This paper is the fifth report for the Intergenerational Commission, which was launched in the summer of 2016 to explore questions of intergenerational fairness that are currently rising up the agenda and make recommendations for repairing the intergenerational social contract. As previous analysis has highlighted, these questions are gaining prominence because there is evidence of a generational living standards challenge in a range of areas. From experiences in the labour market, to asset-building in the form of a house and a decent pension, to what can be expected from the welfare state, the social contract between the generations shows signs of fraying.
Jobs and pay clearly play a central and interconnected role in this wider question of intergenerational living standards. The labour market is the major source of household incomes over working lives (especially so with a working-age welfare state currently in retreat), the vehicle through which saving for retirement takes place, and the main means (outside of the family) of putting together a deposit on a home. On this basis, the finding that millennials who have entered work so far have made no earnings progress on generation X before them is a cause for concern. Of course, much of this will relate to the fact that their labour market experience has so far been dominated by the downturn, but there may be more structural factors at play here too.
In trying to assess how the lifetime incomes of younger (and future) cohorts – and their prospects for retirement – will compare with their parents’ and grandparents’, it is therefore their potential employment and earnings paths that are likely to matter most. The best source for clues as to what the future holds is usually a detailed understanding of experiences to date. As such, this paper explores the structural, compositional and dynamic factors that have driven cohort earnings trends, and in particular the dominance or not of the downturn in explaining what has happened. We track trends over the past couple of decades, and while our focus is on the beginning of careers, we capture the whole of working lives.
Setting aside the UK’s record employment performance and focusing on pay trends, as this report does, the direction of travel over the past 15 years sets the context for faltering outcomes across generations.
Pay growth stagnated in the mid-2000s due to the wedge that opened up between pay and productivity growth, an experience that was followed by the 2009-2014 period of falling pay after the financial crisis. This recent history means that the promise that successive cohorts earn significantly more than those who came before at the same age is under threat, with most five-year birth cohorts currently earning fairly similar wages to those born around 10 years before them.
But these effects are starkest for younger cohorts. Stagnating and then falling pay hit them in the early part of their careers or just as they were entering the world of work, when pay improvements are usually most rapid. As such, the oldest millennials (born 1981-85) are earning £40 per week less around the age of 30 than those born 10 years earlier earned at the same age. And the next group of millennials (born 1986-90) have had the same levels of weekly pay in their early- and mid-20s as those born 15 years before them did.
This report explores the population and jobs market changes that underpin these faltering cohort-on-cohort earnings improvements, and the mechanisms via which year-on-year increases in cohorts’ pay have slowed down.
While much of the story on cohort wages in inextricably intertwined with macro-economic developments in recent decades – and in particular the impact of the downturn – some trends are relatively distinct. For example, we know that population characteristics have changed over time in various ways. For example, the UK has experienced profound changes in both educational attainment and migration levels. So when comparing the pay of cohorts at the same age we may not be comparing like with like.
A detailed look at these ‘compositional’ shifts in individual characteristics reveals:
‘Decomposition’ techniques allow us to look at these changes in combination (along with changes in where cohorts live and their age composition) and estimate their individual and combined impact on the real pay of each five-year birth cohort during 2007-2015 compared to those at the same age 10 years before.
We find that across cohorts, rising educational attainment dominates and has boosted cohort-on-cohort pay changes across the board. Country of birth has dragged on pay slightly for younger cohorts, and rising female participation for older cohorts, but these effects are much smaller. The dominant role of qualifications means that the total compositional effect from changing individual characteristics is positive for all cohorts. In other words, all else equal population changes would be expected to boost pay for each cohort compared to those at that age 10 years prior.
Importantly, though, the size of this effect is not the same for all cohorts. In particular, the ‘qualifications boost’ for the millennial cohorts born since 1981 is less than half its average size for older cohorts. This may partly reflect the fact that the boost is always higher within older cohorts, for example because graduate wages progress more rapidly as they age. But it is also likely to be linked to declining relative increases in educational attainment over time. The implication is that rapid and consistent improvements in educational attainment have previously delivered a large compositional boost to the pay of each cohort compared to those before, and that a new approach to boosting the quantity and quality of qualifications may be needed to make the same gains in future.
As well as the broad population trends discussed above, the type of jobs on offer in the UK labour market have been changing. Two trends stand out in current debates, and merit consideration from a generational perspective:
Overall there is some evidence that non-standard and potentially more insecure employment forms are more common for younger cohorts compared to their predecessors at the same age, but evidence on the combined scale of these shifts and their impact on pay trends is not conclusive.
The same decomposition techniques as used to understand the impact of changing individual characteristics between cohorts allow us to summarise these trends. We find that shifting job characteristics – predominantly from changing occupational shares and part-time working patterns – have boosted mean pay for baby boomer (born 1946-65) and generation X (born 1966-80) cohorts compared to the pay of those at the same age 10 years before.
But changing job characteristics have put downward pressure on cohort-on-cohort wage improvements for older millennials (born 1981-90). In other words, the combination of a shift towards part-time working in the 20s and some downward pressure from occupational changes means that those born in the 1980s are now doing lower-paying jobs than cohorts at the same age were 10 years before them.
Combining thee separate people- and job-specific decomposition analyses, we can summarise compositional effects on the pay of different cohorts during 2007-2015 compared to that of cohorts at the same age 10 years before:
These findings for younger cohorts are profound. This is because a compositional pay boost across cohorts and over time should be considered the norm given overarching qualifications patterns and occupational shifts. A zero or negative compositional contribution to cohort-upon-cohort wage improvements for those born in the 1980s – compared to continuing compositional boosts for older cohorts – looks likely to provide part of the explanation for why pay has fallen back most for younger cohorts in recent years. This makes clear that ascribing all changes in cohort earnings patterns to the financial crisis and the pay squeeze that followed is a limited and inaccurate reading of generational earnings trends.
These compositional shifts don’t tell us the whole story, however. For example, we find basically no compositional effect on the pay of the 1986-90 cohort compared to the pay of those 10 years before them, but we know that they currently earn around £40 per week less. To complete the picture, we also need to consider the changing ‘wage return’ recorded by cohorts after holding compositional factors constant. Such returns apply both to cohorts as a whole and to groups with specific characteristics (such as level of educational attainment or country of birth) within them.
In general, the generational pay picture for the various groups we look at is similar to the overall picture we have described. For example, graduates and non-graduates, and full- and part-time workers in successive generations have each experienced similar wage declines compared to their counterparts in previous generations.
Overall, and fitting the picture we described at the outset of larger cohort-on-cohort pay differences for younger cohorts, the wage returns effect increases gradually with age: it is negative for the cohorts born in the 1970s and 1980s when we compare their pay to those 10 years before, but positive for older cohorts.
Looking at the interaction of compositional and wage return factors, we can split cohorts into three groups:
We have so far discussed changing individual and job characteristics between cohorts and the wage levels associated with these – what might be called a ‘static’ comparison of two points in time. But cohorts are broadly made up of the same people year-on-year, meaning that a picture of how individuals progress in the labour market – a ‘dynamic’ perspective – is also essential for assessing how cohorts have got to where they are today and what the future might hold.
The gains brought by each year-on-year pay increase are of course dependent on what the starting point is: if people begin their careers on much lower wages than they did in the past then future wages will be lower too even if annual increases remain strong. We find that this is exactly what happened for the millennial cohorts born in the 1990s. While starting wages were rising or flat for the 1970s and 1980s cohorts, the 1993-95 cohort had a starting wage (at ages 17-20) that was 25-30 per cent (or £40-£50 per week) lower than the cohort nine years before them. Given this cohort entered the jobs market during 2010-2015, the impact of the downturn here is clear. The open question is the extent to which this lower starting point will feed through to where earnings end up over the course of careers.
Moving forwards in careers from each individual’s (and cohort’s) starting wage, the strongest tool for changing one’s rate of pay is moving from one job to another. This is particularly true when young, when the pay increases associated with moving jobs are much greater, for example the typical real pay rise for someone moving jobs in their mid-20s is around 15 per cent.
In this light, it is concerning that job mobility has fallen across the board, and fallen particularly fast for younger cohorts: millennials so far have been about 30 per cent less likely to move jobs in their 20s than generation X before them. And looking at more narrowly-defined birth cohorts, just 1-in-25 people born in the mid-1980s moved jobs from year-to-year when they were in their mid-20s – half the rate for those who were born a decade before them.
Job moves have also fallen at older ages, but because they are much less common when older and bring a smaller typical pay increase, the declining mobility of younger cohorts appears most damaging for their pay progress.
Given that this decline in mobility began in the early 2000s, and that the move rate remains below where it should be relative to the strength of the labour market, more structural factors than the impact of the crisis are at play here.
And this fall in the job move rate looks particularly concerning when we consider that it may be having wider effects. For example, it may be keeping employees at longer tenures where annual pay rises are lower. Or a low move rate may prevent ‘knock-on’ wage effects on other staff across the age range as a result of departing employees prompting firms to rethink their pay offer to other staff to prevent further losses.
The flipside of falling job mobility – and small reductions in the likelihood of exiting and entering the jobs market at a given age compared to previous cohorts – is an increase in tenure with employers. In particular, longer tenures with firms – spells of five years or more – have increased. For example, around the age of 30 43 per cent of the 1971-75 cohort had been with their employer for five years or longer, a figure which has risen to 47 per for the 1981-85 cohort a decade later.
Crucially, at the same time as tenure has risen, the annual real pay increases that employers offer to their long-serving staff has fallen. For example, referring to the same two cohorts around the age of 30 mentioned above, the typical annual real pay increase for employees who have been with their firms for five years or more has fallen from a healthy 4 per cent for the early 1970s cohort to close to zero today for the early 1980s cohort. Given that many more employees stay in jobs each year than move, even when young, this is a profound shift in our labour market, and one that appears to be enduring beyond the financial crisis at least to some extent.
These trends in job moves and tenure can help us answer the crucial question of why the wages of all cohorts, but particularly younger ones, appear to be improving more slowly year-on-year than they did in the past. Our modelling determines the role that moving jobs, remaining with a firm and exiting or entering work – and the pay rises associated with them – play in explaining changes in progression rates between two cohorts at the same age.
We find that a decline in the pay rises associated with staying with an employer over a year, particularly at longer tenures, is the dominant factor in explaining the slowdown in cohort progression. Declining tenure returns account for on average around four fifths of the reduction in annual cohort progression rates across the age range over the past two decades.
For younger employees only, the falling likelihood of job moves and a reduction in the pay increases associated with them have increased the slowdown in their progression rates, on top of this tenure effect that has been felt by everyone.
From a cohort perspective, this means that those born between the late 1970s and late 1980s – feeling the (connected) effects of less of a pay boost from moving jobs, a lower likelihood of moving jobs, and lower pay rises when they stay with their firm – have experienced a larger slowdown in annual progression rates than older cohorts born before this.
To characterise this, we can imagine how much higher different cohorts’ wages would be if progression rates had not deteriorated over the last 15 years, but rather reflected the performance at each age around the turn of the millennium. In this hypothetical scenario, we find that the 1963-65 cohort would have an average wage 40 per cent higher than it does today, and the 1981-83 cohort 69 per cent higher. Both this millennial cohort and this baby boomer cohort have clearly felt the effects of the cyclical and structural labour market shifts of the past two decades. However, the younger of these two cohorts, at the formative stage of careers, has felt these effects more strongly.
This report lifts the lid on the drivers underpinning slowing cohort wage progression across generations – but particularly for millennials – in recent decades. Much of what has changed is connected to the financial crisis and the pay squeeze that followed it. But as our analysis sets out, a broad range of factors – including growth rates in educational attainment, the impact of changing occupational structures, the rise of atypical working, a structural decline in job mobility, and the enduring impact of lower pay rises when employees stay with firms for long periods – have contributed to stagnating pay growth between cohorts and slower progression rates within them.
Far from putting cohorts’ recent pay experiences entirely down to an ‘accident of history’ – the bad luck of experiencing a large pay squeeze, particularly in the formative stage of careers – there is plenty to consider in terms of how these outcomes can be prevented and unwound in future. A range of policy areas – including further and higher education policy; lifelong learning and human capital development by employers; the interaction between pension contributions and pay; the role of employment law and labour market regulation; and investment and productivity initiatives – are relevant in this regard.
Given the importance of earnings trajectories to lifetime incomes for each generation, the key questions are how we can get current and future cohorts back onto the pay progression tracks of old, and how we can ensure that the promise of generation-upon-generation pay progress is kept in decades to come. To this end, The Intergenerational Commission will continue to develop its understanding of different cohorts’ experience in the labour market and consider what interventions might be warranted as part of a renewal of the intergenerational social contract.