Each year, the Government spends the equivalent of £4,000 per household on subsidised welfare services, including the NHS, state education and social housing. The value of this 'social wage' is, therefore, substantial in relation to people's cash incomes, especially in the poorest households. This study, by Tom Sefton of the ESRC Research Centre for Analysis of Social Exclusion (CASE), looked at how the benefits of welfare spending are shared out, and examined the distributional impact of recent spending increases. The study found that:
- On average, individuals in the bottom two-fifths of income distribution receive around twice the value of benefits in kind as those in the top fifth - over £2,000 per person as against £1,000. Social housing is the most strongly 'pro-poor', but health care, social care and most education services are also pro-poor.
- Differences in income groups' demographic composition are partly responsible for the shape of the distribution. But around half the pro-poor bias is determined by factors related to people's position in the income distribution.
- The overall value of the social wage has increased by 50 per cent in real terms since 1979 and has become gradually more pro-poor. However, this does not necessarily reflect conscious attempts to redistribute in favour of lower income groups. Spending has increased at a faster rate since 1996/97 and the pro-poor bias has increased further. Between 1996/97 and 2000/01, the value of the social wage increased by £260 per person for those in the poorest fifth, and by £50 for those in the richest fifth.
- Changes in the social wage since 1996/97 have reinforced the redistributional impact of tax and benefit changes over the same period. However, their comparative impact has been smaller (around half the size) and less progressive.
- Changes in the social wage did not prevent inequality from rising between 1996/97 and 2000/01. While the social wage has become more pro-poor, it is now smaller in relation to households' cash incomes. These grew faster than their social wage over this period - 12 per cent in real terms as against 8 per cent.
The current Government's welfare reforms have strongly emphasised improved public services. While public spending was tightly restrained for the Government's first two years, spending plans for the rest of its first term incorporated substantial increases, focused on health care and education. The distributional impact of spending on welfare services is of great policy relevance in view of the Government's concern with tackling social exclusion. Although most measures of poverty and inequality ignore the value of benefits in kind, their inclusion is potentially very significant in monitoring the impact of government policies on the poorest households.
Previous research supported by the Joseph Rowntree Foundation examined the distribution of the social wage from 1979 to 1993. This project has updated that work to 2000/01, using better data to generate more accurate estimates of benefits in kind. Three questions were addressed:
- Who benefits most from free or subsidised welfare services, and how do the benefits vary according to people's social, economic and demographic characteristics, focusing on differences among income groups?
- How has the distribution of the social wage changed over time, particularly since 1996/97?
- What impact have recent changes in the social wage had on inequality, and how does this compare with the effects of tax and benefit reforms over the same period?
Current distribution of the social wage
In 2000/01, benefits in kind from welfare services were worth an average of £1,700 per person or nearly £4,000 per household. Health care and education formed the largest components of the social wage, although social housing was very significant in distributional terms (see Figure 1). These benefits in kind were not spread evenly. On average, people in the bottom two-fifths of the income distribution received around twice as much as those in the top fifth, 50 per cent more than the second fifth, and 25 per cent more than those in the middle fifth. There was also a small, but significant, hump in the distribution.
Overall, around half the pro-poor bias can be explained by demographic factors. Lower income groups contain more children and older people, who are the most intensive users of education and health care services. However, a clear pro-poor bias remained even when controlling for differences in the demographic composition of income groups. This shows the importance of taking into account variations in people's use of welfare services which are related to their income and other socio-economic characteristics (factors not taken into account in the official estimates of benefits in kind, produced annually by the Office for National Statistics).
The extent of this pro-poor bias varied markedly among welfare services (see Table 1). Social housing was strongly pro-poor; health care, schools and further education (FE) were moderately pro-poor. Personal social services and the Right To Buy scheme were pro-poor according to the summary indicator used in Table 1 (focusing on the two extremes of income distribution), but the distribution was in fact hump-shaped, favouring those in the middle of income distribution. Higher education was 'pro-rich' if spending was allocated to students' parents. Controlling for demographic differences reduced the pro-poor bias, for the reasons indicated above. For education services, most of the variation among income groups can be accounted for by demographic factors.
Analysis of individual services highlighted a number of factors contributing to a pro-poor bias in the distribution of benefits in kind. Broadly in order of importance, these were:
- differences in the age composition of income groups, except in the case of higher education;
- other factors affecting need, such as longstanding illness in the case of health and social care services, and special needs in the case of schools;
- targeting of services, such as social housing;
- differential use of private alternatives, including private schools and health care;
- means-tested support, for example for residential care, and charges for prescriptions, dental care and other services.
There were also some offsetting factors leading to a more pro-rich (or less pro-poor) distribution than would otherwise be the case. For example:
- participation rates in post-compulsory schooling and higher education for young people from poorer households are lower, although there is some evidence that the gap has been closing;
- better-off social sector tenants have been more able to take advantage of the Right To Buy scheme;
- including housing equity in the means test for residential care support has hit the bottom income group particularly hard, as it contains a high proportion of low income owner-occupiers;
- in the case of social care and to a lesser extent health care, there was an unexplained bias against potential users in the bottom fifth of the income distribution. They reported using fewer services in relation to needs than individuals in higher income groups.
Changes in the social wage
There has been substantial growth in the value of the social wage (see Figure 2). Since 1979, it has risen by around 50 per cent in real terms across the whole population. Lower income groups have benefited more than higher income groups - a rise of over 60 per cent for the bottom fifth, but only 20 per cent for the top fifth. The share of benefits in kind received by the bottom two-fifths has increased incrementally throughout this period. This cannot be accounted for by demographic change, which slightly favoured higher income groups as older people have moved up the income distribution.
This trend continued at a faster rate between 1996/97 and 2000/01. Public spending on health care and education rose by around 3.5 per cent yearly in real terms over this period, compared with just over 2 per cent from 1979 to 1996/97. There was also a further incremental shift in favour of lower income groups. Between 1996/97 and 2000/01, the share of the social wage received by the poorest fifth rose by 0.8 percentage points - from 24.1 to 24.9 per cent - while the share going to the richest fifth fell from 12.6 to 12 per cent. The difference was worth an additional £200 per person for those in the bottom income group.
Changes in the distribution over time are harder to explain than the overall distribution at any given point in time. The amounts involved are relatively small for individual services or specific sub-periods, though the cumulative impact is significant over the longer term and across all services. There is no single or consistent explanation for the increase in the pro-poor bias. Some of these changes reflect a conscious attempt to redistribute the benefits in favour of lower income groups (for example, encouraging higher participation rates in further and higher education for young people from lower socio-economic groups). Some are largely by-products of other government policies, in particular the 'residualisation' of social housing, which has led to an increasing concentration of social-sector tenants at the bottom end of income distribution. Other changes reflect alterations in the pattern of needs, such as the increase in the proportion of individuals in lower income groups reporting a limiting long-term illness, relative to higher income groups.
Impact on inequality
Changes in the social wage between 1996/97 and 2000/01 were progressive. As a percentage of their incomes, households in lower income groups benefited more than households in higher income groups - by 4 to 5 per cent, on average, for the poorest 40 per cent of households, compared with under 1 per cent for richer households. Compared with the effects of tax and benefit policies - the other major instrument for redistribution - changes in the social wage were around half the size and slightly less progressive, but helped to reinforce the redistributional impact of fiscal reforms.
The overall impact of the social wage on inequality was about the same in 2000/01 as it was in 1996/97. Although the social wage is now more pro-poor, it also comprises a smaller share of households' incomes. This is because their cash incomes have grown at a faster rate than their social wage over this period - 12 per cent in real terms as against 8 per cent. These two effects have cancelled each other out. Thus, changes in the social wage have not prevented inequality in final incomes (which include the social wage) from rising as much as inequality in cash incomes.
People in lower income households receive a greater share of benefits in kind from welfare services than those in higher income households. Demographic factors are only partly responsible for the shape of the distribution. The pro-poor bias in the social wage has been rising gradually over the long term. Some changes are by-products of other policies or reflect changing patterns of need, rather than conscious government attempts to target resources at lower income households, though there are some examples of this. Since 1996/97, spending on welfare services has grown faster than in the past and there has been a further incremental shift in favour of lower income groups across all the major services. These changes have reinforced the redistributional effects of tax and benefit policies over the same period, but have not prevented inequality from rising.
About the project
This analysis was based on the Family Resources Survey (FRS), a representative annual survey of around 55,000 individuals in Great Britain. Additional information on the use of other services was taken from the Disability Follow-Up to the 1996/97 FRS, the General Household Survey, and the British Household Panel Survey. The study examined three years (1993/94, 1996/97 and 2000/01) and combined this analysis with earlier work to produce a historical series going back to 1979. This approach sought to improve on official estimates of benefits in kind by:
- allowing for variations in service use related to individuals' socio-economic characteristics and incomes;
- using an economic rather than an accounting approach to estimate the value of housing benefits in kind;
- including services provided to the non-household population, such as people in residential homes and students living in halls of residence;
- simulating the impact of charges and means-tested support for certain services;
- exploring the influence of demographic factors on the distribution of welfare spending;
- using a much larger dataset to obtain more robust estimates.