Social Policy Research 114 - May 1997
The changing distribution of the 'social wage'

A third of all government spending is on welfare services such as the National Health Service, state education, personal social services, and subsidised social housing. The provision of these services can be thought of as an income 'in kind' - a 'social wage' - which represents an important addition to people's cash incomes, particularly for those at the bottom of the income distribution. A study by Tom Sefton at the London School of Economics examined who was benefiting most from this, how the distribution has changed since 1979, and whether this has affected trends in income inequality. The study found:

• The value of the social wage was 70 per cent higher in 1993 for the poorest fifth of the population than for the richest fifth. Social housing subsidies and personal social services are strongly 'pro-poor'. Schools, further education, and health care are moderately pro-poor, but higher education is markedly 'pro-rich'.
• Demographic factors, relating to gender and age, explain much, but not all, of this pro-poor distribution. Controlling for them reduces the advantage of the poorest fifth to 20 per cent compared to the top fifth.
• The value of the social wage increased by 30 per cent in real terms between 1979 and 1993. The distribution became more pro-poor, although most of this was accounted for by demographic change.
• Changes in the value of the social wage partly offset the rise in income inequality since 1979. According to a common measure of inequality, the increase in inequality between 1979 and 1993 was around one-fifth smaller once the social wage is taken into account.
• The cash incomes of the poorest fifth grew by only 6 per cent between 1979 and 1993. Including the social wage, their incomes grew by 6 to 13 per cent, compared to over 60 per cent for the richest fifth. Allowing for higher spending on welfare services thus makes only a limited difference in the extent to which the poorest have benefited from economic growth.

Background

There are conflicting stories about the distribution of benefits from welfare services provided 'in kind' (rather than as cash). On one hand, official analysis suggests that poorer households benefit significantly more than richer ones from the provision of these services. On the other hand, earlier analysis has suggested that the better-off were getting more than their 'fair share', and that the welfare state had failed in its redistributional role. This research addressed three questions about the distribution of the 'social wage':

• Who benefits most from welfare services? (In particular, have public services been 'captured' by the middle classes?)
• How and why has the distribution of the social wage changed since 1979?
• How do changes in the social wage affect trends in income inequality over this period?

Overall distribution of the social wage

The distribution of the social wage is fairly 'pro-poor': lower income groups receive a greater absolute value of benefits in kind from welfare services than do higher income groups. In 1993, the value of benefits in kind received by the poorest fifth of the population was about 70 per cent greater than for the richest fifth. The degree of this pro-poor bias varies significantly between services. Housing and personal social services are both strongly pro-poor. Schools (including further education) and health care are moderately pro-poor, while higher education is strongly 'pro-rich' (see Table 1). Within sectors, there are important differences in the distribution of particular services. For example, higher education for non-mature students is strongly biased in favour of the top income group, whereas further and higher education for mature students have a small pro-poor bias.

Differences between income groups in the use of private alternatives (e.g. independent schools), targeting of services at those in greatest need, and income-related charges lead to a more pro-poor distribution. Differences in the accessibility of services (e.g. higher education) may tilt the distribution towards being more pro-rich. Unsurprisingly, the age composition of the population also has a strong influence on the shape of the distribution. Most education spending is on children and the majority of NHS spending is on elderly people. This favours lower income groups, which contain a disproportionate share of these age groups.

There are two reasons for regarding 'pro-poorness' resulting from demographic factors differently from 'pro-poorness' that is more directly linked to position in the income distribution. Firstly, demographic factors are a good indicator of 'need' for health care, education, and personal social services, and there is clearly a distinction between someone getting more of a service because they need it more and someone getting more because they are poorer. Secondly, everyone starts young and gets older. Over a lifetime, differences in the use of services that are related to age will be ironed out.

If differences in service use related to age and gender are controlled for, the pro-poor bias in the overall distribution of benefits in kind is considerably reduced (from 1.7 to 1.2), though there is still an underlying pro-poor bias (Table 1). The pro-poor bias in the distribution of personal social services virtually disappears, although social housing maintains its strong pro-poor bias.

The conclusion that welfare services contain a small, but significant, pro-poor bias even controlling for demography requires some qualification:

• it does not allow for differences in need other than those related to age or gender. Thus, lower income groups could still be receiving less in relation to their needs than richer people.
• not everyone receives an equal benefit from using a particular service. If children from wealthier families are getting into the better schools, for example, then the distribution allowing for quality will be less pro-poor than the study found.
• the results in Table 1 conceal effects in the middle of the income distribution. For health care and personal social services, the second and third (poorest) fifths of the population both receive more than the bottom fifth, after controlling for demographic factors. If the needs of the poorest are indeed greater, this could mean that services are not reaching those in greatest need.

Distributional changes since 1979

Between 1979 and 1993, the overall value of the social wage increased by around 30 per cent in real (Retail Price Index-adjusted) terms, which benefited all income groups. Lower income groups experienced bigger increases than higher income groups, so the distribution of the social wage has become more pro-poor over the period (see Figure 1).

Contrary to popular opinion, demographic changes appear to have reduced overall pressure on welfare services. The effect of a smaller child population on education spending more than offset the effects of an ageing population on health care and personal social services spending. The reduction in demographic pressures was greater for higher income groups, which is the main reason they experienced a smaller increase in benefits in kind.

Between 1979 and 1993, the proportion of the social wage received by the poorest half of the population increased by four percentage points (to 60 per cent). But, once adjusted for demographic changes, the share received by the poorest half went up by less than one percentage point. This small increase in pro-poor bias can be attributed to non-demographic factors, including changes in government policy.

Impact on income inequality

The social wage is fairly evenly distributed between income groups, whereas cash incomes are unequally distributed. Adding the value of benefits in kind to cash incomes will therefore have an equalising effect on incomes. Changes in the social wage will affect trends in inequality over time.

The size of the social wage has fallen as a proportion of final incomes (cash incomes plus the social wage) from 14.0 per cent in 1979 to 13.1 per cent in 1993. This reduced the potential equalising effect of the social wage. However, the distribution of the social wage became slightly more pro-poor over the period, which strengthened its equalising effect.

Overall, inequality has increased, even if the social wage is included in the measure of income. Thus, changes in the social wage have not prevented inequality from rising. However, changes in the social wage have helped to offset the increased inequality of cash incomes. According to a common measure of inequality, the Gini coefficient, the increase in inequality since 1979 is smaller by around one-fifth, once the social wage effect is taken into account.

An increase in inequality might be of less concern, if all income groups were experiencing growing incomes. While the cash incomes of the bottom fifth income group grew only slightly over this period (by around 6 per cent), final incomes, which include the social wage, grew by 13 per cent in real terms. However, allowing for higher cost inflation in the public services reduces this growth. The rise in final incomes of the poorest will be between 6 per cent and 13 per cent, depending on how much of the higher real cost of public services was offset by productivity growth.

Overview

Any judgement about the distribution of the social wage depends on initial expectations. For those who anticipated that welfare services would benefit mainly lower income groups, it may be surprising that the poorest half of the population are only receiving 60 per cent of the value of these services, or only 54 per cent, once demographic effects are removed. To a large extent, services are directed at children and elderly people (who are poorer than average), rather than at poorer people per se. On the other hand, the view that public services have been 'captured' by the middle classes is not borne out by the evidence. All services, with the exception of higher education, have some pro-poor bias. Generally, universal services are generating universal benefits with a small bias in favour of lower income groups.

Nor has the situation changed dramatically since 1979. The distribution of the social wage became slightly more pro-poor over the period, but again much of this was accounted for by demographic change. Despite the political rhetoric, there is little evidence of a slimmer, more targeted, provision of welfare services. Spending on services has risen in real terms by around 30 per cent, roughly in line with the growth in GDP. This increase benefited all income groups. Only in housing has there been a clear shift in the distribution between income groups, in favour of the poorest. In other areas, higher charges (or income-related contributions), which might have been expected to increase targeting, appear to have had a marginal impact on the distribution. Where an expansion in services has taken place, for example in post-compulsory education, higher income groups have also benefited, often more than lower income groups.

About the study

This research was based on the Family Expenditure Survey, which is a representative annual survey of around 7,000 households in the UK. Additional information was derived from other datasets, including the General Household Survey, the DSS's Households Below Average Income dataset, and the Building Society Mortgages 5% sample. The analysis was carried out for three separate years: 1979, 1987, and 1993.

Compared with previous studies, new developments in this research include allowing for: variations in service use related to individuals' socio-economic characteristics and incomes; expenditure on the non-household population, including people in residential homes and students living in halls; the effects of demographic change on the distribution of spending; and: using an economic approach, rather than an accounting one to put a value on benefits in kind; examining changes in the social wage on - for the first time - a consistent basis over the period 1979-93 and accounting for these changes; exploring local variations in spending.

Further information

The full report, The Changing Distribution of the Social Wage, by Tom Sefton, STICERD Occasional Paper 21, which includes a detailed service-by-service analysis, is available at a cost of £5.00 from Johanna Ruff, STICERD, London School of Economics, Houghton Street, London WC2A 2AE, 0171 955 6698. The author can be contacted via the same address.