The financial costs and benefits of supporting children since 1975

Stuart Adam and Mike Brewer

A comprehensive, quantitative analysis of trends in child-contingent support from the mid-1970s.

How governments should direct money to families with children is a constant topic of political debate. But the complexity of the ever-changing tax and benefit system makes its overall impact on families anything but transparent, and trends in government support for children hard to distinguish.

This report provides a comprehensive, quantitative analysis of trends in child-contingent support from the mid-1970s to the introduction of the new tax credits, and relates this to changes in tax and benefit policy, the characteristics of households with children and the costs of raising children. Drawing on a large-scale survey spanning the last 28 years, the authors analyse entitlements to child-contingent tax and benefit for thousands of households with children and the costs of raising children.

The report:

  • Examines how support has varied across households and over time;
  • Separates the impact of policy from socio-economic changes;
  • Compares government support for children with available estimates of the actual costs of children.
Summary

Summary

This project analyses how the amount of child-contingent support to which households are entitled has changed since 1975, and relates this to changes in taxes and benefits, the characteristics of households with children, and the costs of children. The main findings, by Stuart Adam and Mike Brewer of the Institute for Fiscal Studies, are that:

  • Total spending on child-contingent support has risen from £10 billion to £22 billion per year since 1975.
  • The largest increase has occurred since 1999, with spending rising by half since then. Child-contingent support now accounts for a higher proportion of GDP and total government spending than at any time since 1975.
  • Changes to tax and benefit policies were responsible for only 40 per cent of the increase in spending per child between 1975 and 1999. The rest was due to the changing characteristics of families. The large increases since 1999, however, are almost all due to policy changes.
  • Many programmes have been used to deliver child-contingent support since 1975. Over time, child-contingent support has become more related to parents' income, as means-tested programmes and tax credits have grown. Universal child benefit, although maintaining its real value, has declined in importance from a peak of 79 per cent of total support in 1979 to 42 per cent in 2003.
  • The proportion of child-contingent support going to lone parent households increased faster than the proportion of children in such households between 1975 and 1997. Since 1997, this proportion, but not the level of support, has declined. Changes to programmes have also increasingly emphasised the first child in a household, favoured young children over older children, and paid support direct to the main carer in couples.
  • Although few estimates exist of the costs of children since 1975, they suggest that the proportion of parents who receive child-contingent support in excess of the cost of their children is small but has grown over time.
  • The McClements equivalence scale, frequently used to measure the costs of children relative to adults, has not been adjusted since it was estimated to reflect the changing costs of children relative to adults. If the cost of children is to be used as a guide to setting levels of child-contingent support, more research on how costs vary across household type and change over time is essential.

The project

How governments should direct money to families with children is a constant political and policy debate, and the new tax credits represent one of the largest changes for 30 years. This project used the Institute for Fiscal Studies' tax and benefit micro-simulation model, TAXBEN, to calculate the extra income that each household in the Family Expenditure Survey received because it had children. The definition of child-contingent support included any transfer that an otherwise-equivalent household without children would not receive, rather than being limited to those parts of the system that are explicitly labelled as being child-related.

More expenditure on children

Total spending on child-contingent support has risen markedly since 1975. Total spending has risen from £10 billion to £22 billion per year in 2003 prices (see Figure 1). This is an increase from 1.5 to 2.0 per cent of GDP, or from 3.4 to 4.7 per cent of total government spending. Over this period the number of children has fallen, and so spending per child increased even faster, from £13.41 to £32.57 per week in 2003 prices.

The rate of growth of spending has been uneven. There have been three periods of rapid increase - the late 1970s, the early 1990s and since 1999 - punctuating periods where child-contingent support changed little or declined in real terms. The most dramatic increase has occurred since 1999: in the past four years, total spending on child-contingent support has increased by 52 per cent in real terms.

The changes are due both to changes to tax and benefit programmes, and to changes in the characteristics of households with children. In fact, 60 per cent of the increase in spending per child since the 1970s has been caused by the changing characteristics of households with children, such as the increasing proportion of households with children where no adult is in paid work. This means that changes to tax and benefit policies between 1975 and 1999 were responsible for only 40 per cent of the increase in spending per child over that time, and makes the increase in support since 1999 - almost all of which is due to policy changes - even more marked.

Changes to the way tax and benefit systems recognise children

More than 35 programmes have been used at some point since 1975 to provide direct financial support to families with children. A number of trends are evident (see Figure 2).

Means-tested child-contingent support programmes have expanded. This means that child-contingent support has become gradually less reliant on universal child benefit, the dominant channel of support for most of the period. Although remaining roughly constant in real terms, child benefit has declined as a proportion of total support from 79 per cent in 1979 to 42 per cent in 2003. The growth in spending on income support and its predecessors largely reflects changes in the characteristics of households with children, while the increased spending on in-work benefits largely reflects policy changes to make these programmes more generous.

The tax system played a large role in delivering child-contingent support in the mid 1970s, but was used little between 1979 and 1999. Since then it has re-emerged, and is now the main delivery vehicle with the child tax credit.

Amongst couples with children, child-contingent support has increasingly been paid to the main carer. In the 1970s, the single biggest programme was delivered by reducing the father's tax liability, but, with the new tax credits, couples with children can now choose to have all child-contingent support paid direct to the main carer.

Child-contingent support and household composition

The amount of child-contingent support to which a household is entitled depends on a range of household circumstances. Some of the most important factors are set out below.

Households with more children receive more child-contingent support on average, although they receive less per child. Child-contingent support has increasingly emphasised the first child in a household, and favoured younger children over older children.

Lone parents, on average, receive more child-contingent support than couples with the same number of children. This reflects their lower incomes, higher support for lone parents than for couples with children (for part of the period studied), and lower support for single people than couples in the absence of children. The proportion of total child-contingent support going to lone parent households increased faster than the proportion of children in lone parent households over most of the period. However, since 1997, the proportion of such support received by lone parents, but not the level, has declined.

The costs of children

This study compared what households receive in child-contingent support with estimates of the direct costs of children (i.e. a household's extra expenditure on goods and services necessitated by having children). It did not provide new estimates of the costs of children, but made use of three types of estimates of the costs of children: the "budget standard" or "minimum needs" approach, which provides estimates of either the costs of children or the total costs of a household with children; the costs of children, as a fraction of household income, implied by various equivalence scales; and the costs of children implied by equivalence scales for a household on a poverty line. The research focused on three equivalence scales: the McClements scale, which is used to adjust household incomes when constructing the income distribution in the UK; a scale used in a recent survey of Poverty and Social Exclusion in Great Britain; and a simple mathematical scale recommended by a panel of experts in the United States.

The researchers concluded that none of the estimates of the costs of children provided satisfactory answers.

The minimum needs estimates are only estimates, and are only available for specific household types at specific points in time. To make use of the existing results in other time periods requires estimates of how children's minimum needs have changed over time. In the report, it was assumed that the material needs of children have not changed over time, and this led to estimates of needs of children that appear extremely high in the late 1970s.

Equivalence scales estimate the cost of maintaining a household's pre-child living standards. This concept is not ideally suited to policy analysis; it usually implies that children cost rich parents more than they cost poor parents. The researchers found that the McClements scale has not been adjusted in line with changes in the costs of children relative to the costs of adults since it was first estimated.

The cost of children implied by equivalence scales for a household on a poverty line produces an estimate that rises over time in line with median household income. But the researchers had no way of knowing whether these costs are adequate, nor whether the costs of children do rise over time in line with median income.

Child-contingent support and child costs

The studies of the costs of children provide rough benchmarks against which changes over time in the level of support can be evaluated. However, the researchers were not fully confident that they provide realistic estimates for all households with children. With this proviso, they reached the following conclusions.

During the 1990s, when the three studies of the minimum costs of children used in this report were conducted, less than 10 per cent of couples with two children received child-contingent support in excess of the estimated minimum. Around 75 per cent of two-child families, however, had total disposable income in excess of an estimated "low cost but acceptable" target income. If these estimates of cost and need have not changed in material terms since then, the proportion of households with children who receive child-contingent support in excess of the costs of their children has increased to around 30 per cent by 2003, and fewer than 10 per cent of households with two children now have incomes below the "low cost but acceptable" threshold. The former change is due to the large increases in child-contingent support since that time, and the latter due to increases in both private incomes and child-contingent support. If the costs of children have increased in real terms, then the record will not look as favourable as this.

Between a fifth and a quarter of households with one or two adults and one or two children now receive more in child-contingent support than is suggested by the McClements scale as being the cost of their children. However, this number is much lower, around 10 per cent, when using an alternative scale produced for a recent Poverty and Social Exclusion survey. Changes over time, though, are the same: child-contingent support is becoming more important to the family budget, and the proportion of households whose child-contingent support exceeds the costs of children implied by key equivalence scales has risen gradually over time, and particularly quickly since 1999, while remaining a minority.

Trends in the proportion of households who receive more child-contingent support than the implied cost of children for a household on the poverty line are complicated, and different equivalence scales give different results. The comparisons confirm, however, that the increase in child-contingent support since 1999 is much more significant than the total increase between 1975 and 1999.

One-child households are more likely than larger households to have the cost of their children covered, even using equivalence scales that allow for economies of scale within households; this gap has grown over time. This mirrors the earlier finding that changes in the tax and benefit system since 1975 have tended to favour one-child households, because child-contingent support has become increasingly linked to the presence of any children, rather than the number of children.

Finally, if policy makers wish to use the cost of children as a guide to setting levels of child-contingent support, this study indicates that more research will be needed on how these costs vary across household types and change over time.

About the project

The study examines entitlements to child-contingent support estimated using TAXBEN, the Institute for Fiscal Studies' tax and benefit micro-simulation model, operating on 52,304 households over 28 years. This was used to calculate the extra income that each household in the Family Expenditure Survey received because it had children. It also made use of household-specific inflation rates, calculated by Ian Crawford and Zo‘ Oldfield and as part of a project funded by the Nuffield Foundation ("The distributional effects of price variation", OPD/00101/G).

Downloads

Downloads