THE SWEDISH ECONOMY is currently in its worst postwar recession. Real GDP declined by over 8 percent between the peak in the first quarter of 1990 and the trough in the second quarter of 1993. Sweden’s historic experience of maintaining one of the lowest rates of unemployment among the industrial countries has disappeared during this recession. The unemployment rate (which includes both “open” unemployment and participation in labor market programs) has increased from about 3½ percent of the labor force in 1990 to almost 13 percent of the labor force in 1993. Sweden’s public finances have also deteriorated dramatically in recent years, with the central government’s overall borrowing requirement rising to over 18 percent of GDP in 1993/94.
The current recession was precipitated by the large credit‐led expansion of both consumption and investment in the second half of the 1980s, which eventually paved the way for financial consolidation, resulting in a sharp decline in domestic demand beginning in the second quarter of 1990. However, in addition to the bursting of the asset‐price bubble, the current crisis also has its origins in the underlying structural problems of the Swedish economy—in particular, the long‐term productivity slowdown and high wage inflation that have characterized its recent history. Figure 1 shows that the growth of labor productivity in manufacturing in Sweden has been, in general, slower than the average for the European Union since 1976. This has resulted in an increasing productivity gap with respect to the European Union and a relative decline in Swedish living standards. Although Sweden’s private per capita consumption ranked fifth among the OECD countries in 1970, it had slipped to twelfth by 1990.1 The Swedish economy also suffered from problems of external competitiveness in the 1980s. Following a period of rapid growth of nominal wages and slow growth of productivity, Sweden’s relative unit labor costs rose despite the large devaluations in 1981–82, which resulted only in a transient improvement in competitiveness. This decline in long‐term competitiveness is reflected by a rising import‐penetration ratio and falling export market shares in the 1980s (see Figure 2).
The hypothesis advanced in this paper is that the long‐term problems of slow productivity growth and high wage inflation, although affected by factors that are complex and multifaceted, can be linked in an important way to specific features of the Swedish labor market—in particular, to the way in which the wage bargaining system evolved over time. The following sections of this paper outline the main institutional features of the Swedish labor market and analyze the reasons for the long‐term decline in competitiveness.
I. Institutional Framework of the Labor Market
The so‐called Swedish model, usually identified with an advanced welfare state, has attracted attention from many quarters for its apparent earlier success. One of the distinctive features of the Swedish model has been its unique labor market institution, which combines centralized bargaining with a policy of wage equalization, designed to promote a favorable macroeconomic performance.2
Figure 1.Output and Productivity Trends
Sources:IMF, World Economic Outlook; SCB, National Accounts.
Figure 2.Competitiveness, Import Penetration, and Export Market Share
Sources: IMF, World Economic Outlook; IMF staff calculations.
The main institutional features of the Swedish labor market are as follows. Wage negotiations in Sweden have been conducted since the 1930s between centralized trade unions and a centralized employers’ organization. There are effectively two centralized unions in Sweden— one represents all blue‐collar workers (LO) and the other represents all white‐collar workers (TCO). Since almost 80 percent of Swedish workers are unionized, these centralized unions are truly encompassing and determine the wages for almost all workers. The employers are represented by the Swedish Employers Confederation (SAF). Although wage negotiations are conducted at the central level, the system also allows for “wage drift,” whereby the centrally negotiated wages are topped up at the local level. There is some debate in the Swedish literature about whether the wage drift is independent of the central negotiations or is already accounted for in deciding the centrally determined wage. Wage drift has certainly increased considerably in the second half of the 1980s, signaling a substantial loosening of the centralized bargaining system.
Whereas some form of centralized wage bargaining is common to all the Nordic countries, the Swedish trade unions have been unique in pursuing an explicit policy of “solidaristic wages,” or wage equalization, as an important objective of the bargaining process. The concept of solidaristic wages was initially conceived as equal pay for equal work. However, in the 1970s, the unions pursued an aggressive policy of equalizing wages across both firms and industries, with little attention to the nature of the work performed. Consequently, by the early 1980s, Swedish wage differentials were substantially lower than in other OECD countries. Estimates provided by the Swedish Ministry of Finance indicate that the wage spread for industrial workers, calculated as the difference between the highest and lowest deciles in 1984, was 34 percent for Sweden in contrast to 210 percent for the United Kingdom and 490 percent for the United States. By 1990, the wage spread for Sweden had increased somewhat, to 45 percent, owing to a weakening of centralized bargaining.
The Rehn‐Meidner Model: Solidaristic Wages and Structural Change
The combination of centralized bargaining and wage equalization in the labor market constituted the so‐called Rehn‐Meidner model’s strategy of promoting structural change and growth. This model derives its name from the two trade union economists, Gösta Rehn and Rudolf Meidner, who were instrumental in outlining the economic arguments for the Swedish model in the early 1960s. In terms of the framework set out in this model, the concept of solidaristic wages has two distinctive effects. First, it imposes a high wage cost on low‐technology sectors and inefficient firms by not permitting them to lower wages. Hence, under this wage bargaining system, the less productive sectors either have to rationalize and become more efficient or have to go out of operation. In contrast, workers in the high‐technology sectors are not given the freedom to negotiate higher wages in spite of their relatively higher productivity, allowing the high‐technology enterprises to generate relatively higher rates of profit that can be invested for faster growth. Displaced workers from the low‐technology sectors provide the labor supply for the expanding high‐technology sectors. Centralized retraining schemes aid the workers in shifting from the low‐technology to the high‐technology sectors. Over time, this policy of penalizing the less productive sectors and providing incentives for the more productive sectors increases the share of dynamic, high‐technology enterprises in total production. Since Sweden has traditionally been an economy with high average wages and labor shortages, the wage equalization policy was conceived as advantageous for rapidly transforming the technological basis of the economy and promoting high growth without hitting the labor supply constraint.
A crucial assumption behind the Rehn‐Meidner model is that this forcible retrenchment and retraining of workers ensures a faster process of structural change and growth than in a system in which workers respond to wage differentials in moving from the low‐ to the high‐technology sectors and in acquiring skills. The validity of this assumption is discussed below. The Rehn‐Meidner model is essentially supply‐side oriented. The state is discouraged from intervening with demand‐management policies to absorb retrenched workers. Instead, the state is supposed to operate a tight monetary and fiscal policy, with the emphasis on supply‐side measures such as providing retraining schemes and matching skills with vacancies.
A brief discussion of the Swedish labor market policies is called for in this context. Swedish labor market policies are separated into “active” and “passive” components. The active measures consist of centrally provided training schemes, relief work, subsidized employment, and special measures for youth training. The passive measures are unemployment compensation and early retirement pensions. The main theme of Swedish labor market policies has been to discourage dependency on unemployment benefits and instead foster an environment in which workers are induced to be mobile between occupations through retraining. It has been estimated that in 1985–90, active measures constituted about 57 percent of the total expenditure on labor market policies. This contrasts sharply with a figure of 28 percent for Western Europe. Training programs in Sweden offer the equivalent of unemployment compensation for participants. In contrast, workers on relief work are paid market wages.
The EFO Model: Traded Goods Sector and Wage Determination
The so‐called EFO model constitutes the other main component of the Swedish model. It derives its name from the initials of the three economists of the white‐collar trade union, the employers’ organization, and the blue‐collar trade union—Edgren, Faxen, and Odhner—who were instrumental in outlining the consensus view on wage bargaining between the Swedish employers and workers.3 Whereas the emphasis of the Rehn-Meidner model was on promoting structural change and growth, the EFO model was concerned with preserving the competitiveness of the Swedish economy by controlling the rate of growth of wages. Together, these two models sought to provide the framework in which policies for both growth and competitiveness could be pursued simultaneously.
The main thrust of the EFO model was to maintain the international competitiveness of Swedish industry. Hence, in the competitive sector, or the sector open to foreign trade, the wage increase was to be determined by the sum of international price inflation and the rate of growth of labor productivity in this sector. This method of determining wage increases would maintain Swedish tradable goods prices on a par with those of its competitors. However, the policy of wage equalization adopted by the trade unions simultaneously dictated that an equal wage increase had to be given to the nontraded sector. Since the rate of growth of productivity is relatively lower in nontradables, the wage‐setting process implied by the EFO model built in an inflationary process in the nontradables sector. In theoretical terms, inflation in the nontradable goods sector does not constitute an immediate threat to the competitiveness of the traded goods sector. Nevertheless, over the long run, high rates of inflation in the sheltered sector have the potential to erode the competitiveness of the traded goods sector through the repercussion effects of wage bargaining.
II. Problems of the Swedish Model
Although the Swedish economy performed rather well until the mid–1970s, serious competitiveness problems emerged thereafter. Productivity growth slowed, and wage inflation averaged more than 8 percent a year in the 1980s. While Sweden managed to avoid high unemployment through a large devaluation in 1981–82 and a substantial expansion of the public sector thereafter, its external competitiveness suffered and Sweden’s per capita GNP dropped to the OECD average in 1990 after being 10 percent above average in 1970.4
High Wage Inflation
Sweden’s problems with high wage inflation, especially in the 1980s, concern the way in which centralized negotiations came to be conducted in practice. In theory, a centralized wage bargaining system, by internalizing various externalities of decentralized bargaining, is expected to achieve relatively low wage inflation. For instance, the centralized union is unlikely to push for a wage increase that leads to excessive unemployment; unlike a decentralized union, it cannot pass on the burden of funding unemployment benefits to other unions. Also, it is difficult for a centralized union to secure real wage increases simply by obtaining nominal wage increases, whereas a decentralized union, operating in isolation, can do so.5 Hence, in general, there is less incentive for centralized unions to demand excessively high nominal wages. Sweden was unable to tap these potential benefits of centralized bargaining, especially in the 1980s, because of (1) the multilevel bargaining structure in Sweden in which wage drift accounted for a significant part of the overall settlements; (2) the breakdown of the EFO model in the 1980s; and (3) the impact of labor market programs under very low unemployment. Each of these factors is analyzed below. The analysis is primarily qualitative, and no attempt is made to quantify the importance of the different factors in causing wage inflation.
First, Sweden’s multitiered system of wage bargaining has proved to be a constraint on controlling wage inflation.6 As pointed out earlier, centrally negotiated wages in Sweden are subject to further increases at the local level (wage drift). Anticipating the subsequent wage drift, the central union usually decides on the initial nominal wage increase that is compatible with a given final real wage target. The real wage target is usually set to maintain a low rate of unemployment—an important objective for the central union. This system works fairly well in an environment characterized by relatively high productivity growth and high exogenous price inflation—as happened with the terms of trade shocks in the 1970s. The central union’s concern for the overall rate of unemployment moderates the initial or centrally determined nominal wage increase. However, multilevel bargaining does not work well in the context of low productivity growth and low exogenous price inflation—as happened during much of the 1980s when international inflation was low and there were no major increases in indirect taxes or institutionally set prices in agriculture. The real wage target for the centralized union, in this environment, may dictate negligible, or even negative nominal wage increases, because of the anticipated wage drift. If the central union persists with its real wage target and desists from procuring any nominal wage increases at the central level, it is likely to have severe legitimacy problems with the membership. Hence, the central union is forced into bargaining for a nominal wage increase that is not warranted by the productivity performance or employment target. This seems to have happened especially during the mid–1980s, when the increase in real wages was higher than that warranted by productivity considerations. High unemployment was avoided only by the large devaluations in 1981–82 and the expansion of the public sector thereafter; by the end of the 1980s, public sector employment accounted for almost 30 percent of total employment.
Second, the competitiveness of the Swedish economy declined in the 1980s owing to the gradual abandonment of the rules of the EFO model. The EFO model—based on the leading role of the traded goods sector—failed to provide the basis for wage negotiations in 1983, when the engineering workers broke off centralized negotiations and concluded a separate wage agreement. Simultaneously, workers in the nontraded goods sector started negotiating their wages independently of the implications for the competitiveness of the traded goods sector. This was particularly true for public sector workers. Since the nontraded goods sector did not have any exogenous norm as an anchor for determining the wage increase, the wages negotiated in this sector were far too high to maintain the competitive requirements of the traded goods sector. With the breakdown of the discipline provided by the EFO model, uncoordinated, competitive wage increases in the form of increasing wage drift occurred in all sectors of the economy. In the 1970s, wage drift accounted for between 25 and 30 percent of total wage increases. By the end of the 1980s, wage drift constituted more than 40 percent of total wage increases.
Third, contrary to conventional wisdom, labor market programs may at times tend to have a wage‐increasing impact.7 This particular viewpoint contrasts with the more traditional view that labor market programs reduce real wage growth and contribute to low unemployment.8 The argument for the wage‐reducing impact of labor market programs is based on the “insider‐outsider” framework of labor market analysis. Accordingly, it is argued that labor market programs—especially training programs and relief work—which help to avoid problems of de‐skilling of the labor force and long‐term unemployment, promote more effective competition for jobs and reduce upward pressure on wages. However, a recent paper argues that at very low rates of unemployment (as in Sweden between 1974–77 and 1987–1990, when the “open” unemployment rate averaged less than 1½ percent), the wage‐reducing impact of certain types of labor market programs may cease to operate.9 Labor market programs (especially “relief work," which is paid at market wages) now effectively give employed workers, or “insiders,” the guarantee of avoiding open unemployment rather than enabling the unemployed “outsiders” to compete effectively for jobs. In other words, at very low rates of unemployment, the wage‐increasing impact of “active” labor market programs predominates over its wage‐reducing role.
Slow Productivity Growth
The analysis of long‐term trends in productivity is complex. A number of different factors, such as savings‐investment behavior, educational attainment, and the incentive mechanisms available for innovativeness, influence the productivity performance of the economy.10 In Sweden, additional factors influencing the long‐run productivity performance are its large public sector and generous welfare state.11 However, the literature has found it difficult to identify the empirical importance of these various factors in explaining the productivity performance of an economy. In analyzing Sweden’s relatively poor productivity performance after the mid‐1970s, this paper focuses mainly on the impact that the incentive mechanisms operating in the labor market had on productivity—particularly the role of solidaristic wages. Other explanations are not analyzed in this paper—not because they are unimportant, but because there has so far been no systematic analysis of the link between wage bargaining institutions and productivity in manufacturing. The analysis is basically qualitative, suggesting possible causation mechanisms, rather than providing an exhaustive empirically testable explanation for the productivity slowdown.
The Rehn‐Meidner model’s strategy of combining centralized bargaining and wage equalization seemed appropriate in the 1960s and early 1970s when the main purpose was to shift workers from the low‐ technology to the high‐technology industries, where they could take advantage of the rapid international transmission of technical knowledge. This strategy contributed toward enhancing both the average levels of productivity and living standards. However, Sweden’s successful period of technological “catching‐up” implied that by the mid‐1970s a substantial part of Swedish industry was already technologically advanced.12 Therefore, further productivity growth could not be sustained by shifting workers between different sectors of the economy through the solidaristic wage policy. Instead, rapid growth of productivity could only be obtained endogenously, by increasing the efficiency with which existing enterprises operated. This required the creation of appropriate incentive mechanisms that would allow firms to increase productivity by restructuring their enterprises and motivating workers to enhance their human capital.
A wage bargaining system that relied on centralized bargaining and wage equalization did not prove appropriate to the changed circumstances of the 1980s and 1990s. In addition to the diminished prospects for a further period of catching up, new developments in the organization of work during this period also warranted a change in the wage formation regime. These developments, which have involved a shift from standardized assembly line production (“Fordism”) to flexible work practices (“post‐Fordism”) in many industrial countries, have been documented in a number of studies.13 A distinctive feature of the “post‐Fordist” environment is the much greater diversity between individual firms. The level of effort, diligence, and skills that firms expect of their workers under a system of flexible work practices varies greatly. Consequently, efficiency wage considerations require large variations in the remuneration schemes offered by different firms for increasing productivity. In this context, to persist with a labor market institution that compressed wage differentials drastically was not compatible with providing both enterprises and workers the right incentives for increasing productivity endogenously.14
The system of centralized training schemes that operated in Sweden also proved inappropriate for obtaining productivity increases in the new circumstances. Centralized training schemes, by their very nature, impart mainly general skills. However, the new technology of the “post‐Fordist” environment required a much greater emphasis on firm‐specific skills, which, in turn, implied that firms had to play a greater role in the training of workers. The policy of solidaristic wages mitigated against greater reliance on firm‐specific training, however. To find it optimal to offer in‐house training, firms needed the freedom to devise their own internal wage differentials (for example, to be able to offer a steeply increasing age‐related wage profile) and promotion schemes to motivate workers to enhance their human capital, as well as to avoid the moral hazard problems of workers leaving the firm after receiving training. Once again, the rigidity imposed by the system of centralized bargaining and solidaristic wages precluded firms from tailoring their own unique remuneration schemes to offer the optimal amount of internal training to enhance productivity.
This paper has outlined some of the structural problems underlying the current crisis in the Swedish economy. High wage inflation in the 1980s can be traced to the combined impact of multilevel bargaining, the breakdown of the leading role of the traded goods sector in wage determination, and the operation of certain types of labor market programs in conditions of extremely low unemployment. Slow growth of productivity, in turn, has been attributed to the inappropriateness of solidaristic wages in an environment in which flexible working practices have become more prevalent and there is less scope for catching up. These structural problems suggest the need for far–reaching institutional changes in the labor market—particularly increasing wage differentials and dispensing with multilevel bargaining. Reforms to the labor market programs are also needed to make them more appropriate to the changed circumstances rather than abandoning them altogether.
BaumolWilliam J.Sue AnneBatey Blackman and Edward N.WolffProductivity and American Leadership: The Long View (Cambridge, Massachusetts: MIT Press1989).
BosworthBarry P. and Alice M.Rivlineds. The Swedish Economy (Washington: Brookings Institution1987).
CalmforsLars (1993a) Centralization of Wage Bargaining and Macroeconomic Performance: A SurveyInstitute for International Economic Studies Seminar PaperNo. 536 (Stockholm1993).
CalmforsLars (1993b) “Lessons from the Macroeconomic Experience of Sweden,”European Journal of Political EconomyVol. 9 (March1993) pp. 25–72.
CalmforsLars and JohnDriffil“Bargaining Structure, Corporatism and Macro–economic Performance,”Economic PolicyNo. 6 (April1988) pp. 14–61.
EdgrenGostaKarl–OlofFaxen and Clas–ErikOdhnerWage Formation and the Economy (London: Allen $ Unwin1973).
EliassonGunnar and othersThe Knowledge Based Information Economy, (Stockholm: Industrial Institute for Economic and Social Research1990).
LayardRichardStephen J.Nickell and RichardJackmanUnemployment: Macroeconomic Performance and the Labour Market (Oxford, England: Oxford University Press1991).
LindbeckAssarThe Swedish Experience,Institute for International Economic Studies Seminar PaperNo. 482 (Stockholm1990).
LundbergE.“The Rise and Fall of the Swedish Model,”Journal of Economic LiteratureVol. 23 (March1985) pp. 1–36.
MilgromPaul and JohnRoberts“The Economics of Modern Manufacturing: Technology, Strategy and Organization,”American Economic ReviewVol. 80 (June1990) pp. 511–28.
PioreMichael J. and Charles F.SabelThe Second Industrial Divide: Possibilities for Prosperity (New York: Basic Books Inc.1984).
The author is an Economist in the IMF’s European I Department. He holds a Ph.D. from Cambridge University and was a Fellow at Queens’ College, Cambridge, before joining the IMF. The author is grateful to Francesco Caramazza, Martin Fetherston, John Green, and Desmond Lachman for comments on earlier drafts of this paper. He also benefited from the discussion on this paper in the European I lunchtime seminar series.
See Lindbeck (1990).
See Lindbeck (1990) for a more detailed description of these developments.
See Calmfors (1993b).
See, for instance, Layard, Nickell, and Jackman (1991) for a detailed discussion of the impact of labor market programs.
See Calmfors (1993b).
See Baumol, Blackman, and Wolff (1989) for a detailed analysis of the factors influencing productivity in the long run.
Lindbeck (1990) has argued that the generous welfare state and the large public sector are partly to blame for Sweden’s poor productivity performance.
The changing technological structure of Swedish industry is documented in Eliasson and others (1990).
See, for instance, the discussion in Piore and Sabel (1984) and Milgrom and Roberts (1990). Organization of firms on “post‐Fordist” lines is said to have become increasingly important since the mid–1970s. Eliasson and others (1990) document some of these changes for Sweden.
See Ramaswamy and Rowthorn (1993) for a more formal treatment of the relationship between the wage bargaining system, profitability, and productivity.