Skip to main content

The impact of trade unions on jobs and pay

Trade union representation has shrunk over the last twenty years, but could stabilise with the new statutory recognition procedures.

Written by:
Neil Millward, John Forth and Alex Bryson
Date published:

This report uses large-scale survey evidence to examine the effects of unions upon the employment levels of workplaces and upon the pay of their employees. Although some of the findings show that trade unions are associated with adverse outcomes, these effects are absent when employers jointly determine pay and employment matters with the unions.

The impact of trade unions on work and pay contains findings and policy implications on:

  • employment growth Whether the presence of trade unions slows down employment growth, taking other factors into account.
  • workplace closures The extent to which unions precipitate the closure of workplaces in certain sectors.
  • bargaining scope How comprehensive bargaining agendas can ameliorate trade unions' effects on employment.
  • pay levels The effects of different types of union representation on pay levels and the size of recent pay settlements.


Recent legislation has provided trade unions with a right to recognition by employers in certain circumstances. Use of the statutory procedure may help to slow down the decline in union representation that has occurred over the past two decades. But what impact have unions had recently at the workplace? This study examines the effects of unions upon the employment levels of workplaces and upon the pay of their employees. Neil Millward, John Forth and Alex Bryson analysed the recent Workplace Employee Relations Survey and found that:

  • Workplace closures were more likely among unionised plants than non-union plants in manufacturing between 1990 to 1998.
  • One exception was plants with a comprehensive bargaining agenda. For these the likelihood of closure was no greater than for non-union workplaces.
  • There was no difference in closure rates between union and non-union workplaces in the service sector.
  • In the economy as a whole, non-union workplaces grew on average by 1.4 per cent per annum between 1990 and 1998, whereas unionised workplaces shrank on average by 1.8 per cent.
  • This difference remained significant when other factors, such as the size, activity and age of workplaces, were taken into account.
  • In 1998 trade unions did not, on average, negotiate higher pay for the employees they represented, when other factors affecting wages levels were allowed for.
  • But in the private sector, unions did achieve higher pay where they bargained for a sizeable majority of the workforce, or where multiple unions were involved.
  • Pay settlements in 1998 were similar, whether negotiated by unions or not. But pay increases were lower where union negotiations covered most employees, suggesting a long-term decline in the ability of unions to enhance pay.

The issues

In the early 1980s, when trade unions were more widespread and powerful than they are now, research studies showed that they raised pay and constrained employment growth. Legislative restrictions to reduce union power, introduced throughout the 1980s, were tightened further in the 1990s. Membership continued to decline, at least until very recently.

A new assessment of the unions' role in shaping pay and employment levels is therefore appropriate. This study contributes to that assessment by reporting detailed statistical analyses of large-scale surveys that are representative of workplaces and the employees within them. It addresses the following questions:

  • do trade unions have an impact on employers' decisions about jobs?
  • do unions still negotiate higher pay than would otherwise be the case?
  • are other conditions of employment affected in similar ways?
  • and do any non-union forms of employee representation make a difference to these matters?

Workplace closure

The complete closure of a workplace, where jobs are not transferred elsewhere, represents a large-scale loss of jobs. Many factors influence the likelihood of a workplace closing. Its economic performance in the preceding period is a good guide. But statistical analysis shows that closure also depends upon industrial characteristics, the type of workplace, the nature of the workforce and how it is managed.

Taking these factors into account, workplaces with recognised trade unions were no more likely to close between 1990 and 1998 than workplaces without unions.

In manufacturing industry, however, there was a difference. Unionised plants were, on average, 15 per cent more likely to close than non-union plants. This effect was not universal throughout manufacturing, however. It was particularly apparent where recognised unions represented only manual workers but not the rest of the workforce. It was also clear where unions were excluded from negotiating with management about employment matters such as recruitment and staffing levels. But where managements did allow unions a role in determining these aspects of employment, the chances of closure were no different from those of non-unionised plants.

On average unions had no impact on the likelihood of workplace closures in service industries, whether in the private or public sectors.

Employment growth

Two-thirds of private sector workplaces continued to operate with 25 or more employees between 1990 and 1998. Employment in these workplaces grew at a very modest rate of one-third of one per cent per annum, on average.

Employment in the typical unionised workplace in the private sector declined at a rate of 1.8 per cent per annum, whilst employment in the average non-union workplace grew at a rate of 1.4 per cent per annum. This difference persisted after controlling for other factors known to have an impact upon employment levels - thus union recognition restricted the growth of continuing workplaces in the private sector over the 1990s. This negative effect of unions on employment growth was slightly larger in service industries than in manufacturing. However, it was confined to cases in which unions negotiated over wages, but had no role in determining staffing levels or recruitment. The rate of employment growth among service sector workplaces where unions negotiated over wages and employment was no different to that seen among workplaces without recognised unions.

Pay levels

Trade unions continued to raise pay in the private sector in 1998, keeping it above the levels in comparable non-union workplaces. This was so despite their weakened state compared with earlier times. The effect was, however, not a general one. It was particular to two types of situation. One was multi-union representation. The other was where pay-setting arrangements covered at least 70 per cent of employees in the workplace. These circumstances apply to about a half of employees covered by union bargaining or about 15 per cent of all private sector employees. Typically their pay was 9 per cent higher than for similar employees in comparable, non-union workplaces.

Other, smaller groups of employees at these same workplaces also benefited from the unions' bargaining activity, even though they were not included in the arrangements and their pay was set unilaterally by the employer. A few were low-skilled employees, but most commonly they were managers and higher-paid professional workers.

Figure 1 summarises how these direct and indirect effects bore upon employees at different parts of the pay distribution in the private sector in early 1998. The direct impact of union bargaining affected a mere 6 or 7 per cent of employees at the bottom end of the pay distribution. The most extensive impact of unions was for people being paid between £5 and £10 per hour - at least a quarter of them benefited directly from union bargaining. At higher pay levels the direct effect of unions was less widespread. But the indirect impact was at its most extensive in this upper section of the pay distribution - at £10 an hour or more, over 15 per cent of employees benefited from the spill-over from union bargaining on behalf of, not themselves, but other employees at their own workplace.

Separate analysis showed that trade unions were also associated with more widespread provision of some fringe benefits in much the same circumstances as where they achieved higher pay. The benefits studied were enhanced sick-pay benefits and employer-provided pensions.

Pay settlements

Basic pay is adjusted annually for most employees and this is particularly the case where trade unions are involved. But unions appear to affect the process of pay determination more than the outcome.

Pay increases in the private sector in 1997/8 were no greater where unions were involved in negotiations, when other factors were taken into account. If anything, union settlements were smaller than the pay increases given by employers acting on their own.

Particular types of union representation generated lower settlements. Notably, employees covered by multi-union representation had smaller pay increases than employees in comparable non-union workplaces. When this result is taken with the findings on pay levels, the research suggests that the ability of unions to enhance wages and salaries is in long-term decline.

Non-union consultation arrangements

The study also examined consultation arrangements that made no use of union representatives. Although they were widespread in general, consultation specifically about pay increases or about changes in employment levels was less common. Where the arrangements existed they had no impact on pay levels or upon the size of pay settlements. Nor did they affect the likelihood of closure or the rate of employment growth in continuing workplaces. Any new consultation arrangements brought into being by European Union directives therefore seem unlikely to have an impact on these important matters.


The researchers conclude that, although many of the findings show trade unions in an unfavourable light - especially in relation to the loss of jobs - these situations are generally avoidable. Where managements allow unions a role in determining not just pay but employment matters as well, there are less widespread adverse effects of union representation on employment growth and the chances of workplace closure. But they question whether this is the type of situation that the new statutory recognition procedure, contained in the Employment Relations Act, is likely to encourage.

About the study

The study of pay involved statistical analysis of the nationally representative Workplace Employee Relations Survey of 1998, entailing interviews with managers in 2,191 workplaces with 10 or more employees and questionnaires from 28,237 of their employees. The data on pay levels and pay settlement were gathered roughly a year before the National Minimum Wage was introduced in April 1999.

The analysis of employment was based upon the 1990-98 Workplace Employee Relations Panel Survey. Interviews were conducted in 1990 with a senior personnel manager in 2,061 workplaces with 25 or more employees. Follow-up telephone interviews in 1998 determined whether the workplace had closed or survived. A random sample of 834 of the survivors was interviewed in 1998. The analysis of employment growth used this panel of continuing workplaces that had at least 25 employees in both 1990 and 1998.

The 1990 and 1998 management surveys had response rates of around 80 per cent, while the 1998 employee survey, used in the analysis of pay levels, had a response rate of around 65 per cent.


681.pdf (29.15 KB)
185935291x.pdf (196.3 KB)
Music producer in a music studio sat on a chair.

This report is part of the work topic.

Find out more about our work in this area.

Discover more about work