Unemployment (MIT Press, Cambridge, Massachusetts).

    • Search Google Scholar
    • Export Citation
  • Ljungqvist, Lars, and Thomas J. Sargent, 1998, “The European Unemployment Dilemma,Journal of Political Economy, Vol. 106, pp. 514550.

    • Search Google Scholar
    • Export Citation
  • Nickell, Stephen J., and Martyn Andrews, 1983, “Unions, Real Wages and Employment in Britain 1951-79,Oxford Economic Papers, Vol. 35, pp. 183206.

    • Search Google Scholar
    • Export Citation
  • Scarpetta, Stefano, 1996, “Assessing the Role of Labour Market Policies and Institutional Settings on Unemployment: A Cross-Country Study,OECD Economic Studies, Vol. 26(1), pp. 4398.

    • Search Google Scholar
    • Export Citation
  • Schindler, Martin, Bas B. Bakker, Helge Berger, and Antonio Spilimbergo (eds.), 2014, Jobs and Growth: Supporting the European Recovery (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Thomas, Alun, 1998, “The Wage Bargaining Structure in Norway and Sweden and its Influence on Real Wage Developments,IMF Working Paper 98/74 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • World Bank, 2014, Doing Business (Washington: World Bank Group).


Prepared by Martin Schindler.


See the Selected Issues chapter on “Potential Output and Unemployment” for the 2013 Article IV for a discussion.


Sweden ranks at or near the bottom also if the ratio of 90th to 10th percentile is used or inequality across family incomes is used (see CIA World Factbook, where Sweden ranks as the most equal society).


While a detailed, sector-by-sector comparison in Germany is not available, as of April 2014, branch-specific, collectively bargained hourly wages in Germany averaged €10.5/hour, compared to average hourly wages of about €20 in the 4th quarter of 2013. See


As the analysis below indicates, unemployment outcomes and unions are linked via the wage channel.


While this set-up is simple and serves to illustrate equilibrium relationships between unemployment and the various policy variables without accounting for dynamics or causal effects, the estimated orders of magnitude are consistent with those found in other contexts. For example, in a DSGE model calibrated to France, Espinoza and Pérez Ruiz (forthcoming) find that a 1 ppt. tax wedge reduction would raise employment between 0.1 (short run) and 2/3 (long run) percent, compared to about a 0.4 percent employment increase implied by the tax wedge coefficient in column 7. Also based on a model simulation, Anderson et al. (2014) find that the average country would experience long-run employment effects of about 3 percentage points from labor market reforms that close the gap with a set of benchmark economies.


Expenditure on ALMPs is measured as total expenditure per unemployed individual, which helps to control for the possibility that expenditures as a percent of GDP increase when unemployment rises, without necessarily a change in spending per individual, which is what matters most.


Theoretically, the effect of firing restrictions is ambiguous: EPL makes job destruction more costly, reducing the flow into unemployment, but it also raises firms’ cost of creating a vacancy (by reducing the option value of a job), thus reducing the flow out of unemployment.


See also IMF (2014).


In an alternative specification, the tax wedge was also included, but was not statistically significant, consistent with the notion that the tax wedge would affect the level of wages, but not their growth rate or dispersion. To maximize sample size, it is excluded from the estimates reported here.


Including the unemployment rate instead of the output gap leads to similar results.


A large literature following Ljungqvist and Sargent (1998) has shown how the performance of labor market institutions can vary with changes in the economic environment.


Advance notice requirements for redundancies are among the highest in the world, especially for workers with more than five years of tenure. Sweden also has burdensome redundancy rules, including certain retraining and/or reassignment obligations before redundancy dismissals can take place, as well as priority rules for redundancy dismissals, lay-offs and re-employment. See


Net job creation associated with wage subsidies is often small, as employers often switch existing hiring plans into subsidized areas, rather than creating new positions. There is also evidence that some of the Swedish programs (New Start) have little to no lasting employment impact on the most vulnerable groups they are intended for.


Such calculations must be qualified: the estimation methodology does not account for interactions, and it is likely that the cumulative effect is not equal to the sum of the individual impact estimates.


See, e.g., proposals in the OECD’s 2011 Reviews of Vocational Education and Training—Learning for Jobs and the Swedish government’s policy plans announced in the 2014 Fall Budget Bill.

Sweden: Selected Issues
Author: International Monetary Fund. European Dept.