Monitoring the Impact of the National Minimum Wage

Angela Bowring, Alastair Hatchett, Laura James, Simone Melis, Ken Mulkearn, Anna Warberg, Lois Wiggins and Louisa Withers
(Incomes Data Services, IDS)

IDS evaluated the impact of the National Minimum Wage across 173 organisations in 6 low-paying sectors. It looked at recent developments in these sectors, the impact of the increase in the minimum wage in October 2008 and the potential impact of the upratings in October 2009. IDS also looked at the impact of the recession on pay and employment in low-paying sectors. This research built on similar work conducted by IDS for previous Low Pay Commission reports.

The report was written in the autumn of 2009 based on research covering the period from September 2008 to September 2009.

The research methodology included the following.

• Surveys of low-paying sectors: retail; housing and social care; childcare; hotels; fast food, pubs and restaurants; and leisure.
• Interviews with HR managers in these sectors about pay increases, pay rates, changes to pay structures, and other pay and employment practices.
• A time-series analysis of rates of pay for 15 companies since the National Minimum Wage was introduced 10 years ago.
• An analysis of the lowest rates of pay in organisations across the economy.

Key findings included the following.

• The fast food, pubs and restaurants sector had the highest proportion of companies setting their lowest rate of pay at the level of the National Minimum Wage.
• Around 55 per cent of surveyed organisations had to increase their lowest rate of pay in order to comply with the National Minimum Wage.
• Employers focused more on the impact of the recession on their businesses than on the impact of the rise in the National Minimum Wage.
• Around a third of surveyed firms claimed that wages had risen further up the pay scale as a result of the minimum wage. In retail, this had led to a reduction in the differential between geographic pay zones.
• In its panel of large companies, differentials had risen slightly in the last 2 years to almost 5 per cent in 2008 following moderate increases in the National Minimum Wage.
• The long-term trend across most sectors for employers to pay adult rates from age 18 onwards continued, although youth rates continued to be used extensively in some sectors such as fast food, pubs and restaurants.
• Few employers said that an increase in the National Minimum Wage had affected training or development negatively.
• The long-term trend to incorporate premium payments for bank holidays, weekend work and night work into basic pay continued.
• The increase in statutory annual leave affected employers in the fast food, pubs and restaurant sector the most.

A Qualitative Study of Apprenticeship Pay

Kayte Lawton and Emma Norris (Institute for Public Policy Research, ippr)

The research looked at how apprentice pay rates affected young people’s decisions to start and complete an apprenticeship. It also examined employers’ use of minimum wage exemptions and how they might be affected by an apprentice minimum wage. It was a qualitative study which sought to understand the different factors involved in complex decision-making, both by young people and employers. The focus of the research was on industries where apprentices were known to be relatively low-paid. It was conducted between July and September 2009.

It drew on three sources of original data.

• Focus group research with current and former apprentices. This comprised 7 focus groups, with a total of 54 participants across various sectors and locations of the UK.
• A survey of young people, which complemented the focus group research.
• Interviews, with a total of 24 organisations: employers; training providers; careers advisers; and charities working with disadvantaged young people.

The central aims of the project were as follows.

• To gain an insight into how apprenticeship pay rates differ across the UK, including by country, apprenticeship level, industry sector, age and gender.
• To understand the importance of pay in the take-up and completion of apprenticeships relative to other factors (such as peer effects, family background and careers advice), and how this varies by gender, age, sector and country.
• To investigate how employers might respond if there were a national minimum apprentice wage.

Key findings included the following.

• Receiving a wage was a motivating factor for young people starting an apprenticeship, but was rarely the main reason for starting: some other factors, such as family influences, had a greater impact.
• Where pay was considered, the EMA was most commonly used as a point of comparison and apprentices were satisfied they would receive more in work than in full-time education.
• Most apprentices recognised the trade-off between low pay now and higher pay in the future.
• There did not appear to be a strong case for introducing a minimum wage in order to increase demand from young people for apprenticeships: most places were oversubscribed. However, some evidence was found that a minimum wage may help to increase apprenticeship completion rates. Although low pay was not the primary reason for leaving, it was one factor, when alongside low job satisfaction and poor quality training.
• A minimum wage for apprentices may also attract different kinds of people into apprenticeships, and may have an effect on the calibre of apprentice candidates.
• There was a lack of clarity and understanding around minimum wage rules for apprentices. Introducing apprentice pay within the National Minimum Wage framework could help simplify pay arrangements and improve understanding and compliance.
• Employers who made use of the apprentice exemption were very concerned about the potential impact of an apprentice National Minimum Wage on the supply of apprentice places.
• A minimum wage for apprentices would have more of an effect on female apprentices and those in the main low-paying sectors, particularly hairdressing.

Minimum Wage and Staying-on Rates in Education for Teenagers

Augustin De Coulon, Elena Meschi and Anna Vignoles
(Institute of Education), Joanna Swaffield (University of York) and Jonathan Wadsworth (Royal Holloway, London)

This research investigated the impact of the minimum wage on the decisions of young people to stay on in education or participate in work or training. It focused on 16–17 year olds.

The researchers used secondary analysis of large data sets (Longitudinal Study of Young People in England; National Pupil Database/Pupil Level Annual School Census; Local Education Authority and School Information Service; Annual Survey of Hours and Earnings (ASHE); and Annual Population Survey).

Building on the modelling used by Dickerson and Jones (2004), research we had commissioned for the 2004 Report, they analysed the decision to participate in education and training or to enter the labour market, using a multinomial probit model. They performed robustness checks by using different wage measures, econometric specifications and geographical units of analysis.

The research evaluated the impact of the introduction of the 16–17 Year Old Rate using a differential impact approach and exploiting geographical variation. They defined high and low-wage regions, identified whether wages increased for 16–17 year olds (using kernel densities and regression estimates) and then evaluated whether the staying-on rates were affected in geographic areas where wages were lowest.

Recognising the potentially conflicting impact of the introduction of the Education Maintenance Allowance (EMA) on 1 September 2004, a month before the introduction of the 16–17 Year Old Rate, they also looked at the large increase in the National Minimum Wage in 2006.

Key findings included the following.

• There was no evidence that the introduction of the National Minimum Wage for 16–17 years olds in 2004 had any impact on participation in education. There was also no evidence that the large increase in 2006 had any impact on participation in education.
• Overall, the decision to stay on in full-time education did not seem to be driven by the local wage available to 16–17 year olds. Rather the decision was mainly a function of academic ability, social class and other personal/family characteristics.
• There was no evidence of reduced participation in general among youths in low-wage areas compared with high-wage areas.
• However, in some of their analysis, they found evidence of wages affecting the staying-on decision, especially for males with low GCSE results. Pupils with low ability and from a low socio-economic group tended to be more sensitive to changes in local wages.
• There was evidence that some characteristics of the labour market affected participation in education. A higher unemployment rate at the regional level significantly reduced the probability of being in employment and increased the likelihood of staying in education.
• Although there was no evidence that the introduction of the 16–17 Year Old Rate had affected the decision to remain in education, the researchers noted that it was difficult to disentangle this effect from the introduction of EMA in September 2004.

The Impact on Employment of the Age Related Increases in the National Minimum Wage

Richard Dickens (University of Sussex and Centre for Economic Performance, London School of Economics),
Rebecca Riley and David Wilkinson (LLAKES Centre and National Institute of Economic and Social Research, NIESR)

This project revisited the impact of the National Minimum Wage on employment. It proposed a novel approach to exploit the age thresholds of the National Minimum Wage at ages 18 and 22. Some 22 year olds gain a pay rise of up to 20 per cent, and some 18 year olds receive a rise of around 35 per cent, as a result of this legislated change. The research focused on what happened to the probability of employment, unemployment and inactivity at these age thresholds.

In this study, they used a regression discontinuity approach to examine the impact of qualifying for a higher minimum wage rate when an individual became either 18 or 22 years old. They compared the labour market outcomes for those just a few months before an age threshold with those just above it. They assumed that those a few months either side of an age threshold were similar, except that those above were entitled to a higher minimum wage than those below it.

The research used the quarterly Labour Force Survey (LFS) as it contained the individual’s actual birth date and the actual day of the interview. Those who were just above and below the age of 18 and 22 at the time of the survey were identified.

The researchers pooled data from 1999–2009 to increase their sample sizes but argued that the relativities between the youth minimum wages and the adult rate had been similar over time.

They tested robustness using different age thresholds, gender, skill levels, and data prior to the introduction of the National Minimum Wage.

Key findings included the following.

• There was a positive and statistically significant change in probability of employment at the age of 22. Comparison of low-skilled individuals either side of their 22nd birthday, using quasi-experiments on impacts of the minimum wage, suggest a 5 percentage point rise in employment rate on turning age 22, from around 55 per cent to 60 per cent.
• A plausible explanation for this result was a labour supply effect as young, unskilled people responded to a large pay rise by entering work more readily.
• Around two-fifths of this rise was accounted for by a fall in unemployment and three-fifths by a fall in inactivity.
• The results suggested a stronger employment effect for women (around 6 percentage points). The increase in employment for women was mainly achieved from a reduction in inactivity. There was little effect on unemployment. In contrast, the increase in employment for men was a result of declining unemployment rather than inactivity.
• These results were robust to various specifications.
• No labour market activity effects were found at age 21 or 23, nor was there any effect prior to the introduction of the minimum wage.
• Comparison of low-skilled individuals either side of their 18th birthday were inconclusive. There were other regulatory changes that happen at this age and certain jobs were only available to those aged 18 and over.
• The research concluded that there would be little harm in legislating to start the adult minimum wage rate at age 21 as the evidence did not suggest this would lead to deterioration in employment rates for young low-skilled people. It may even raise employment rates.

Taxes, Benefits and the National Minimum Wage

Mike Brewer, Richard May and David Phillips,
(Institute for Fiscal Studies, IFS)

This project looked at the interaction of the minimum wage with the tax and benefits system.

IFS used the Family Resources Survey (FRS), LFS and TAXBEN (IFS’s tax and benefit micro-simulation model) to:

• analyse the characteristics of those who earn the National Minimum Wage and families containing such individuals;
• investigate the position of families with a National Minimum Wage earner in the working-age income distribution;
• investigate the interaction of the National Minimum Wage with the tax and benefits system by calculating marginal effective tax rates (METRs) and participation tax rates (PTRs) for minimum wage earners, and compare these with those of other workers; and
• see how reforms to the tax and benefit system in the period 1999 to 2011 affected or will affect National Minimum Wage earners and their families.

The main dataset used was the quarterly LFS. The researchers pooled 4 quarters covering the financial year 2007/08. A later year was available but they used this to be consistent with the latest FRS (2007/08). In order to analyse the effect of the minimum wage on net income, information was required on different income sources and housing costs. Many of these were not available in LFS so were imputed from FRS using regression techniques.

Key findings included the following.

• National Minimum Wage workers were more likely to be women than men, more likely to be aged under 25, but unlike findings from previous work, not likely to be part of a couple. In addition, they were more likely to be from a ethnic minority background, less educated than other workers and more likely (than other workers) to live in social housing. They also tended to be part-time and work in retail or hospitality.
• Families where the main worker was paid the National Minimum Wage for their main job were concentrated towards the bottom of the working-age income distribution.
• Workers earning the National Minimum Wage in their main job were more likely to have very low METRs (generally the very poorest or second earners in couples), and more likely to have very high METRs (those subject to means-tested benefits and withdrawal of tax credits), than workers paid more than the minimum wage.
• Those with very low METRs gain the most (and those with high METRs the least) from an increase in the minimum wage.
• Families containing National Minimum Wage earners gained more from changes in the tax and benefit system between October 1999 and April 2007 than other working families.
• The expansion of means-tested tax credits between 1999 and 2007 increased the METRs faced by many minimum wage workers.
• Proposed changes to the tax and benefit system between 2007 and 2011 have a much smaller distributional impact than those up to 2007, and so families containing a National Minimum Wage worker do not gain as much, relative to other working families, as they did in the earlier period (1999–2007).

Estimating the Impact of the 7th NMW Uprating on the Wage Growth of Low-wage Workers in Britain

Joanna Swaffield (University of York)

This research attempted to address the question of what would have been the wage growth of employees directly affected by the minimum wage if the minimum wage had not been introduced or uprated. It extended and updated the analysis in Swaffield (2008) to the October 2006 upratings.

She compared a group of employees that were affected by the minimum wage (treatment group) with a group of employees who were similar to the affected group but who were paid above the minimum wage (control group).

Using ASHE and LFS, the researcher estimated difference-in-difference regressions for each of the first seven upratings to the National Minimum Wage.

She used three measures of real wage growth – an absolute measure of wage change, a growth (or percentage) measure and a measure of the probability of receiving real wage growth. For each measure, the experience of wage growth of individuals directly affected by the minimum wage was compared (before and after the introduction/uprating of the minimum wage) with the experience of a similar group not affected in this way. The control group was defined as those paid less than 10 per cent above the forthcoming minimum wage. Various definitions of treatment group were used.

A variety of sensitivity tests were carried out to check the robustness of her results.

Key findings included the following.

Testing for Spill-over Effects of the National Minimum Wage

Mark Stewart
(University of Warwick)

This research investigated whether the minimum wage upratings have effects on wages beyond the increases required to bring those previously below the new minimum wage up to it, and if so how extensive these ‘spill-over’ effects were. Three approaches were adopted to look at this issue.

• Analysis of individual wage changes – he looked directly at the individual wage changes of those just above the new minimum and compared them with a counterfactual based on price inflation and average earnings increases.
• Comparison of wage quantiles before and after an increase in the minimum wage – he used a method similar to Dickens and Manning (2004), comparing quantiles of the observed wage distribution with an estimated counterfactual ‘compliance change’ distribution. The vertical difference in the percentiles was then used for the quantile regressions that explored ‘spill-overs’ in greater depth.
• Comparison of wage distribution functions before and after an uprating – he used a non-parametric approach to compare the observed wage distribution with an estimated counterfactual distribution constructed by making appropriate adjustments to the observed wage distribution before the increase.

Data from ASHE were used to test whether the introduction of the minimum wage and subsequent upratings had led to any spill-over effects.

Key findings included the following.

• Stewart found that the definition of the counterfactual wage distribution was crucial to his analysis. He noted that, not surprisingly, the estimates and tests of the spill-over effects were found to depend importantly on the assumptions made for the specification of the counterfactual.
• In his analysis of individual wage changes, he found some evidence of spill-over effects when the base year used was 1999–2000 but not when it was 1997–1998. He found no systematic evidence of any spill-over effects and concluded that there was no evidence of spill-over effects at all for models with interaction restrictions.
• In his comparison of wage quantiles before and after an increase in the minimum wage, he found significant but limited spill-overs up to the 6th percentile resulting from the introduction of the minimum wage in 1999. There was slightly stronger evidence of spill-overs up to the 11th percentile as a result of the 2001 upratings and up to the 20th percentile in estimates for the October 2002 upratings. Using an alternative counterfactual, the evidence on spill-overs was much reduced but was still around 1 per cent (up to the 20th percentile) for the October 2002 upratings.
• Using comparisons based on the estimated wage distribution functions, Stewart found that there was some evidence of significant spill-overs from the October 2002 and October 2004 upratings. Under alternative counterfactual assumptions, the finding for 2004 disappears but significant spill-over effects remain for October 2002.
• He expressed concerns about the wage distribution method and concluded that the individual wage and wage quantile approaches were more robust. However, these two approaches obtained conflicting results.

The Impact of the National Minimum Wage on the Wage Distribution

Tim Butcher (Low Pay Commission), Richard Dickens (University of Sussex) and Alan Manning (London School of Economics)

Previous work on the impact of the National Minimum Wage on the wage distribution found little effect on differentials. However, more recent evidence (Butcher, 2005) showed relative wage increases for the bottom 20 per cent of workers suggesting some spill-over effects of the National Minimum Wage over a longer horizon or some other factors at work. Other work using the LFS (Dickens and Manning, 2006) also suggested much larger spill-over effects than had originally been found.

The researchers investigated this issue further using data from ASHE, which was the only data source providing accurate pay data on a consistent basis over the period of the introduction of the National Minimum Wage.

This research built on the model of spill-overs developed by Lee (1999) in looking at the increase in United States wage inequality in the 1980s and 1990s.

The researchers estimated direct and spill-over effects at the aggregate level and separately by gender. They then estimated their model for around 135 areas that broadly captured local labour markets.

Key findings included the following.

• The National Minimum Wage was introduced 10 years ago and, at about the same time, wage inequality at the bottom of the earnings distribution started to fall, having risen over the preceding 20 years. The falls relative to the median went up to the 25th percentile. It was tempting to assign this to the minimum wage but the minimum wage only directly affects around 5 per cent of employees.
• The direct effect of the minimum wage had not been enough to explain the observed fall in wage inequality. However, allowing for modest spill-over effects, their model fitted the decline in inequality at the bottom of labour market reasonably well.
• They found that the direct effect was largest at the bottom percentile, raising wages by nearly 30 per cent. This effect rapidly declined and only reached up to the 6th percentile. There were modest spill-over effects. It was largest at the 6th percentile, raising wages by about 7 per cent more than in the absence of the minimum wage, and stretched further up the pay distribution. At the 10th percentile it was 4 per cent and it was still over 1 per cent at the 17th percentile.
• For women, the spill-over effect was greater and reached further up the distribution. It was largest at the 8th percentile and small effects were still observed at the 20th percentile.
• They also found that areas most affected by the minimum wage, the lowest-paying areas, had the largest spill-overs with effects evident up to the 25th percentile.
• They concluded that spill-over effects may be larger than previously thought and were much greater than any purely direct effect.

Investigating the Impact of the Minimum Wage Regime on the Labour Market Behaviour of Young Workers: Moving Towards a Structural Dynamic Approach

Haroon Chowdry, Costas Meghir and Jonathan Shaw
(Institute for Fiscal Studies, IFS)

This research noted that most existing empirical literature into the impact of the minimum wage had been non-structural, involving some statistical comparison of labour market outcomes for those workers directly affected by the minimum wage with those who were not.

There were, however, limitations to this treatment and control group (difference-in-difference) methodology. It typically relied on assuming that there were no equilibrium effects (that is, no spill-overs to workers just above the minimum wage and no substitution between different types of workers). Our commissioned research findings for this year and in previous years call this assumption into doubt.

Instead, the researchers suggested using a structural dynamic approach, which they claim can incorporate the equilibrium effects above.

They conducted a survey of the literature on dynamic structural models of education and the labour market and then investigated the availability of data required to estimate such models.

Key insights were used to propose a detailed set of steps required to construct a model capable of answering the empirical issues around the impact of the National Minimum Wage on young people (and any other group for that matter).

Key findings included the following.

• The literature review concluded that search and wage posting models could in principle be used to analyse issues around young people and the minimum wage. Nevertheless, most of these models to date have been too narrow in their focus for this purpose, with no insights given into the relationships between minimum wages and participation in education or training, or minimum wage age bands and substitution effects across different age groups. Furthermore, only a couple of models in the literature were actually able to allow for ambiguous effects on employment. Most imposed theoretical constraints that did not allow for a positive impact. A substantial amount of modelling work remains to be done to bring such models up to the level of sophistication required to address these issues.
• The research laid the foundations for future modelling work by outlining the desirable features of a sufficiently detailed equilibrium search model and how they might be implemented.

1. Endogenous firm behaviour (both firm and worker behaviour determined within the model). The wage posting models seemed more suitable in this regard.

2. ‘Agnostic’ predictions about the impact of the minimum wage e.g. flexible predictions about the impact on employment.

3. Age-related minimum wage rates (and experience), by segmenting the labour market by age group or allowing different ages to compete with each other for jobs.

4. Education choices, by using segmented markets, or by treating workers with different skills as different labour inputs.

• Such models could, in principle, be estimated from labour force panel data on employment/unemployment durations and wages, such as LFS or the British Household Panel Survey.
• Because of the amount of modelling work required, the recommendations were necessarily tentative. But the researchers remained optimistic that it would be feasible to construct an advanced structural model of minimum wages and labour market outcomes for the UK.