About us

Leading international thinktank and political network


Register for all the latest updates in our regular newsletter

Home Opinion Turning over the 'hourglass' labour market argument
Equality • Women • Pay gap

Turning over the 'hourglass' labour market argument

Craig Holmes - 27 November 2014

Women and men at the bottom of the labour market have both fared poorly since the recession. But even a closing of the pay gap for higher-paid women is not quite all it seems

Since the onset of the 2008 recession, labour market experiences have differed for men and women in the UK. The initial loss of jobs was felt far more heavily amongst men, while job losses resulting from austerity measures had a greater impact on the female workers. The initial fall in total employment up to 2010 resulted in more female self-employed workers and less male self-employed workers. As total employment has subsequently risen, female self-employment has continued to grow steadily while male self-employment has increased more markedly. Part- time employment has also risen for both groups over the whole time period. During the first few years, the much larger share of women in part-time jobs fell, while some of the fall in full-time employment for men was offset by moving to part-time work. However, part-time employment has increased for men and women as the proportion of adults in work has risen, meaning that underlying the decrease in measured unemployment or inactivity is a fall in hours and job quality. These outcomes, which are well-known, are summarised below:

A report by the Fawcett Society has raised concerns that many women are finding work in low-paid or insecure employment, and increasingly in jobs that use zero-hours contracts. Table 2 suggests that such developments have largely been consistent with pre-recession inequality. Male workers have experienced a slightly larger increase in low paid work, while female workers have been more marginally more affected by the increasing use of zero-hours contracts.

The final well-established trend relates to pay. Obviously, more people in insecure or part-time employment means that weekly incomes will fall due to reduced hours. At the same time, average hourly pay rates have declined as nominal wage growth has slowed below inflation. Figure 1 shows real median hourly pay for men and women since 2006. Since the start of the period, average wages have fallen by 8.7 per cent and 5.8 per cent for men and women respectively.

There are, however, two further issues relating to the quality of employment quality and earnings for male and female workers since the start of the crisis which are worth examining. These are, first, the phenomenon known as polarisation, hollowing out, or the ‘hourglass’ labour market, and, second, the manner in which these occupational changes have had an impact on the distribution of male and female wages, and how these changes compare to other factors that have shaped pay.

The hollowing out of employment

The polarisation hypothesis advanced by Maarten Goos and Alan Manning suggests that the past four decades have seen a growth of ‘lovely’ and ‘lousy’ jobs and a decline in the middle of the distribution, which has typically covered jobs that are more easily replaced by technology such as semi-skilled manual occupations and clerical workers. One question is whether this trend has continued throughout the downturn. Figure 2 shows employment share changes by occupation and gender, where each occupation is ranked by its mean hourly wage in 2006, for women and men respectively. A quadratic best-fit line indicates the overall trends.

For men, polarisation appears to have continued – the best-fit line illustrates the increasing employment shares in the highest-paying and lowest-paying occupations, and declines around the middle. For women, however, the relationship is much closer to linear rather than u-shaped. Employment shares have generally fallen in the lowest-paying occupations and average-paying occupations, and have grown for those jobs that have typically experienced higher wages such as managerial positions and the professions. What we likely see here is the net effect of two ongoing trends: a decline in middle-skilled jobs in line with the idea of hollowing out, as well as a narrowing of gender gaps in access to the best jobs between successive cohorts.

The distribution of wages

Figure 3 shows hourly wage growth for men and women at different points of their respective distributions – the axis along the bottom goes from worst paid to best paid (note that a position on the distribution for a male workers represents a higher initial wage than it does at the same position for women). The figure reveals that the lower average wage fall for women over this time period is not evenly distributed: for men and women in the bottom half of their distributions, real wage falls were similar. The lowest-paid men experienced the largest real wage decrease, although compared to the lowest-paid women, there was further to fall before hitting the lower limit created by the national minimum wage. The bottom half of both distributions have become more spread out, which implies an increase the incidence of low pay.

Women in the upper half of the distribution, on the other hand, have done relatively well compared to higher earning males. Wage dispersion has been minimal for the top 40 per cent (with only the very top pulling away at all in relative terms). For men, there is a four percentage point difference between the fall in real wages for workers around the middle and the worker who is just inside the top decile, compared to a less than two percentage point increase for women in the same part of their respective distributions. As indicated by the average wages discussed previously, gender pay gaps have fallen for all women, but this is clearly strongest for the better paid.

Everything else being equal, hollowing out – as experienced by male workers – should increase wage inequality (as more people move to the tails of the distribution) while upgrading (that is, a general move towards better jobs) has more ambiguous effects as some workers decrease their relative earnings compared to the top but increase them compared to the worst paid. Figure 4 shows the effect of the change in the employment share of seven categories of occupation (reflecting broad groups of low-, middle- and higher-skilled jobs) on the distribution of male and female wages, if we could hold everything else the same as it was in 2006. Figures 4 and 5 come from a decomposition technique that has been applied to earlier time periods in the UK.

For women, the relationship between the two lines is much closer than it is for men, meaning that occupational composition is a better predictor of the overall change in wages for women than it is for men. This means that factors other than occupational composition have had more effect on the distribution of male earnings. These factors include other ways in which the composition of the workforce has changed (such as increasing levels of education and falling union membership) as well as the way wages respond to the composition of the workforce (for example, the premium received for, or return on, having a degree or particular useful skills). Figure 5 shows the sum total of the first group of effects – labelled ‘composition’ effects – and the second group of effects – labelled ‘wage returns’.

The figure illustrates two important differences for men and women over this period of time. First, if nothing had changed about the way factors like education, occupation and other relevant characteristics corresponded to wages, female earnings dispersion would have increased more as the composition of those factors changed. This would have meant that the incidence of low pay would have increased by more than it ultimately did, as shown in the upper half of Table 3.

Second, the wage-return effects have acted in opposite directions. For women, the dispersion of pay has been pulled back by these returns, implying a larger fall or smaller rise in the premium attached to education, experience, use of skills and other wage-enhancing characteristics for those with more of these factors. On the other hand, their male equivalents (i.e. the ones with the most education and skills in the jobs that make most use of these) have been relatively better rewarded and pulled away from the rest, including those around or just above the middle. The combination of all of these changes has led to two outcomes. First, women are slightly better represented in higher-wage jobs than they were in 2006 (as indicated by the bottom half of table 3). Despite this, gender pay gaps at the top have risen, which raises the question as to why the increasing numbers of women in good jobs with high levels of education have not experienced the same opportunities that have existed for their male counterparts.

As we have seen, in the period since the recession the male labour force has experienced more hollowing out of middle-skilled jobs, while female workers have moved up the job distribution during this time. Consequently, female pay in the upper half of the distribution has grown faster than for men, reducing (but not eradicating) pay gaps. Nonetheless, the top male workers have experienced the highest growth in pay relative to all workers, even compared to the women in the labour market with the highest levels of education in the most desirable occupational groups, which suggests that women continue to face obstacles at the top.

At the bottom end, the experience of men and women has been largely similar. Female workers began the period at a disadvantage in terms of both the absolute levels of pay and their exposure to low-paying jobs. Over this period, male wages at the low end have fallen faster (again, a consequence of them having further to fall before they reaching the floor imposed by the minimum wage) and their exposure to low-paying jobs has increased by more. Obviously, this narrowing of female disadvantage is nothing to celebrate, as both men and women at the lower end have experienced marked declines in wages, job security and the quality of employment. That the largest decline in wages has been experienced right at the bottom stands in contrast to the experience of the previous decade up to the start of the recession, where it was the lowest paid who experienced the highest wage growth. Real increases in the minimum wage played an important role during this period by protecting workers from a number of other labour market changes which would have left them far worse off than the average worker, such as the hollowing out of occupations and the decline of union bargaining power) Conversely, allowing the real value of the minimum wage to fall since 2006 has had the opposite effect on the lowest paid.

Craig Holmes is a teaching fellow in economics at Pembroke College, Oxford University and a research fellow of SKOPE at the Oxford University Department of Education.

This is a contribution to Policy Network's work on Social Policy and Changing Welfare States.

Tags: equality , recession , labour market , pay gap

Add comment


Enter the code shown:

The Policy Network Observatory promotes critical debate and reflection on progressive politics. It is centre-left orientated but determinedly challenges social democracy. It is pro-European but restlessly questions EU institutions and practices.

Most read this month

Search Posts

search form
  • Keyword
  • Title
  • Author
  • Date posted