Mobility and the squeezed middle


A common criticism of the recent concern over rising inequality is that it looks at the distribution of earnings and wealth at one point in time. Individuals are mobile and tend, on balance, to find better paying jobs over the course of their lifetime, which might mean that looking at their wages at any one point overstates the differences in standards of living taken from a whole life perspective. This point was emphasised in a recent report by the Social Market Foundation(SMF). In its analysis of earnings outcomes, the report found that of all the households who were in the middle quintile of wages in 2007/8, 41 per cent had moved upwards to a higher quintile of the distribution four years later. Following on from this, they ask whether we have become too concerned about the ‘squeezed middle’.

It is always important to work out what a particular piece of evidence shows, and what it doesn’t show.  As Gavin Kelly argues in an earlier blog post, this analysis emphasises the point that there is always a range of different experiences that sit below aggregate measures. In this case, many of these experiences appear to be far better than has been widely accepted. However, there are some caveats even on that point. For example, as Kelly notes, the measure is a relative one, which means that for any individual that moves up, somebody has to move down. Therefore, in a labour market with falling real wages across the distribution, any upward gains will be more than offset by downward moves in absolute terms.

One thing that we don’t know from these data is whether the recession has been a period of particularly high upward mobility. A quick review of existing comparable studies suggests that the 41% upward mobility figure isn’t particularly high, and might be a little lower than previous estimates. Jenkins and Jarvis (1998) use earlier versions of the same dataset used by the SMF (British Household Panel Survey, 1991-1994) to calculate the likelihood of individuals moving between deciles of the distribution for a shorter two-year period. They find that 30 per cent of those in the middle quintile moved upwards between 1991/2 and 1993/4 (Table 1, below, reduces their decile transition matrix into one using quintile transitions). Chen (2009) shows that around 38 to 40 per cent of all individuals move upwards by at least one decile in various four year periods from the BHPS (starting with 1991-1995 and finishing with 1997-2001). Although this isn’t calculated for each quintile separately, we can see from Table 1 that historically the middle group tends to be the most upwardly mobile. High income groups have little room to move upwards, while low income groups get trapped in low-end jobs, despite the greater scope for upward mobility. This implies that mobility would be higher than 40 per cent for the middle group.

Source: Jenkins and Jarvis (1998), Table 2; own calculations

Table 1 also illustrates some other interesting comparisons. The SMF report shows that 10 per cent of the middle income quintile in 2011/12 came from the bottom quintile four years earlier. This number is relatively comparable to the earlier data, where 5 per cent made such a move in half the time. On the other hand, mobility from quintile 1, 2 and 4 towards the middle group was 9, 19 and 22 per cent respectively in the SMF analysis. These proportions are almost identical to those found for the shorter time period shown in Table 1, which suggests mobility has fallen.

One thing that studying mobility patterns over a period of time can’t do is say anything about the experience of new entrants. During a four year period, a significant proportion of the working population will leave the labour market and be replaced by school and university leavers, those returning to work from a spell of unemployment or inactivity, or migrants. For this group, the snapshot measure of their earnings matters for a number of reasons. The living standards of new labour market entrants are more directly affected by their current real income than those of older workers. They have fewer savings (or more debts) to smooth out short-term losses of earnings, and are less likely to be living with a partner and sharing living costs. Secondly, their early labour market experience might influence their future living standards. It could be argued that as growth returns and real wages start to rise again, many of these individuals will move up the distribution as earlier generations have. However, there is evidence from the US andCanada that entering the labour market during a recession, even with high levels of qualifications, has a persistent negative effect on earnings which means they have less upward wage mobility that other cohorts have typically experienced.

This raises a second point, which brings us back to the variety of experiences that lie behind aggregate statistics discussed earlier. The finding that that 41 per cent of the middle move up the wage distribution may indicate that in the four year period from 2007/8 to 2011/12, all individuals had the same opportunities to move upwards. But in reality this statistic likely masks significant inequalities of opportunity between different types of individual and household in the middle group. Regional, educational and family background differences play a role in shaping mobility opportunities and complicate how we go from thinking about aggregate statistics to understanding the “swirling flux” of individual experiences.

Craig Holmes is a lecturer in Economics at Pembroke College, Oxford and research fellow at SKOPE . He tweets as @CraigPHolmes.