0 Preliminary: Comments Welcome, Not to be Quoted Without Authors’ Permission The Economic Effects of the Introduction of the UK National Minimum Wage Stephen Machin * , Alan Manning ** and Lupin Rahman *** July 2000 Revised February 2001 * Department of Economics, University College London and Centre for Economic Performance, London School of Economics ** Department of Economics and Centre for Economic Performance, London School of Economics *** Centre for Economic Performance, London School of Economics Abstract Between 1993 and April 1999 there was no minimum wage in the UK (except in agriculture). In this paper we study the impact effects of the introduction of a National Minimum Wage (NMW) in April 1999 on one heavily-affected sector, the residential care homes industry. We look at the impact on both wages and employment. Our results suggest that the minimum wage raised the wages of a large number of care homes workers, causing a very big wage compression of the lower end of the wage distribution thereby reducing wage inequality. There is some evidence of employment and hours reductions but no effect on home closure. Acknowledgements We would like to thank the numerous students who punched the care homes data in to the computer and participants in seminars at the Reserve Bank of Australia, STICERD’s Inequality Conference, Warwick and the EEEG Conference in Southampton.
Transcript
0
Preliminary: Comments Welcome, Not to be Quoted Without Authors’ Permission
The Economic Effects of the Introduction of the UK National Minimum Wage
Stephen Machin*, Alan Manning** and Lupin Rahman***
July 2000 Revised February 2001
* Department of Economics, University College London and Centre for Economic Performance, London School of Economics
** Department of Economics and Centre for Economic Performance, London School of Economics
*** Centre for Economic Performance, London School of Economics Abstract Between 1993 and April 1999 there was no minimum wage in the UK (except in agriculture). In this paper we study the impact effects of the introduction of a National Minimum Wage (NMW) in April 1999 on one heavily-affected sector, the residential care homes industry. We look at the impact on both wages and employment. Our results suggest that the minimum wage raised the wages of a large number of care homes workers, causing a very big wage compression of the lower end of the wage distribution thereby reducing wage inequality. There is some evidence of employment and hours reductions but no effect on home closure. Acknowledgements We would like to thank the numerous students who punched the care homes data in to the computer and participants in seminars at the Reserve Bank of Australia, STICERD’s Inequality Conference, Warwick and the EEEG Conference in Southampton.
1
I. Introduction
In April 1999 the UK government introduced a National Minimum Wage (NMW)
to the UK labour market. This is the first time that the UK labour market has had an
economy wide minimum wage and, as the old industry-based Wages Council system was
abolished in 1993, the NMW was introduced into a labour market which had no
minimum wage legislation in operation.1 Because of this, the introduction of the UK
NMW provides a very good testing ground for evaluating the economic effects of
minimum wages. This is what we undertake in this paper.
Our analysis is based upon a large scale survey that we carried out before and
after the introduction of the NMW. This survey focused upon the whole population of
residential care homes in Britain, collecting information on all workers in the sampled
homes, and on an array of home characteristics. We chose to look at this sector for
several reasons. First, it contains many low wage workers (a lot of whom are part-time
female workers), so if the minimum wage is likely to have had an important impact on
economic outcomes it ought to be in an industry like this one. Second, we chose this
sector because it is not unionised. Thirdly, it consists of large numbers of small firms
(average employment being somewhere in the range of 15-20 workers) doing a very
homogeneous activity in geographically concentrated markets. The small size of these
firms means that monitoring problems are unlikely to be severe because the owner
normally also works in the home and also that collection of good quality data on all
workers is feasible. Finally, the product market side in this sector is interesting. A
substantial fraction of the old people in these homes have their care paid for by the
1 Except in agriculture where the Agricultural Wages Board was not abolished along with the other Wages Councils.
2
Department for Social Security (DSS). But, the amount they will pay is capped and was
not increased when the minimum was. Homes whose residents are paid for by the DSS
are then in a situation where they are unable to pass any of their cost increases on to
prices: this is likely to make the employment effect worse.
We consider what happened to wages and employment in the care homes sector in
the period surrounding the introduction of the NMW. The analysis confirms that the
choice of sector is a good one for studying the likely impact of minimum wages. Pay is
very low, with the average hourly wage being around £4 just before the minimum wage
introduction. Before its introduction around 40 percent of workers were paid less than
the minimum wage. In April 1999 we see a spike in the wage distribution of around 30
percent at the minimum wage. Clearly the minimum wage ‘bit hard’ in this sector. It
therefore provides a very good environment for looking at the impact of the minimum
wage on employment.
We look at the employment effects of the minimum wage by considering what is
now a standard technique in the empirical literature on minimum wages and employment
by relating changes in employment before and after the minimum wage introduction to
the fraction of low paid workers in the pre-minimum wage period (see, for example,
Card’s, 1992, state based study of an increase in the US federal minimum wage in the
early 1990s). Our results point to a small negative impact of the minimum wage on
employment and hours in the care homes sector.
The rest of the paper is structured as follows. Section II presents a brief history of
minimum wage legislation in the UK. Section III describes the data and presents some
descriptive statistics. Section IV presents our empirical findings on the wage effects of
3
the minimum wage. Section V then moves on to consider the employment and hours
effects. Section VI then concludes.
II. Minimum Wages in the UK
Unlike in other countries, minimum wages have not historically had an important role to
play in the UK labour market. There used to be an industry-based system of minimum
wage floors, the Wages Councils, which operated from the start of the century up to their
abolition in 1993. In earlier work we have studied the impact of the Wages Councils,
concluding that their activities did not harm employment (Dickens, Machin and Manning,
1999; Dolado et al, 1996)
Following their election in 1997, Tony Blair’s Labour Government was
committed to introducing a National Minimum Wage. It set up a Low Pay Commission
consisting of academics and representatives of employers and workers to report on a
suitable form and level. Eventually, a minimum wage of £3.60 per hour was introduced
in April 1999 for those aged 22 or older, with a lower youth rate of £3 per hour for those
aged 18-21 inclusive (those aged less than 18 were not covered). This was approximately
x% of average hourly earnings and it was expected that x% of workers would be directly
affected.
III. Data Issues and Descriptive Statistics
Most existing UK data is either not yet available or not well suited to carrying out a
before and after evaluation of the economic impact of the introduction of the minimum
4
wage to the UK labour market.2 For this reason we decided to collect our own data. We
wanted to focus on a situation where the minimum wage has the potential to have an
important impact and so chose to collect data on employers and workers in residential
care homes.
There are several reasons for choosing this sector. First, it is a leading employer
of many low wage workers. The principal occupation, care assistants, is one of the
lowest paid occupations in the UK. Second, most homes are reasonably small (average
employment size < 20; median employment size = 15) and this enables us to collect data
on all workers within the homes. Third, there are basically no trade unions to distort
wage-setting procedures in this sector.
Sample Design
Our sample design was to sample the population of UK care homes before and after the
introduction of the minimum wage. We obtained lists of all homes from the Yellow
Pages Business Database in July 1998 (for the pre-minimum sampling) and in May 1999
(for the post-minimum sampling). There were 11635 care homes in the former and
11036 homes in the latter. As one of the things one might be interested in is the extent to
which employers adjusted wages before the minimum wage introduction we sampled
(based on area stratification) one-ninth of the homes in each of the nine months before
minimum wage introduction, and then we re-sampled the homes (including new homes),
and again one-ninth at a time, in the nine months following the introduction of the wage
floor. We also identified home closures that occurred over this time period.
2 For example the survey containing the best wage data, the New Earnings Survey, has several drawbacks. First, it is carried out in April of each year which is unfortunate as the minimum wage was introduced in April 1999 resulting in wages contained in the survey being a mish mash of pre- and post-introduction wages. Second, and even more important, it undersamples low wage part-time workers (as workers below
5
The questionnaire was mailed to the manager of the care homes and asked a range
of questions about the home, about the views managers (who are often home owners) had
about the minimum wage. For obvious reasons, the precise nature of the attitudinal
questions was different for questionnaires sent out before and after the introduction of the
minimum wage. Managers were also asked to provide data on job title, sex, age, length
of service, possession of a nursing qualification, weekly hours and hourly wages for all
workers. The exact questionnaires are reproduced in the Appendix to the paper. For a
postal survey we achieved a reasonable response rate (of the order of around 20 percent)
and, in terms of number of residents, area and employment size of homes, the responses
were representative of the population as a whole. Managers were less likely to complete
all the information on worker characteristics. Where there is missing information on
hourly wages and/or hours we imputed them using the average for job within that firm.
We report both results including and excluding imputed figures below.
Descriptive Statistics
Some features of the sample are described in Table 1. The first two columns
show the characteristics of the sample of workers and homes pre and post minimum wage
introduction, for all homes that we obtained some responses on worker characteristics.
It is clear that wages are very low in the sector we have chosen to study. Before
the introduction of the minimum wage average wages were about £4 per hour (whether
we include imputed results does not make much difference). The choice of sector
therefore clearly satisfies one of the criteria we wanted to emphasise, namely that we
wanted to look for the economic effects of minimum wages in a situation where the
the National Insurance weekly earnings lower limit are not well picked up). Other surveys may well be characterised by measurement errors in wages as they are employee, rather than employer, surveys.
6
imposition of the minimum wage floor has the potential to affect a reasonably large
number of workers. What is also interesting is the fact that average wages rose
significantly after the minimum wage was introduced, going up by somewhere around 25
pence per hour (or around 6 percent). We consider this in much more detail in the next
Section where we discuss the ‘bite’ of the minimum wage introduction.
The Table also documents other characteristics of the sample. Average home size
is fairly small, both in terms of workers and residents. Mean employment is in the range
of 16 to 18 workers, and the average number of residents is just over 20 per home. They
both remain fairly constant before and after the minimum wage introduction.
As stated earlier the principal occupation in this sector is care assistants who
comprise around 62 percent of the workers in the sample. The workers are typically
older workers (average age = 40), overwhelmingly female, working an average of 25
hours per week. Only about 10% have a nursing qualification (other educational
qualifications are not relevant in this sector). The average job tenure is 3.5 years.
Finally, we have collected other home level information on the occupancy rate of
beds and, since the sector has price regulation operating through local authorities, on the
percent of residents who pay local authority prices for beds. This latter feature of the
sector is, of course, very interesting in the context of minimum wage effects. One
argument sometimes put forward in the literature is that one may not observe
employment responses to minimum wages if employers are able to pass minimum wage
increases on to consumers in the form of higher prices. This seems unlikely to happen in
the care homes sector as prices are, in many cases, controlled by local authorities. In this
regard, it is useful to note that prices do not seem to go up by much (and certainly by
7
nowhere near as much as average wages) after the minimum wage introduction. We
return to this later, after considering the wage and employment effects of the minimum
wage introduction.
Because we are interested in comparing employment before and after the
introduction of the minimum wage, most of our estimation uses a panel of homes for
whom we have a response before and after the introduction. The third and fourth
columns present the descriptive statistics for these homes. They are broadly
representative of the whole sample. Some of these homes have quite large numbers of
imputed hours and hourly wages so the final two columns report the descriptive statistics
restricting attention to those homes where less than half of the workers have missing
hours or wage information.
IV. Wages and the Minimum Wage
The main aim of this paper is to look at the employment consequences of the minimum
wage introduction in the low wage sector that we have sampled. However, before
considering this, one clearly needs to establish that the minimum wage has had the effect
one expects on wages and the distribution of wages. Confirming that the minimum wage
introduction had real ‘bite’ and affected the wages of low wage workers in the expected
direction is clearly a prerequisite before one goes on to look at the impact on
employment.
The ‘Bite’ of The Minimum Wage Introduction
The UK National Minimum Wage was introduced in April 1999 at £3.60 per hour for
workers aged 22 or more, and at £3.00 per hour for 18-21 year olds. When presenting
measures of the impact of the minimum wage we sometimes use these age-specific
8
minimum wages and sometimes just the adult minimum: the reason for this is that there is
a spike in the youth wage distribution at the adult minimum after the introduction so that
one could argue that the adult minimum is the effective minimum (see Katz and Krueger,
1992, for evidence that in the US the youth sub-minimum is rarely used). Table 2 reveals
that a large number of care home workers were paid below the minimum wage before
April 1999. Something like one-third of workers were paid less than their age-specific
minimum wage and 38% paid below the adult minimum rate.
Table 2 also presents measures of the ‘wage gap’ the percentage increase in
wages needed to bring workers up to the minimum. The wage gap in firm i is computed
as ∑∑ −
=
jjiji
j
minjiji
Wh
)max(Wh ji
i
WGAP where hji is the weekly hours worked by worker j in
firm i, Wji is the hourly wage of worker j in firm i, and Wjimin is the minimum wage
relevant for worker j in firm i (this might be the age-specific or the adult minimum).
Table 2 shows that the wage gap averages 4% using the age-specific minimum and 4.7%
using the adult minimum.
After the minimum wage is introduced, there seems to be very little under-
payment. One might be concerned that those firms that subsequently pay illegally below
the minimum wage do not respond to our survey. However, there is no evidence thst this
is the case. If this was true we would expect to see the initial wage levels in our sample as
a whole being below those in our balanced panel as the latter group also repsond after the
9
minimum wage is increased: in Table 1 one can see that there is no evidence that this is
the case3.
Table 2 shows there is a very noticeable spike at the minimum after April 1999.
Something like 28% of workers are paid the age-specific minimum and 30% the adult
minimum. Not surprisingly the minimum wage introduction therefore had a sizable
impact on wage dispersion: the gap between the 50th and 10th percentiles of the
log(hourly wage) distribution narrowed from 0.21 to 0.09. It is also interesting to notice
that the upper half of the distribution was unaffected with the gap between the 90th and
50th percentiles of the distribution not changing. A companion paper (Machin and
Manning, 2001) investigates the impact of the minimum wage on the wage distribution as
a whole.
Measuring the Impact on Care Home Wages
The statistics of Table 2 present evidence about the impact of the minimum wage on the
overall wage distribution. But, for our empirical analysis, we would also like evidence
that the homes that we would most expect to be affected were, indeed, the most effected.
This is the purpose of this section. More specifically we estimate wage change equations
of the form:
itit11ti,11it e? XdMINßa)? ln(W +++= − (1.1)
where ∆lnWit is the change in wages for home i in the period surrounding minimum wage
introduction (t-1 denotes the period before, t the post-minimum period), MINi, t-1 is a
measure of the importance of the minimum wage for home i (defined below), ∆Xit is the
change in other home and worker characteristics between t-1 and t and ε is an error term.
3 We also investigated whether the sample response rates fell disproportionately in the low-wage regions
10
The parameter β1 is what we want to estimate and can be thought of as measuring the
relation between wage changes and the minimum holding constant the other factors we
control for through the change in X.
There are several practical concerns with this kind of equation. First, and most
important, is how one measures MINi, t-1. We use the two measures of the impac tof the
minimum wage already discussed in Table 2: the proportion of workers paid less than the
minimum wage in the period just before its introduction, and the wage gap. It is not clear
which is the better measure. For example, if the minimum wage caused all workers
intially paid below it to lose their jobs, then the headcount might be the better measure,
but if it is more difficult to raise the productivity of those a long way below the minimum
wage than those near it, then the wage gap measure might be better.
Because the minimum wage introduction in the UK was at national level the
variation across homes in the impact of the minimum wage all comes from variation in
the initial level of wages. β is then only a true measure of the impact of the minimum
wage if, in the absence of the minimum wage, there would be no relationship between the
initial level of wages and the change in wages i.e. if wages follow a random walk. This is
the implicit identification assumption in using an equation like (1.1) to estimate the
impact of the minimum wage.
We test this identifying assumption by also looking at the relationship between
wage changes and initial wages using some data we collected in an earlier time period
(1992/93) for care homes on the south coast of England (see Machin, Manning and
wherethe minimum wage had more bite. There was no evidence this was the case.
11
Woodland, 1993, for more detail on this earlier survey).4 This enables us to consider
whether one observes a different association in the period surrounding minimum wage
introduction as compared to an earlier time period where no such policy was put in place.
Estimates of Wage Equations
Tables 3a and 3b report estimates of care home-level wage change equations for hourly
wages (Table 3a) and for weekly wages (Table 3b). The first four rows estimate wage
equations including the low-pay and wage-gap measures of MINi , t-1 described above with
and without additional controls. These all show evidence of bigger wage increases in
homes with more low-wage workers. We will concentrate our discussion on the hourly
wage results in Table 3a though the weekly earnings results in Table 3b are very similar.
The associations are very strong in statistical terms and are sizable. For example,
according to column (2) of Table 3a a home with one third of its workers paid less than
the minimum saw average wage growth of 3.5 percent higher than one which had one
tenth of its workers paid less than the minimum. This is large given that average wage
growth was around 6 percent in this time period. It is worth nothing that the wage gap
measure appears to do a much better job in explaining the change in average wages than
the headcount measure.
These results appear to establish an important impact on wages of the minimum
wage but, as noted above, it may merely be because there has always been a link between
wage growth and initial low pay. So rows (5) and (6) of the Tables compare the
relationship between wage growth and initial wages in the 1998/99 and 1992/93 surveys.
4 This 1992/93 survey was carried out for the same reason as the current one, namely to evaluate the impact of minimum wage introduction on care homes. The first wave of the survey was carried out before the April 1992 election as the Labour Party manifesto had committed to introduce a minimum wage if they
12
First, we estimate the more general wage equation of the following form relating wage
change to the initial period average wage:
it1ti,22it )ln(Wßa)? ln(W υ++= −
where the focus is now on estimating β2 , the association between wage changes and the
initial average wage (υ is an error term).
The significantly negative coefficient on the initial wage in row (5) of Tables 3a
reconfirms that, in the period surrounding minimum wage introduction, wage growth
was higher in firms with lower wages in the initial period. Column (6) looks at the
situation in 1992/93. There is also a negative coefficient on the initial wage, but it is
clear that its magnitude is nowhere near as marked as in 1998/99.
We also report the results of wage growth equations for the 1998/99 data that
include both the initial wage and the low-pay and wage-gap measures of the impact of the
minimum wage. Even controlling for the initial wage, there is still evidence of a
significant impact of the wage gap measure on wage growth: in fact the coefficient on the
initial wage is reduced a level similar to that found in the 1992/93 data. It is also worth
noting that when we restrict the sample to the ‘clean panel’ in which less than half the
worker information is imputed, the coefficient on the wage gap measure increases (rows
(9) and (10).
Some graphs also serve to make the same point. Figure 1a plots the relationship
between average wage growth and the initial log wage in 1998/99 and Figure 1b does the
same for the 1992/93 data. While it is very clear that there is a negative relationship in
were elected. Their loss of the election meant our plan to carry out a before and after analysis of minimum wage introduction was scuppered.
13
both periods, the strong diagonal effect on homes with initial low wages is very apparent
in the 1998/99 data. This is the impact of the minimum wage.
County Level Analysis
So, far we have considered differences in wage growth across firms with different
initial levels of wages. But, we might also be interest in the impact across areas with
different initial levels of wages. Because of regional differences in wages (largely the
result of regional differences in house prices), the impact of the national minimum wage
was very different in different parts of the country. In our sample there were no worker
initially paid below the minimum wage in Berkshire, Buckinghamshire, Hertfordshire or
Oxfordshire while over 60% of workers were low-paid in Glamorgan, Humberside,
Lincolnshire, Merseyside, Northumberland, South Yorkshire and Tyne and Wear. We
would expect to find bigger increases in wages in those counties with the lowest initial
level of wages. Here, we supplement our data with data on care assistants from the
Labour Force Survey.5
Table 4 reports county-level hourly and weekly wage equations from aggregated
care homes data and from the Labour Force Survey. A reassuring similar pattern
emerges. Counties with more low wage workers in the period before minimum wage
introduction had faster wage growth (hourly and weekly) and the negative, strongly
significant, relationship between wage changes and the initial average wage is again
present. The impact of the wage-gap measure seems larger in the county-level than the
5 This exercise is essentially the same as Card’s (1992) analysis of state level wage and employment changes before and after the US 1991 federal minimum wage increase. He relates state-level changes to the initial proportion of workers paid less than the new minimum, arguing that identification comes from the fact that more workers wages were affected in some states than others.
14
firm-level data possibly reflecting spillovers from low-wage to higher-wage firms within
an area.
The evidence of this Section has established a clear and important positive effect
on wages in the period when the national minimum wage was introduced. We investigate
the possible employment effects in the next section.
V. Employment and the Minimum Wage
We analyse the employment consequences of the minimum wage introduction using the
same methodology as the wage analysis i.e. we estimate equations of the form:
4 4 4it i,t 1 it it?ln(N ) a ß MIN d X ω−= + + +
As before, the implicit identification assumption here is that, in the absence of the
minimum wage, there would be no relationship between employment growth and the
level of wages. It is not obvious this is the case (e.g. homes that are doing less well might
pay lower wages and have lower employment growth) so we start by investigating the
relationship between employment growth and initial wages in the 1992/93 data.
Changes in Employment and Initial Wages
Table 5 begins the analysis of employment links by considering the associations between
changes in care home employment and the initial wage for the period surrounding
minimum wage introduction (1998/99) and for the earlier period (1992/93) where no such
policy intervention occurred. The following equation is estimated:
15
it1ti,33it )ln(Wßa)? ln(N ξ++= −
where it is now employment change for home i, ∆ln(Nit), that we relate to the initial wage
(ξ is the equation error).
Table 5 presents estimates of this equation for two measures of employment, the
number of workers and total workers hours, for the 1998/99 period and for the earlier
1992/93 period. The first thing to note is that the correlations of employment changes
with initial wages are much weaker than the correlation of wage growth with initial
wages. There is, however, some evidence that the associations between employment
change and the initial wage are more positive in 1998/99 than in 1992/93. The average
number employed effects are weak but there is more evidence of an effect on total hours.
For example, the t-statistic for the hypothesis that the gap between the coefficients on the
initial log hourly wage is zero is 1.43 for the total employment and 1.96 for total hours.
Changes in Employment And Initial Minimum Wage Variables
Figures 2 and 3 plot the basic data used in the regressions that are discussed
below. The change in log total employment is plotted against the initial proportion low-
paid in Figure 2a and against the initial wage-gap in Figure 2b. Figures 3a and 3b do the
same but with the change in log total hours on the left-hand axis. The regression results
are reported in Table 6, again for total employment and hours in the upper and lower
panels respectively. Columns (1) and (5) report the coefficient on the initial proportion
low-paid and the initial wage-gap respectively in a regression where the dependent
variable is the change in total log employment and there are no other controls. The
estimated impact of the minimum wage is negative though not significantly different
from zero. However the addition of controls (columns (2) and (4)) and restricting the
16
sample to those with relatively complete information on worker characteristics makes the
coefficients both larger in absolute terms and more significant. When other controls are
included and the restricted samples is used, the minimum wage variables are estimated to
have a significant negative effect on employment growth. We also report implicit
elasticities if the minimum wage was raised by 10p (the adult rate going from £3.60 to
£3.70) and 40p (the adult rate going from £3.60 to £4.00). These elasticities ar ein the
range of –0.2 to –0.3.
The bottom half of Table 6 uses the change in the log of total hours worked as the
dependent variable. The coefficients become more a little bit more negative and a little
bit more significant.
While the employment effects are clearly nowhere near as strong as the wage
effects considered, they do suggests that employers cut employment and hours in
response to the minimum wage.
County Level Analysis
As before, we also consider whether there is any evidence of difference in employment
growth between high- and low-wage counties. The results are reported in Table 7. The
minimum wage variables generally have the opposite sign to Table 6 but the standard
errors are so large that they are never significantly different from zero. However, these
results do suggest there is no evidence at county-level that low-wage areas were
particularly badly affected by the minimum wage.
Home Closures and Openings
Thus far we have restricted attention to the balanced panel of continuing homes.
However, the sector does have high turnover of workers and businesses. We have
17
therefore also looked at whether home closure and opening rates are connected to the
incidence of low wage workers in the period before minimum wage introduction.
We have carried out two sets of analysis here. The first looks at home closure and
initial low pay incidence using the home level data. The second compares county-level
closure and entry rates to the low pay structure of the area before the minimum wage
came in.
Table 8 contains the home-level closure equations. A variety of specifications are
reported, using all three minimum wage variables, and either including or excluding the
control variables. A very clear pattern emerges. In none of the specifications is there any
evidence that the homes with more low wage workers shut down because of the
imposition of the minimum wage. Indeed, several of the coefficients are actually
estimated to be negative, rather than the positive that would be predicted if one
hypothesised that the rising wage costs ensuing from the minimum wage would cause
closure.
The second analysis, now at county-level, is considered in Table 9. Here we
report the same kinds of specifications as in Table 8 for county-level closure rates in the
upper panel of the Table and for county-level entry rates in the lower panel. Again, none
of the associations is significant in statistical terms and in both cases both positive and
negative coefficients emerge. It seems that differential entry and exit related to wage
levels are unlikely to have contributed to reducing employment in the care homes sector
over the minimum wage introduction period.
18
So, to summarise, the results point to a very sizable wage impact of the minimum
wage and there is some evidence of an impact on jobs and hours in the firm-level
regressions.
Other Outcomes and the Minimum Wage
This section briefly investigates the impact of the minimum wage on other
outcomes. It is possible that they may have ‘passed on’ increased wage costs from the
minimum wage introduction through higher prices though the extent to which this is
possible may be limited by the extensive price regulation. Second, there may have had to
have been re-organizations that could raise productivity (e.g. quality of care
improvements, or increases in care worker productivity, or simply making people work
harder for their higher wages).
We consider these two possibilities in Table 10. The upper panel reports price
change equations and the lower two panels consider two productivity measures. The first
is changes in residents per worker hour, the second comes from managers responses to a
question about whether they think worker effort went up, stayed the same, or fell in the
period around minimum wage introduction.
Perhaps not surprisingly given the nature of price regulation in the care homes
sector we find no evidence that prices rose by more in the initial low wage firms. Whilst
all the estimated coefficients in the upper panel are positive none of the coefficients
approach anything near statistical significance (all have t-ratios < 1). As such there
seems to be no evidence that minimum wage increases might have been passed on
through higher prices in this sector.
19
There is more evidence in line with the re-organization/productivity improvement
idea considered in the lower panels of Table 9. The estimated coefficients on the
minimum wage variables in both the change in residents per worker hour and subjective
effort change equations are all estimated to be positive. The pattern of significance is,
however, somewhat mixed but the results are nevertheless suggestive that low wage
employers may have been able to sustain the sizable wage increases that occurred by re-
organizations to increase worker effort.
VI. Conclusions
In this paper we present empirical work on the wage and employment consequences of
the recent introduction of a minimum wage to the UK labour market. We focus on a
specific low wage sector, the care homes industry, on which we carried out our own
survey before and after the minimum wage was introduced. We find a very important
wage compression effect on the bottom half of the wage distribution in this low wage
sector. Before its introduction around 40 percent of care homes workers were paid less
than the minimum wage. In April 1999 we see a spike in the wage distribution of around
30 percent at exactly the minimum wage. This resulted in wage growth being
considerably higher in the period surrounding minimum wage workers in homes who had
more low paid workers before the minimum came in. This seems to establish that the
minimum wage had considerable ‘bite’ on average wages and wage structure. Turning to
the employment effects we find some evidence of employment and hours reductions, but
no effect on home closures. Of course, the sector we have examined is particularly
vulnerable to the minimum wage as it has very many low-paid workers and is unable to
20
pass wage costs on into prices. Hence, one should be cautious in drawing conclusions
from this study about the impact of the introduction of the National Minimum Wage on
the UK as a whole.
21
References
Card, David (1992) Using regional variation in wages to measure the effects of the federal minimum wage, Industrial and Labor Relations Review, 46, 22-37. Card, David and Alan Krueger (1995) Myth and Measurement: The New Economics of the Minimum Wage, Princeton: Princeton University Press. Dickens, Richard, Stephen Machin and Alan Manning (1999) The Effects of Minimum Wages on Employment: Theory and Evidence From Britain, Journal of Labor Economics, 17, 1-22. Dickens, Richard and Alan Manning (1995) After Wages Councils, New Economy, 2, 223-27. Dickens, Richard, Stephen Machin, Alan Manning, David Metcalf, Jonathan Wadsworth and Stephen Woodland (1995) Minimum Wages and UK Agriculture, Journal of Agricultural Economics, 49, 1-19. Dolado, Juan, Francis Kramarz, Stephen Machin, Alan Manning, David Margolis and Coen Teulings (1995) 'The Economic Impact of Minimum Wages in Europe', Economic Policy, 23, 317-72. Machin, Stephen, Alan Manning and Stephen Woodland (1993) Are workers paid their marginal product? Evidence from a low wage labour market, Discussion Paper No.93-09, University College London, minor classic that sadly remained unpublished.
22
Table 1: Nature of Survey
All Firms Balanced Panel Balanced Panel (excluding firms with
lots of missing worker info)
Pre-Minimum
Post-Minimum
Pre-Minimum
Post-Minimum
Pre-Minimum
Post-Minimum
Number of Homes 1865 2144 643 643 617 617 Number of Workers 17.23
(12.16) 17.19
(13.02) 16.43 (9.71)
16.70 (11.9)
16.48 (9.67)
16.90 (11.9)
Hourly Wage (not imputed)
4.03 (0.85)
4.28 (0.81)
4.00 (0.80)
4.25 (0.74)
3.97 (0.74)
4.24 (0.72)
Hourly Wage (imputed)
4.01 (0.82)
4.26 (0.75)
3.98 (0.75)
4.23 (0.72)
3.98 (0.74)
4.23 (0.70)
Weekly Hours (not imputed)
25.6 (6.5)
25.2 (6.1)
24.9 (6.3)
24.9 (5.8)
24.7 (5.9)
24.8 (5.7)
Weekly Hours (imputed)
25.8 (6.5)
25.4 (6.0)
25.2 (6.2)
25.2 (5.8)
24.9 (5.9)
25.0 (5.6)
Weekly Earnings (not imputed)
103.9 (38.9)
108.4 (37.8)
100.4 (37.8)
106.8 (35.4)
99.0 (33.7)
105.9 (34.6)
Weekly Earnings (not imputed)
104.2 (37.1)
108.8 (35.9)
100.6 (34.5)
107.3 (34.5)
99.7 (33.6)
106.3 (33.6)
Proportion of workers with missing information
0.10 0.11 0.06 0.04 0.04 0.04
Proportion Female 0.94 0.95 0.95 0.95 0.95 0.95 Average Age 38.2
% Occupancy Rate 0.87 0.89 0.88 0.89 0.88 0.89 Average Weekly Price per Bed (£)
252.4 (86.1)
258.2 (92.2)
250.2 (78.2)
257.6 (79.2)
249.7 (78.2)
257.0 (78.8)
% DSS/Local Authority
0.55 0.58 0.52 0.57 0.52 0.57
Notes: 1) Standard errors in parentheses 2) Pre-Minimum observations refer to responses received before April 1999 and post-minimum to responses received
after March 1999. 3) Care Assistants include senior, day and junior carers but exclude night carers and sleep-ins.
23
Table 2: The “Bite” of the MinimumWage Introduction And Subjective Views on Employment Impact
All Firms Balanced Panel Balanced Panel
(excluding firms with lots of missing worker info)
Pre-Minimum
Post-Minimum
Pre-Minimum
Post-Minimum
Pre-Minimum
Post-Minimum
% Paid Less Than Minimum Wage
32.2 1.0 31.3 0.8 31.7 0.7
% Paid Less Than Adult Minimum Wage
38.2 4.2 37.8 4.3 38.3 4.2
Wage Gap 0.040 0.002 0.041 0.003 0.038 0.002 Adult Wage Gap
0.047 0.006 0.050 0.007 0.048 0.007
% Paid Exactly Minimum Wage
8.7 27.7 9.3 28.3 9.4 28.6
% Paid Exactly Adult Minimum Wage
8.6 30.0 9.0 30.7 9.2 31.0
Notes: 1) Pre-Minimum observations refer to responses received before April 1999 and post-minimum to responses received
after March 1999.
24
Table 3a: Changes in Average Log Hourly Wages
And Initial Period Wage Measures
Dependent Variable: Change in Log Average Hourly Wage
Data Low-Pay Proportion
Wage Gap
Initial Log Wage
Controls R2 Number of Homes
(1) 98/99 0.145 (0.012)
No 0.19 643
(2) 98/99 0.156 (0.022)
Yes 0.36 593
(3) 98/99
0.800 (0.070)
No 0.29 643
(4) 98/99
0.830 (0.087)
Yes 0.44 593
(5) 98/99
-0.360 (0.040)
No 0.29 643
(6) 92/93
-0.174 (0.057)
No 0.07 231
(7) 98/99 0.039 (0.019)
-0.306 (0.061)
No 0.30 643
(8) 98/99
0.567 (0.064)
-0.178 (0.043)
No 0.40 643
(9) 98/99 clean panel
0.905 (0.046)
No 0.36 617
(10) 98/99 clean panel
0.702 (0.058)
-0.126 (0.031)
No 0.39 627
Notes: 1) Sample is balanced panel of homes. 2) Wage-gap is the wagebill shortfall from the minimum wage as a proportion of the total wagebill. 3) Low paid are defined as those aged 18-21 inclusive paid less than £3.00 per hour and those aged 22+ who are paid
less than £3.60 per hour. 4) Control variable are the initial proportion female, proportion with nursing qualification, proportion of care
assistants and average age (all workers), occupancy rate, proportion of local authority/dss residents, county and month dummies.
25
Table 3b: Changes in Average Log Weekly Wages
And Initial Period Wage Measures
Dependent Variable: Change in Log Average Weekly Wage
Data Low-Pay Proportion
Wage Gap
Initial Log Wage
Controls R2 Number of Homes
(1) 98/99 0.137 (0.025)
No 0.04 643
(2) 98/99 0.156 (0.022)
Yes 0.18 593
(3) 98/99
0.669 (0.119)
No 0.07 643
(4) 98/99
0.715 (0.143)
Yes 0.21 593
(5) 98/99
-0.324 (0.031)
No 0.21 643
(6) 92/93
-0.272 (0.055)
No 0.07 231
(7) 98/99 0.034 (0.028)
-0.311 (0.037)
No 0.22 643
(8) 98/99
0.318 (0.082)
-0.296 (0.035)
No 0.23 643
(9) 98/99 clean panel
0.832 (0.134)
No 0.07 617
(10) 98/99 clean panel
0.386 (0.127)
-0.268 (0.032)
No 0.21 627
Notes: 5) Sample is balanced panel of homes. 6) Wage-gap is the wagebill shortfall from the minimum wage as a proportion of the total wagebill. 7) Low paid are defined as those aged 18-21 inclusive paid less than £3.00 per hour and those aged 22+ who are paid
less than £3.60 per hour. 8) Control variable are the initial proportion female, proportion with nursing qualification, proportion of care
assistants and average age (all workers), occupancy rate, proportion of local authority/dss residents, county and month dummies.
26
Table 4: County-Level Analysis of Changes in Average Log Wages
And Initial Period Wage Measures, Care Assistants, Aggregated Care Homes Data 1998/99 and Labour Force Survey 1998/99
Dependent Variable: Change in Average Log Hourly Wage (1) (2) (3) (4) (5) (6) Care Homes
1998/99 Data Labour Force Survey 1998/99
Data Initial Proportion Paid Less Than Minimum Wage
0.215 (0.044)
0.410 (0.153)
Initial Wage Gap 1.308 (0.309)
1.609 (0.797)
Initial Log Hourly Wage -0.419 (0.084)
-0.449 (0.184)
R-squared 0.38 0.33 0.43 0.08 0.06 0.08 Number of Counties 55 55 55 55 55 55
Dependent Variable: Change in Average Log Weekly Wage
(7) (8) (9) (10) (11) (12)
Care Homes 1998/99 Data
Labour Force Survey 1998/99 Data
Initial Proportion Paid Less Than Minimum Wage
0.187 (0.088)
0.546 (0.205)
Initial Wage Gap 1.114 (0.497)
2.618 (0.847)
Initial Log Hourly Wage -0.385 (0.098)
-0.702 (0.204)
R-squared 0.10 0.09 0.22 0.09 0.11 0.22 Number of Counties 55 55 55 55 55 55 Notes: 1) Weekly gap is the wagebill shortfall from the minimum wage as a proportion of the total wagebill 2) Low paid are defined as those aged 18-21 inclusive paid less than £3.00 per hour and those aged 22+ who are paid
less than £3.60 per hour. 3) Robust standard errors in parentheses.
27
Table 5: Changes in Log Employment and Hours And Initial Period Wage Measures, All Workers,
Care Homes Data 1998/99 and 1992/93
Change in Log Number
Employed (1) (2) 1998/99
Data 1992/93
Data Initial Log Hourly Wage 0.108
(0.082) -0.133 (0.147)
R-squared 0.002 0.002 Observations 643 237 Change in Log Total Hours (3) (4) 1998/99
Data 1992/93
Data Initial Log Hourly Wage 0.152
(0.093) -0.220 (0.165)
R-squared 0.004 0.009 Observations 643 231 Notes: 1) Robust standard errors in parentheses.
28
Table 6: Changes in Log Employment and Hours And Initial Period Minimum Wage Measures, All Workers,
Care Homes Data 1998/99 Dependent Variable: Change in Log Number Employed
R-squared 0.000 0.050 0.000 0.050 0.001 0.002 0.032 0.052 Number of Counties 55 55 55 55 55 55 55 55 Notes: 1) Demographic variables: changes in proportion female, proportion with nursing qualification and average age (care
assistants). 2) Weekly gap is the sum of the wagebill shortfall from the minimum wage as a proportion of the total wagebill 3) Hourly gap is the sum of the hourly shortfall from the minimum wage as a proportion of the total hourly wagebill 4) Low paid are defined as those aged 18-21 inclusive paid less than £3.00 per hour and those aged 22+ who are paid
less than £3.60 per hour. 5) All regressions control for county-level growth in male employment for male employees aged 25-64. 6) Standard errors in parentheses.
30
Table 8: Home Closures and the Minimum Wage Home Closures (1) (2) (3) (4) Initial Proportion Paid Less Than Minimum Wage
0.017 (0.178) [0.004]
0.044 (0.222) [0.009]
Initial Wage Gap
1.381 (0.889) [ 0.294]
1.473 (0.994) [0.292]
Demographic Varaibles Yes Yes Home Characteristics Variables Yes Yes Response Month Dummies Yes Yes Log-Likelihood -266.05 -237.36 -264.86 -236.29 Number of Homes 682 647 682 647 Number of Closures 90 85 90 85 Notes: 1) Sample is drawn from balanced panel including crossover homes and 105 closed homes who replied in the pre-
minimum period. 2) Probit coefficient estimates, standard errors in parentheses 3) Marginal effects in square brackets 4) Demographic variables: changes in proportion female, proportion with nursing qualification, proportion of care
assistants and average age (all workers). 5) Home characteristics variables: changes in occupancy rate and proportion of local authority/dss residents. 6) Wage gap is the sum of the wagebill shortfall from the minimum wage as a proportion of the total wagebill. 7) Low paid are defined as those aged 18-21 inclusive paid less than £3.00 per hour and those aged 22+ who are paid
less than £3.60 per hour.
31
Table 9: County-level Closures and Entrants and the Minimum Wage County Closure Rate (1) (2) (3) (4) Initial Proportion Paid Less Than Minimum Wage
0.074 (0.242) [0.007]
0.285 (0.268) [0.025]
Initial Wage Gap
0.458 (1.347) [0.040]
1.490 (1.430) [0.132]
Demographic Variables Yes Yes Home Characteristics Variables Yes Yes R-Squared 0.03 0.18 0.03 0.18 Number of Counties 55 55 55 55 County Entry Rate (5) (6) (7) (8) Initial Proportion Paid Less Than Minimum Wage
-0.418 (0.430) [-0.022]
-0.265 (0.463) [0.014]
Initial Wage Gap
-1.080 (2.409) [-0.056]
0.254 (2.481) [0.013]
Demographic Variables Yes Yes Home Characteristics Variables Yes Yes R-Squared 0.02 0.22 0.00 0.21 Number of Counties 55 55 55 55 Notes: 1) Sample is county level panel data compiled from the unbalanced panel of care homes. 2) Closure rate refers to proportion of estimated home closures of all homes by county; dependent variable is the
logistic transformation log [p(1-p)]. 3) Entrant rate refers to proportion of estimated new homes of all homes by county; dependent variable is log
((p+1/(2n))/(1-p+1/(2n))), with the correction included due to zero entry rates in 2 counties. 4) Demographic variables: changes in proportion female, proportion with nursing qualification, proportion of care
assistants and average age (all workers). 5) Firm characteristics variables: changes in occupancy rate and proportion of local authority/dss residents. 6) Wage gap is the sum of the wagebill shortfall from the minimum wage as a proportion of the total wagebill. 7) Low paid are defined as those aged 18-21 inclusive paid less than £3.00 per hour and those aged 22+ who are paid
less than £3.60 per hour. 8) All regressions control for county-level growth in male employment for male employees aged 25-64. 9) Standard errors in parentheses. 10) Marginal effects in square brackets.
32
Table 10: Prices, Productivity and the Minimum Wage
Change in Log Average Price (1) (2) (3) (4) Initial Proportion Paid Less Than Minimum Wage -0.001
(0.024) -0.013 (0.027)
Initial Wage Gap
0.116 (0.145)
0.048 (0.155)
Demographic Variables Yes Yes Home Characteristics Variables Yes Yes Response Month Dummies Yes Yes R-Squared 0.00 0.11 0.00 0.11 Number of Homes 572 501 572 501 Change in Log Residents Per Worker Hour (5) (6) (7) (8) Initial Proportion Paid Less Than Minimum Wage 0.103
(0.044) 0.068
(0.047) Initial Wage Gap
0.334
(0.275) 0.111
(0.273) Demographic Variables Yes Yes Home Characteristics Variables Yes Yes Response Month Dummies Yes Yes R-Squared 0.01 0.21 0.00 0.20 Number of Homes 586 514 586 514 Subjective Responses on Change in Worker Effort (9) (10) (11) (12) Initial Proportion Paid Less Than Minimum Wage 0.063
(0.183) 0.100
(0.201) Initial Wage Gap
0.530
(1.085) 0.922
(1.166) Demographic Variables Yes Yes Home Characteristics Variables Yes Yes Response Month Dummies Yes Yes Log-Likelihood -245.24 -210.91 -245.18 -210.72 Number of Homes 561 486 561 486 Notes: 1) As for Table 6. 2) Effort variable coded as an ordered response based on answers to the question “Has the minimum wage had an
impact on work effort in your business? No/Yes – Decrease/Yes-Increase”. It is ordered from 0 (decrease), 1 (no change), to 2 (increase). Ordered probit coefficients (and associated standard errors) reported for this variable.
33
Figure 1a The Relationship between Wage Growth and Initial Wages: 1998/99
Figure 1b The Relationship between Wage Growth and Initial Wages: 1992/93
34
Figure 2a The Relationship between the Change in Log Total Employment and Initial Proportion Low-Paid
Figure 2b The Relationship between the Change in Log Total Employment and Initial Wage Gap
35
Figure 3a The Relationship between the Change in Log Total Hours and Initial Proportion Low-Paid
Figure 3b The Relationship between the Change in Log Total Hours and Initial Wage Gap
