18 March 2011

Social Mobility and the Conscience of the Rich

Most of us do not make considerable fortunes – so who are we to criticise those who do or how they choose to spend their largesse? If rich men want to buy Picassos and lend them to the Tate, surely it’s up to them? Of course, there is a long history of charitable donations by the wealthy from Andrew Carnegie to Bill Gates, and recently two Britons who made fortunes in the financial services boom have set up charitable foundations to examine issues relating to social mobility.

The Sutton Trust was founded in 1997 by Sir Peter Lampl after a career in private equity in the US. “The main objective of the Sutton Trust is to improve educational opportunities for young people from non-privileged backgrounds and increase social mobility”. Clive Cowdery, who made a fortune in the insurance industry, in 2005 set up the Resolution Foundation, “an independent, not-for-profit research and policy organisation dedicated to improving the well-being of low-to-middle earners (LMEs) in today’s mixed economy”.

The Trust and the Foundation are registered charities with a similar modus operandi. The Sutton Trust sees itself as:
a 'do-tank' which not only undertakes research and advocates policy change, but also funds, develops and tests innovative practical solutions to educational inequality. The Trust is now focusing on research and policy work.
The Resolution Foundation undertakes:
original research and economic analysis to understand the challenges facing LMEs today. We develop practical and effective policy proposals to tackle the issues we identify, and we engage with policy makers and other key stakeholders to influence decision making and bring about change.
Both organisations employ “bright young things” to get the work done, “clever old things” as trustees, and are chaired by their founders. Both get frequent media coverage of their activities, the longer established Sutton Trust probably being better-known, but the Resolution Foundation recently sponsored a high-profile event with Ed Miliband as a speaker. Both undertake research relating to social mobility, the Sutton Trust, because of its interest in the inherently long-term benefits of education, tends to look at inter-generational issues. Resolution has recently published a report, “Moving on up? Social mobility in the 1990s and 2000s”. It compared how intra-generationally mobile people a group of people in their 30s had been in the 1990s with another younger group’s experience in their 30s during the 2000s.

The Resolution report received good press coverage, eg in the Financial Times on 11 March 2011 under the headline “Social mobility ‘increased 22% during 2000s’”:
Social mobility – at least as measured by earnings – increased during the past decade, according to a study that challenges the widely held view that social mobility is static or even in decline. Research focused on the earnings of people now in their thirties and early forties shows the chances of making a significant move up or down the earnings ladder remain pretty low.
But the work, by the Resolution Foundation, calculates there was a 22 per cent increase in the probability of moving significantly up the earnings distribution table in the 2000s compared with the 1990s. Back then, less than 3 per cent of people who started the decade in the bottom 20 per cent of earnings made it into the top 20 per cent by the end of the decade. In the 2000s that percentage remained low but had almost doubled to nearly 6 per cent.
Social mobility is a politically-loaded topic. Comparisons between the 1990s and the 2000s are to a large extent between the periods of Conservative and New Labour government, so evidence of improved social mobility after 2000 reflects well on Labour. Not that any party wants to neglect the issue. The Coalition Deputy Prime Minister, Nick Clegg, gave a speech on social mobility in August 2010 in which he said “our long-term social policy goal is social mobility” (though this was primarily directed at inter-generational mobility). It wasn’t surprising, then, that two members of the UK twitterati with long political antennae, Paul Waugh and John Rentoul, picked up on reports of the Resolution study, asking “What does this mean for the Coalition's narrative?” and “What does this mean for the media narrative?”, respectively. This post attempts to answer their questions – and fails.

The devil always lies in the detail, so the starting point in coming up with an answer was the Resolution report, in particular Section 3, which explains that the data for the older group came from the National Child Development Study (NCDS) and for the younger group from the British Cohort Study (BCS). NCDS attempts to follow through their lives all the people born in one week in England, Scotland and Wales in 1958. The second has aimed to follow all the children in England, Scotland and Wales born in one week in 1970. Among the data collected by NCDS and BCS from the individuals who make up these “cohorts” has been information about incomes. This provided the basis of the Resolution study which used changes in earnings as the measure of mobility. Table 1 of the report (page 9, reproduced below) summarises the data sets that were used.
It is important at this point to realise that “Sample size” here refers to the number of individuals whose data was selected by Resolution from the respective NCDS and BCS cohorts. How and why this was done is explained in Appendix A and attention should be paid to Tables A1 and A2. In essence, individuals who had negative or zero earnings and individuals who were self-employed (“due to the unreliability of earnings data for that group”) were excluded.

Using the data in Table A1 it is possible to reconstruct the “Economic Activity” of the total NCDS cohort of 11469 (apparently) individual and compare it with that of the 5903 in the Resolution sample:

Again, the effect of the exclusion applied to produce the Resolution sample from the total NCDS cohort can be seen in the “Social Class” percentages:

Something similar happens with the BCS-derived sample:

Appendix A recognises the apparent over-representation of the higher social classes, particularly in the BCS sample, but goes on to argue:
To test the effect of this on our findings of mobility we re-ran the regression of log earnings found in Table A5 of the Appendix B but with respondents from the managerial and skilled manual social classes excluded from both cohorts. The results are shown in Table A3 and are purely indicative as exclusion of the managerial and skilled manual classes removes 57 percent of respondents from the NCDS sample and 65 percent from the BCS sample. Using these reduced samples, mobility is still significantly higher among the 2000s BCS cohort than the 1990s NCDS cohort. However, we acknowledge that the increased probability of inclusion in the BCS cohort due to respondents being a member of the managerial or skilled manual classes may have increased the overall estimate, but not reduced its significance.
The Resolution report then says:
Another point to note is that the rate of exclusion overall is much higher in the BCS cohort where the excluded sample contained 2,105 more respondents than the final research sample. In comparison, the NCDS excluded sample contained 337 fewer respondents than the final research sample.
This is more easily understood by reference to the table below:
The report then states:
The reason for this is likely to be the data collection methodology of the 2008 BCS survey which used a telephone interview rather than the face-to-face interview used in other NCDS and BCS surveys. Telephone interviews tend to have lower response rates than face-to-face interviews and overall attrition between the start and end points of the BCS survey (2000 and 2008) used in this research was higher than between the start and end points of the NCDS survey (1991 and 2000).
To understand the expression “start and end points” it is helpful to look at the website of the Centre for Longitudinal Studies (CLS, part of the University of London’s Institute of Education). Their NCDS page explains that the NCDS cohort (considered to be 16240 individuals) was surveyed in 1991 and 1999/2000 at the ages of 33 and 42 (cf Table 1 quoted above) and that samples of 11407 and 11419 (about 70%) were achieved. Neither of these figures is exactly the 11469 Resolution figure, but not far off, and there certainly wasn’t any “attrition” between the two surveys.

The Centre’s BCS pages are more problematic. They confirm the Resolution figure of 11261 30-year olds (out of a target 16068, again 70%) surveyed in 1999/2000 (Table 1 again) but provide little data about the 2008 telephone survey, although “the dataset will be available from the UK Data Archive in September 2010”. It turns out that the National Centre for Social Research (NCSR) was contracted by CLS to undertake the 2008 survey, and their Technical Report is available. This reveals that the intention was to survey 11843 of the BCS cohort as selected by CLS and that 8874 telephone interviews were achieved (75%) by NCSR. However, how many of these 8874 were excluded by Resolution as being self-employed or without income is not clear. If 40 % were excluded, as was the case for the 1999/2000 BCS dataset, that would have given a sample of about 3600 individuals. However, this would not be consistent with the statement in Section 3 of the report: “The size of the cohort samples which are above 4,500 in both [NCDS and BCS] cases”.

Not being a social scientist (probably apparent in previous posts about IEA and IPPR) I accept that I lack a proper appreciation of the difficulties such work presents in terms of data. Having acquired the earnings numbers, Resolution, I’m sure, crunched them competently to produce the tabulations of transitions and estimates of mobility. I wonder, however, about the errors in the data that might derive from the exclusions made by Resolution from both the NCDS and BCS datasets. Also, it is not clear how Resolution treated the much smaller 2008 BCS dataset. Its reduced size was not exactly due to “attrition” (a problem which develops over time with all longitudinal surveys) but to a change in methodology. Is it possible that moving to a telephone survey introduced a bias towards individuals with higher earnings?  Anyway, putting it crudely, the report seems to show that in the 1990s, when incomes generally were rising, the NCDS group were all getting better off with small relative changes (all boats rising with the tide) ie there was high absolute mobility.  In the 2000s, a period of largely stagnant incomes, the younger group experienced bigger relative changes (snakes and ladders).  The characteristics of the winners and losers in relative terms is going to be the subject of future Resolution research.    
So, to return to the Waugh/Rentoul questions – I don’t think I can help! Apart from pointing out that the government is more interested in inter-generational than the intra-generational mobility which Resolution was attempting to address.

What should rich men with a conscience, and a broader interest in society than making money out of it, do with their money? Of course it’s up to them if they want to set up think tanks to investigate hobby-horse issues. (Personally I would want to emulate Bryan Ferry and collect modern British art!) I do wonder, though, whether mainstream politics, where the full breadth of social issues and claims on public money have to be addressed, might be a better outlet for these undoubtedly clever and energetic individuals (quite a few of the current Cabinet seem to have made a pile). It seems undemocratic to me that certain single issues can gain more attention than they deserve, by comparison with others which don’t happen to be rich men’s fancies.

No comments:

Post a Comment