Good Housekeeping: Ensuring the Basis for Sustained Poverty Reduction

Holly Sutherland ¹
Research Professor
Institute for Social and Economic Research
University of Essex

Abstract

Reducing levels of child poverty (as well as poverty in general) in a way that can be sustained over time requires not only policy measures that create opportunities and future capabilities. It also depends on having a social protection system that keeps pace with economic and social change and is appropriate across all communities, and which does not itself preclude individual risk-taking and initiative. Some conflict is inevitable between the goal of providing adequate social protection for those unable to support themselves and that of maintaining incentives to work and to save for those who have the potential to do so. Nevertheless, there are aspects of the design and evolution of cash benefits (and associated policies) that can improve the terms of the trade-off between the two goals. This paper draws on the experience of the United Kingdom's child poverty agenda over the last eight years, and on assessments of the prospects for a sustained reduction in child poverty in the future, to explore what these features might be. It also considers evidence from international comparisons of poverty and social protection systems.

Introduction

This paper considers "social investment" in two ways. First, it considers what a framework for policy might look like that lays the basis for sustained poverty reduction. Second, it focuses on obvious ways to invest in the future: investment in children and reduction in poverty among children. The term "good housekeeping" in the title refers not to Margaret Thatcher's exhortation to "live within one's means" ² but rather, since investment usually involves borrowing of one kind or another, it is intended as a reference to the need for social policy to have a stable and consensual framework to work within. Drawing on the experience of the United Kingdom's child poverty agenda over the last eight years, and on assessments of the prospects for a sustained reduction in child poverty in the future, I explore what lessons can be learned that might be relevant in other countries.

The next section explains how the UK targets for child poverty were set, briefly describes the associated policy reforms, and provides an assessment of their effects and prospects for the future. Several key issues for the longer term are identified and two of them are discussed in the following two sections: respectively, the basis for the regular uprating of benefits (and tax thresholds) and the complexities of understanding what matters in determining whether child poverty will fall in a sustained way. A subsequent section draws on evidence from the other countries of the EU about the relationships between cash support for children and child poverty. The concluding section summarises the main points of the paper and finishes with some open questions for discussion.

Child poverty in the UK: what has happened and what might happen?

As is well known, the UK government has set targets for reducing child poverty and eventually "eliminating" it by 2020. The original pledge was made unexpectedly by Tony Blair in a lecture in March 1999 where he undertook to "end child poverty forever" within 20 years (Blair 1999). An immediate target was set of cutting the child poverty rate by a quarter against a relative target between 1998/99 and 2004/05.  Since then a more elaborate series of targets has been established, which to some extent recognise the multi-dimensional nature of poverty, some of which are backed up by the authority of Public Service Agreements (PSA). 

• There is a PSA target to halve the number of children in relative low-income households between 1998/99 and 2010/11. Relative low-income households are those with income below 60% of the contemporary equivalised median, measured without deducting housing costs. Meeting this target would mean reducing the number of poor children in Britain to 1.7 million.
• There is a second tier of the 2010/11 target (without a PSA) to reduce the number of children in "absolute low income" to under 1 million. Absolute low income is having an income below the 1996/97 poverty line, indexed for inflation. The latest estimate, for 2004/05, is 1.4 million children below this poverty line. 
• A third tier - currently in the form of a commitment to set an additional target (see Harker 2006) - consists of a measure of material deprivation and low income combined: a child is considered poor if it is both measured as being materially deprived (lacking a series of essential goods and services) and being in a household with income below 70% of the contemporary median.

Child poverty will be considered to be falling if all three measures are moving in the right direction (DWP 2003). Success in "eradicating child poverty" - the implicit target for 2020 - is being interpreted as having a material deprivation rate for children that approaches zero, together with having a relative child poverty rate that is "among the best in Europe". Exactly what this means is not clear but has been assumed to be equivalent to a child poverty rate that is in single figures like those of the Nordic countries. Figure 1 shows the latest available figures for child poverty rates in the EU, corresponding to 2003 incomes. The value of 22% is lower than that shown for the UK at the time when child poverty reduction became a high profile policy goal (using 1998 incomes) at 29%. Nevertheless it remains far short of being one of the lowest rates in Europe, at the sixth highest in the EU25.

Figure 1  Child Poverty Rates in the European Union, 1998 and 2003 Incomes

Note: Estimates for the two years come from different sources and the 2003 income sources differ across country.
Source: Eurostat 2007 Table 4.1 and Eurostat 2003:66

I return to how the UK might learn lessons from its EU neighbours in the section "What do other countries do?" Here, I consider further the extent to which British child poverty rates have fallen and how far child poverty rates must fall if the targets for 2010/11 and 2020 are to be met.

In 1998/99 the British child poverty rate (at 24% using national conventions for measurement) was among the highest in Europe and the OECD. It fell to 19% (according to the latest estimate for 2004/05), at least partly due to the policy measures introduced in response to concern about this situation. The trajectory is shown in Figure 2 alongside the path that child poverty rates should follow if there was to be a smooth reduction in child poverty to meet the 2004/05 target followed by the 2010/11 and 2020 targets (the level corresponding to "among the best in Europe" being assumed to be 5%). As can be seen, the 2004/05 target was not quite met: child poverty fell by 23% of its level in 1998/99 rather than the targeted 25%. In order to speculate whether the fall in child poverty can continue on a sustained basis, or even accelerate in order to meet the targets, we need to understand what contributed to the reduction observed so far.

Figure 2  Great Britain Child Poverty Rates: Actual and Targeted Levels

Note: Relative threshold is 60% of contemporary median.
Source: DWP 2006 Table H2

Effects of Policy Reforms

From 1999 onwards a whole battery of policy reforms was introduced with the aim of reducing child poverty. These ranged from increasing benefit payments for children, to devising new forms of cash support for parents with low earnings, to encouraging and enabling entry into paid work (particularly for lone parents), to setting up new services for disadvantaged children. For more information about the specific policy measures, see HM Treasury (2004).

It is not straightforward to establish how much the extra measures cost or how great their effect. One estimate is that up to 2004 the cash income changes amounted to an increase in spending equivalent to 0.6% of GDP on child-contingent parts of the tax-benefit system and 0.3% on services for children, not including education (Hills and Sutherland 2004). The Treasury itself expressed the spending on benefits and credits for children over the same period as a "real terms rise of 72%" (HM Treasury 2004). This serves to pose the question of what to assume about the counterfactual policies (what would have happened otherwise) and, relatedly, what to assume about how taxes and benefits should adjust over time. This latter point is considered in the following section, "Principles for regular tax–benefit uprating".

Generally the approach to policy reform has been, on the one hand, to seek solutions through "work for those who can; security for those who cannot" (DfEE/DSS 1998) and, on the other hand, to "focus both on the direct ways of helping today's children and on improving the capacity of tomorrow's parents to lead fulfilling lives, free of poverty" (Hirsch 2006:49). In essence, this latter agenda has been seen as being best addressed through improvements in the effectiveness of formal education.⁷

The focus of this paper is on the medium-term agenda. Delivering a higher-capacity generation of parents through improved educational outcomes may well contribute to lower child poverty rates, and policy with this aim is clearly an appropriate component of any long-term strategy. At the same time, if child poverty rates do not continue to fall in the short and medium term, and the medium-term target in 2010/11 is missed by a long way, it is implausible that improved parental capacity alone can deliver the eradication of child poverty by 2020. (It is also, of course, uncertain whether educational outcomes can be improved enough and fast enough, or whether such changes necessarily lead to the returns to work that are required to have a major impact on child poverty.)

The evidence in Figure 2 suggests that the medium-term target might be met through "more of the same", given that the strategy so far has been quite successful, even though the 2004/05 target was not met. The trend appears to be downward at about the right gradient. However whether this is feasible depends on our assessment of why child poverty rates did fall. The two main drivers were an increase in parental employment and the direct boost to incomes provided through increased out-of-work benefit rates (for children) and a new and more generous system of in-work tax credits. While it is generally agreed that work is the best route out of poverty, it has been estimated that only one-sixth of the child poverty reduction shown in Figure 2 is due to growth in parental employment (Hirsch 2006). Entry into work is not a guarantee against poverty, even with a tax-credit system in place, and there are limits to the amount of new employment that can be created and sustained that is both in the right place and of the right type for parents of young children to take up. Between 1997 and 2001 employment increases accounted for about half the fall in child poverty (Sutherland et al. 2003) but taking the period 1998/99 to 2004/05 as a whole it seems that the main driver was increased incomes for families who qualified for increased cash support. This included both couples (one-earner) and lone parents, as well as workless families (Brewer, Goodman et al. 2006).

Moreover, whether the increase in employment was directly due to government employment-related policies is another matter. The tax credits were intended to increase parental employment by making work pay and there is evidence of small increases in employment that would not have happened otherwise (Brewer and Browne 2006, Gregg et al. 2006). It seems likely that employment effects due to a growing economy overwhelm the responses to changes in benefits and credits. As Hirsch (2006:46) puts it, "Quantifying the contribution made by public policy is extremely difficult, but best estimates are that it has raised the proportion of couples with work by about 1 per cent and the proportion of lone parents with work by about 5 per cent (Gregg et al. 2006)."

Arguably, the main roles of the in-work income support system have been to (a) increase the incomes of low-paid parents and (b) to maintain incentives at a time when incomes of parents not in work were being raised.

Looking Ahead

Any strategy to keep child poverty rates falling as shown in Figure 2 requires continued efforts to encourage parents to take paid work, while also recognising that cash payments to families with children will have the major role, either performing a social protection function, or in maintaining (and improving) work incentives. But simply maintaining the value of cash payments in real terms will not be sufficient on either count. Relative poverty is measured by comparing disposable incomes with those at the overall median. Many components of market income typically rise faster than prices, so the poverty line rises in nominal terms at a faster rate than cash payments, which are usually indexed for inflation. To even keep poverty rates constant, benefit payments have to be increased at the same rate as (net) market income or targeted increases in particular benefits need to be made. The latter corresponds to the UK government's strategy and we have seen benefit and tax credit rates for children increased by more than inflation (and the child tax credit amount per child will be increased with earnings over the medium term). Of course, to increase incomes of recipients of these transfers relative to the median (and hence reduce child poverty rates) requires regular increases of more than the growth in average incomes. In other words, these benefits need to take a larger share of the public cake. There are three risks associated with this strategy.

First, and most obviously, other groups might legitimately resist their share of cake becoming smaller. The alternative of a bigger cake – increases in public spending overall – is not considered in full here, but may well be inevitable in the short or medium term if real investment is to be made. Second, and relatedly, hiking up payments for children risks unbalancing the structure of payments. Some children in the UK are already entitled to higher payments than certain adults. This may not be ruled out from a design point of view – since it could be argued that developing children have greater needs than some adults. But experiencing a sudden drop in income – on one's birthday – could be problematic for individuals and their families and is not likely to be popular. Thirdly, it may actually be less efficient in reducing child poverty measured using household income to target children rather than all individuals in the household. This is clearly the case in a mechanical sense: one child in a household of several adults would require a very large benefit increase to lift them all (including the child) above the poverty line. There may also be subtler ways in which relying on targeting children with cash payments, rather than supporting adults – or anyone who can benefit from support at their particular stage in the life cycle – might not be effective in correcting the underlying problems.

Another relevant factor is the number of children. This is predicted to fall in Britain and this has two effects on the prospects. First, there is again a mechanical point: the targets are carefully worded in terms of the reduction of numbers of children in poverty. If the rate of poverty remains constant while the overall number of children falls, the number of poor children also must fall. The more complex point is whether a society with a lower proportion of children allocates more resources (of all sorts) per child or whether the slice of cake shrinks and resources follow the demographic groups that are growing in relative size (in this case, quite obviously, the elderly).

So what are the prospects for meeting the 2010/11 target of reducing UK child poverty by a half since 1998/99? One estimate, based on rolling out existing policies, projections of demographic change and some fairly optimistic forecasts of employment growth among parents⁸ suggests that child poverty rates will be only very marginally lower in 2010/11 than in 2004/05: 2.65 million children will be in poverty instead of 2.70 million in 2004/05 (Hirsch 2006). The same analysis estimates that the 2010/11 target could be (almost) met with a substantial 43% real increase in the level of child payments in the Child Tax Credit (CTC). The CTC is targeted at low-income families with children whether or not they work; the child payments are tapered away rather slowly and after a large effective disregard for those in work, so the vast majority of poor and near-poor children are in households entitled to the CTC. However, receipt of the CTC, even at the greatly increased level modelled for 2010/11, would not be a guarantee against poverty. The proposed increase would cost £4.3 billion (or around 0.3% of GDP) in 2010. There are, of course, other options and, for example, the same study explored the effects of using some of the resources on extra benefit for large families and on using the universal child benefit as a vehicle for part of the transfer to children.

All such options are worth considering, and a finely tuned package of extra spending on cash support for children is undoubtedly necessary for the medium-term target to be met. However, there are two other factors that help in understanding why it is so hard to reduce child poverty rates and provide clues to what other issues need to be addressed. These are considered in some detail in the following sections. First, there is the question of how benefit rates and thresholds are updated as prices and incomes rise. If poverty is to be assessed against contemporary median incomes, then what can be done about the "running down the up escalator" aspect of the problem? At the level of principle this is a matter that applies similarly to all systems and countries and is discussed in the next section.

Second, it is useful to reflect on the fact that most of the attempts between 1999 and 2005 to predict the effect of policy reforms on child poverty came up with more optimistic outcomes (including meeting the 2004/05 target) than in fact occurred. There are a number of plausible explanations. Some are related to the techniques and data used for the predictions and others to the complexities of behavioural reactions to policy changes. These are discussed in the section "Predicting child poverty: What matters?"

Principles for regular tax–benefit uprating⁹

This section considers the principles underlying different ways of adjusting benefit rates and thresholds as prices and incomes rise. If benefits were designed to achieve across-the-board reductions in relative poverty, their values would need to rise at least with the benchmark used to set a poverty standard. In the context of the UK at present, this would mean keeping up with a measure of median household disposable income adjusted for household size. Similarly, if the tax and benefit system is constructed with the aim of affecting household income inequality, then adjustment in line with mean net or disposable income growth would be appropriate as a starting point. If this is not done, the shares of different people's income taken in tax and granted in benefits will change, and with them the shape of the distribution. By contrast, if all incomes grow by the same proportion, and with them benefit rates and tax brackets, the relationship between net and gross incomes would stay the same, as would direct tax revenues as a share of aggregate income.

As a variant of this, some argue that the benefit rates that effectively determine society's minimum income should be set explicitly to allow a minimum living standard set in relation to contemporary views to be achieved. Adjusting benefits in line with such "minimum income standards" requires a two-stage process: periodic recalibration of what it is that is required for a minimum living standard, and then finding out what this costs to achieve given contemporary prices. Adjustments then follow a combination of the changing minimum (consumption) standards, and changes in prices.

On the other hand, if benefits are seen in some way as a return on contributions, as is (in the UK case) loosely embodied in the national insurance system, then their values may be linked to what people have paid in or the "contribution base". This might suggest a link with, for instance, gross earnings.

However, policy may be driven by questions of affordability, rather than by what might ideally be achieved. In this case it may be growth in the economy or potential tax base that is most relevant, for instance, GDP growth. At times of high growth, benefit levels would rise such that recipients had some share in the country's increasing prosperity.

By contrast, if benefits are designed to guarantee a particular real, but unchanging, standard of living, adjustment would need to be by prices. ¹⁰ Generally this is the principle on which uprating takes place in the UK, as in the non-contributory parts of systems in many other countries. Given the relative movements of prices and the other benchmarks – earnings, GDP and mean and median disposable income – in recent years, this has tended to result in benefit levels (and tax thresholds) progressively falling behind the levels that would be achieved under any of the other uprating principles set out above (except, arguably, the minimum income standard principle).

However, such principles take for granted that the starting levels are what are in fact desired, and the only issue in adjustment is to keep them in line with the appropriate benchmark for the objective. However policymakers are often faced with the challenge of reforming a system from what is, from their point of view, an unsatisfactory starting point. Transitions to a new system are often made by using more or less generous uprating than implied by the underlying principle, until the intended structure is reached. This can be seen as a way of exploiting "money illusion" to disguise the losses that are occurring for some, or as a way of phasing in structural changes slowly in a way that minimises disruption to household budgets.

The difficulty of such adjustment from "policy disequilibrium" is that what is happening is seldom made explicit – indeed, sometimes the objective is precisely to disguise from some of those affected what is going on. So a final principle to guide uprating practice might be that of flexibility: to minimise the extent of statutory uprating, leaving the government free to make the adjustments that follow from their political priorities and the economic circumstances, and permitting it to claim credit for increases that in fact only maintain the status quo (according to one principle or another). Having no rule at all could be quite consistent with sustainable and desirable outcomes if there were full public understanding of the issues at stake and if changes in benefits and taxes, as well as redistribution of incomes, were a major item of public discussion and democratic debate. This is rarely the case.

Predicting child poverty: what matters?

Predicting child poverty for some date later than the reference period for the available household micro-data can only be done approximately. Understanding why modelled predictions and eventual direct measures diverge can not only improve projection methodologies but also help us understand what is important for actual child poverty outcomes. First of all, demographic and other changes between the survey year and modelled year are difficult to account for precisely. Secondly, difficulty in predicting behavioural reactions to policy changes as well as to the changing economic circumstances introduces discrepancies and uncertainties. For example, non take-up of means-tested benefits and tax credits – difficult to measure and take account of in any case – has been higher in the case of the Child Tax Credit (CTC) than originally envisaged (Brewer, Goodman et al. 2006). ¹¹

It also seems likely that analysis based on static tax-benefit modelling does not capture all the relevant change in household circumstance that in fact occurred over the six-year period (and might also over the following six years up to 2010/11). It is quite plausible that the reforms, combined with other trends and changes, have led to a somewhat different set of trade-offs for parents between paid work and not working. Indeed, it may in some circumstances be a lot easier for parents to achieve below-poverty-line levels of incomes than it is to construct combinations of work and care that offer just-above-poverty levels of income. Put another way, some parents may have taken the opportunity to reduce the amount they work once increased support through tax credits was available. Furthermore, it is still possible that parents do not understand the new system, as is suggested by low levels of take-up. And it is plausible that the system is somehow inappropriate for some types of families, as indicated by high levels of worklessness and child poverty within some ethnic minority populations (Platt 2006).

It is the group of children in one-earner couple families who contribute the largest number to the remaining stock of children in poverty. Current debates in the UK centre around whether these families should be targeted further with state support (either through the tax system or extra help through the introduction of an additional couple element in the tax credits) or whether the barriers to work for second earners due to the work disincentive effects of the existing tax credits should be addressed (Chote et al. 2007 Chapter 12). The tax credits are means-tested on joint income resulting in second earners having a high effective marginal tax rate on the first £ earned. What this comes down to is an argument about the roles of mothers and whether they should be supported to be in paid work or not to be in paid work. Also important is the persistence of the gender pay gap, to the extent that it is this that turns women into "second" earners.

Wage rates more broadly are clearly a factor too. For work to be effective as a child poverty reduction strategy, wage rates need to be high enough so that, when combined with in-work benefits and credits, family income is above the poverty line, as well as high enough to provide a real incentive to combine paid work with the tasks of parenthood. In Britain, in spite of a national minimum wage that is relatively high in international terms, it remains the case that 35% of poor children have a parent in full-time work (DWP 2006a Table E5).

There may also be specific factors at work in some parts of the economy. For example, in London child poverty rates are higher than in the UK as a whole and have been more resistant to reduction. ¹² Reasons for this include the low level of parental employment in London, particularly among lone parents and particularly for part-time working (Buck et al. 2007). Not only does it appear that the supply of part-time jobs is lower than elsewhere but also – and relatedly – that part-time working for parents is costly or infeasible if travel is expensive and journey times long. (Similar issues may apply in rural areas.) Since the poverty risk of children in workless households or one-earner couple households is high, encouraging entry into part-time work (along with the provision of such jobs) may seem like a good idea, but this will only be effective if the returns for such working are worthwhile.

Furthermore, the relationship between parental employment, child poverty and ethnicity is complex. Child poverty rates are lower for children of White parents than for any other of the major UK ethnic groups. However, employment rates are higher for Black Caribbean lone parents than White lone parents, and children with both parents earning are more common among Indian couples than among other groups (Buck et al. 2007). While the measures to tackle child poverty in lone-parent families have been shown to be relatively effective among Black children (where 69% of those in poverty are living with a lone parent), poverty reduction among children living in Bangladeshi and Pakistani families – where lone parenthood is relatively rare – is much lower (Platt 2006). At least part of the explanation is in the prevalence of single-earner couples and the low rates of pay received by Bangladeshi and Pakistani men. As noted by Harker (2006), "the ethnic minority employment gap is too often dismissed as resulting from 'cultural' differences when there is clear evidence of the significant barriers that some ethnic minority groups face in entering and progressing in work" (p.28).

Finally, one obvious failure of the tax credit strategy is that take-up rates remain low: some families simply do not receive their entitlements. The latest official estimates suggest that while the national take-up rate by lone parents for the combination of tax credits to which they are entitled while in low-paid work is between 86% and 95%, the corresponding range for couples with children is lower: between 71% and 75%. In London, the range for both groups of families is between 58% and 73% (HMRC 2006). Thus large proportions of low-income families have lower incomes than they should. This, one way or another, is due to the complexity of the entitlement criteria and the claim process. The credits are means-tested and there may be stigma attached to a claim or to receipt. At least in the first years of introduction there were many overpayments which had to be repaid, possibly leading the system to be mistrusted by potential recipients. The claim process is complicated, requiring both partners in a couple to provide information and make a joint claim. There may even be misunderstandings related to the term "credit": some potential claimants may believe that the payment is a loan and has (always) to be paid back. In any event, the means-tested character of the credits reduces their effectiveness relative to the universal child benefit, which is also cheap to administer and widely accepted.

What do other countries do?

As explained above, the success of the UK child poverty strategy will be judged in the long term as having a child poverty rate that is among the best in Europe. We have seen that the child poverty rate remains one of the highest, even though the position has improved since 1998. The countries of the EU take very different approaches to the design of social protection systems and policies to make work pay. They also have widely diverging rates of lone parenthood, of parental worklessness and also fertility, earnings distribution and maternal employment. "Policy learning" across countries is part of the agenda to promote social inclusion and reduce poverty in the EU (Marlier et al. 2006). But it would be mistaken to think that a policy approach or package that was apparently successful in one country could be applied mechanically in the UK and be equally successful.

At the same time, it is instructive to compare the systems of support for children in the context of the child poverty rate, and the extent to which payments for children succeed in reducing the child poverty rate. Figure 3 shows the distributional incidence of child-contingent cash support in each of the EU15 countries in 2001. "Child-contingent" support includes not only family and child benefits but also additions or complements within other benefits (e.g. social assistance) that are paid by virtue of the presence of children, as well as tax concessions. The effect of benefits is indicated by the lighter parts of the bars. The effects of taxes are shown by the dark part of the bars. In the case of tax concessions such as child tax allowances, these are shown as positive. In the case of taxes paid on child-contingent benefits, the effect is shown as negative. The net effect is the net amount received. (The amounts are the average payments per child and in order to make comparisons across countries with different income levels they are expressed as a percentage of average disposable income per head of population. ¹³) The most striking feature is the very great differences across the 15 countries in terms of the size of the payments to children, how they are distributed across the household income distribution and whether delivery is through (net or gross) cash payments or tax concessions.

Figure 3 Spending per Child on Child-Contingent Cash Payments in 2001 by Decile Group

Note: Spending is expressed as a percentage of total per capita disposable income.
Source: EUROMOD

The UK appears as a country with a relatively generous system for children, heavily, although not exclusively, targeted at households with low, and low-to-middle, incomes. (Since the calculations are for 2001, they pre-date many of the reforms discussed above. Such a diagram for today's system of cash support for children would show higher amounts in the bottom half of the income distribution.)

One might expect child-contingent spending targeted at low incomes to be associated with a high degree of child poverty reduction, or put another way, to play a major role in protecting children from poverty. Figure 4 indicates whether this is so by comparing the child poverty rate in 2001 with what it would have been without the parts of the tax and benefit systems that are child-contingent (see Corak et al. 2005). This picture (dark bars) does not show what child poverty rates would look like were there no state support – in this event people would have to reorganise their lives in order to survive. Instead, it is a measure of the importance of state support to household incomes in keeping children above the poverty line. Countries are ranked by the child poverty rate using 60% of the national median as the poverty line (shown by the paler bars).

Figure 4 The Child Poverty Rate in EU15 in 2001, With and Without Child-Contingent Incomes and Without All Benefits (including Public Pensions)

Source: EUROMOD

It is not the case that the successful countries depend more heavily than the UK on the child-contingent system for keeping their children above the poverty line (indicated by the distance between the top of the paler bar and the top of the darker bar). The differences lie in the level of poverty risk before child-contingent support, which is much lower in the low child poverty countries like the Nordic countries, Belgium and Austria. They also lie in the nature of the support systems and here a number of distinct features are relevant: the extent of means-testing, the extent of child-targeting, the size of the system of state cash support for children and the way in which state support influences household composition. These are considered in turn, comparing the UK with countries where particular differences are evident.

The child poverty rate in Austria is half that in the UK. The Austrian system is largely contingent on circumstances and it is not primarily targeted by income, and yet the Austrian child support system has a similar effect on child poverty as that in the UK, which is heavily reliant on means-testing (Levy et al. 2007).

Furthermore, some support is provided indirectly for children through the non-child-contingent parts of state support received by adults in the same household (such as state pensions or unemployment benefits, or social assistance payments for adults). The extent to which all cash benefits (including state pensions) protect children from poverty is shown by the dark points in Figure 4. ¹⁴ In countries like the UK the additional role of adult payments is rather small. In Austria it is larger and this effect plus the effect of child-contingent payments reduces Austrian child poverty from the fourth highest (based on primary incomes) to the fifth lowest (based on disposable incomes). In Denmark and Sweden the effect on children of "adult" benefits is much larger. Poverty rates based on market incomes are rather similar in Sweden and the UK. Swedish child poverty is among the lowest in the EU15 countries with the contribution of benefits for adults being larger than the contribution of child-contingent payments. Bradshaw (2006) includes similar findings for Sweden.

Several countries have small systems of child-contingent support: Spain, Greece and also the Netherlands are notable in this respect (Figure 3). They also have rather low rates of child poverty before the addition of child-contingent support and, in the case of the Netherlands and Greece before the addition of any benefits and pensions, where their rates are the lowest of the 15 countries (Figure 4). Low rates of poverty on the basis of primary incomes can arise because inequality in the distribution of primary income is relatively low, with few households receiving low or zero incomes. The Netherlands is one example of such a case. ¹⁵ The other way in which low primary income child poverty arises is simply a result of a lack of a generous support system: households organise themselves such that dependants are supported by family members who do have market sources of income, or else prospective parents do not start families until they are able to support them. This is typical of the Southern European countries where fertility is low, where households may often contain three generations and where the age of leaving home on average is much later than in Northern or Central Europe (Aassve and Iacovou 2006). While it may appear that there is less for a child support system to do to protect children from poverty, this masks other long-term problems related to low fertility and delayed household formation.

To summarise the lessons from other countries for the UK, it seems that a high child poverty rate based on primary income, while not in itself desirable, is not an insurmountable barrier to low child poverty rates based on disposable income (Sweden, Austria). It is not necessarily the case that high child-contingent payments are necessary for the child poverty rate to be low: a mix of child and adult instruments is in some cases more effective (Denmark, Sweden). Means-testing, while in the short run cheaper than providing the same high level of support to all children, may be only somewhat less effective in terms of poverty reduction than a system that targets by contingency (Austria, and also Denmark, Sweden and Belgium). ¹⁶ We do not know what effect family means-testing has in the long run on labour market and household formation behaviour. Nevertheless it appears from Figure 4 that systems that have relied on means-testing to a substantial extent for some time (UK, Ireland) are associated with high primary income poverty.

Conclusion

I have considered aspects of social investment in two ways. Firstly, this has been done through the channel of children and by considering policy designed to improve the lot of children now, with a view to them reaching adulthood unscarred by poverty. Secondly, I have tried to identify the factors to be considered when thinking about tax and benefit policy, which could have a sustained effect in reducing child poverty over the medium and longer term. In the short term, targeting by income and on children as the group of concern may seem to follow obviously from the policy priority. Thus indexation of all payments and thresholds may seem like a secondary issue or one implying unnecessary increases in public spending. However, with a longer-term perspective there are strong arguments for:

• a comprehensive and explicit uprating strategy
• less reliance on means-testing
• less reliance on child-targeting
• broader appreciation of the costs and benefits of working for parents.

With this focus I have inevitably neglected some important issues. These include the share of spending that is devoted to services, rather than income support, for children. Related to this is the conceptualisation of poverty as something additional to, or distinct from, low household income. Many initiatives that are intended to improve the wellbeing of children or their future prospects as adults do not involve cash transfers and their effects would not be captured directly in income poverty measures. In the UK outcomes for children in some other important dimensions are monitored alongside income poverty in the form of Opportunity for All indicators. These include, for example, school attendance, smoking among children aged 11–15 and re-registrations on the child protection register (DWP 2006b). Over the period I have considered, most but not all the indicators have moved in the right direction or stayed constant. Exceptions include obesity in young children, the education gap among looked-after children, infant mortality and families in temporary accommodation. In comparative perspective an aggregate index of an even broader set of indicators of child wellbeing recently ranked the UK at the bottom of 21 OECD countries (UNICEF 2007).

However, it is extremely difficult to judge the extent to which either cash-based or non-cash public policies in general, or even particular policies, have had an impact on these outcomes. For example, Sure Start is a programme initiated in 1999 to deliver "the best start in life for every child", bringing together at a local level services such as early education, childcare, health and family support for families in disadvantaged areas. Although established as a long-term programme with long-term goals, Sure Start Local Programmes (SSLPs) have been subject to an early evaluation, which claimed that SSLPs benefit children of less socially deprived parents at the expense of those most deprived (Belsky et al. 2006). However, this evaluation has been contested on methodological grounds with the argument that some form of randomised control trial should have been used and that the poor outcomes for some children may have been due to unmeasured poor prospects at the start. Establishing "What works" is not straightforward.

It is also important to highlight the importance of one particular resource and investment in children that cannot be provided by government but which can be influenced by it, and that is parental time and capacity to parent effectively. The focus on "work for those who can" risks ignoring the pressures placed on parents. This is well illustrated by a parent at a feedback event within the recent Joseph Rowntree Foundation project on What Will it Take to End Child Poverty? ¹⁷ who said:

All my tax credits and most of my wages go to keeping my four kids; they're in three different kinds of childcare and I've been working for six years and I'm not any better off for working. I'm playing at being an employee and playing at being a mum – can't do either fully. (quoted in Hirsch 2006:29)

Being better off in work involves more than a mechanical financial calculation: it also requires the right sort of job with hours, flexibility, convenience and childcare to fit, providing enough pay and leaving enough time to spend being an effective parent.

I will summarise this paper with a set of questions that it has, explicitly or implicitly, raised. They are none of them new, nor do they apply exclusively to the UK.

• Should the mechanisms, formulae and size of annual increases to benefits (and tax thresholds) be something determined by each government on a year-to-year basis, or should conventions be sought around which either consensus or dialogue might be developed?
• If most of the decrease in worklessness has been due to a growing economy (rather than welfare-to-work policies), what kind of "making work pay policy" will be effective in a downturn? What policies encourage the creation of "good" jobs that can be combined with parenting?
• What are the most effective approach(es) to increasing the income of one-earner couple parents?
• Is targeting (cash) support at children rather than adults the best way to protect children from poverty?
• What are the long-term effects of a mainly means-tested system of support?

References

Aassve, A. and M. Iacovou (2006) "Youth poverty and transition to adulthood in Europe" Demographic Research, 15:21–40.

Belsky, J., E. Melhuish, J. Barnes, A.H. Leyland and H. Romanuik (2006) "Effects of Sure Start local programmes on children and families: Early findings from a quasi-experimental, cross sectional study" British Medical Journal, 332:1476.

Blair, T. (1999) "Beveridge revisited: A welfare state for the 21st century" in R. Walker(ed.) Ending Child Poverty, Policy Press, Bristol.

Bradshaw, J. (2006) A Review of the Comparative Evidence on Child Poverty, Joseph Rowntree Foundation, York.

Brewer, M. and J. Browne (2006) The Effect of the WFTC on Labour Market Participation, IFS Briefing Note BN69, Institute for Fiscal Studies, London.

Brewer, M., J. Browne and H. Sutherland (2006) Micro-Simulating Child Poverty in Great Britain in 2010 and 2020, National Poverty Center Working Paper #06–31, University of Michigan.

Brewer, M., A. Goodman, J. Shaw and L. Sibieta (2006) Poverty and Inequality in Britain: 2006, IFS Commentary 101, Institute for Fiscal Studies, London.

Buck, N., H. Sutherland and F. Zantomio (2007) Implications of Demographic and Economic Change for Child Poverty in London, Report for the London Child Poverty Commission.

Chote, R., C. Emmerson, A. Leicester and D. Miles (eds.) (2007) The IFS Green Budget: January 2007, Institute for Fiscal Studies, London.

Corak, M., C. Lietz and H. Sutherland (2005) The Impact of Tax and Transfer Systems on Children in the European Union, Innocenti Working Paper No. 2005–04, UNICEF Innocenti Research Centre, Florence.

Department for Education and Employment, and Department of Social Security (1998) A New Contract for Welfare: The Gateway to Work, Cm 4102, The Stationery Office, London.

Department for Work and Pensions (2003) Measuring Child Poverty, DWP, London.

Department for Work and Pensions (2006a) Households Below Average Income, 2004/5, The Stationery Office, London.

Department for Work and Pensions (2006b) Opportunity for All – Eighth Annual Report 2006, Cm 6915, The Stationery Office, London.

European Observatory on the Social Situation (2005) Network on Social Inclusion and Income Distribution Final Report.

Eurostat (2003) Living Conditions in Europe: Statistical Pocketbook, Data 1998–2002, Eurostat, Luxembourg.

Eurostat (2007) Living Conditions in Europe: Statistical Pocketbook, Data 2002–2005, Eurostat, Luxembourg.

Gregg, P., S. Harkness and L. Macmillan (2006) A Review of Issues Relating to the Labour Market and Economy, Particularly in Terms of the Impact of Labour Market Initiatives on Children's Income Poverty, Joseph Rowntree Foundation, York.

Harker, L. (2006) Delivering on Child Poverty: What Would it Take? A Report for the Department for Work and Pensions, Cm 6951, The Stationery Office, London.

Hills, J. and H. Sutherland (2004) "Ending child poverty in a generation? Policies and prospects in the UK" (mimeo).

Hirsch, D. (2006) What Would it Take to End Child Poverty? Firing on All Cylinders, Joseph Rowntree Foundation, York.

HM Revenue and Customs (2006) Child Tax Credit and Working Tax Credit Take up Rates 2003–04, HMRC, London.

HM Treasury (2004) Child Poverty Review, The Stationery Office, London.

Levy, H., C. Lietz and H. Sutherland (2007) "Swapping policies: Alternative tax–benefit strategies to support children in Austria, Spain and the UK" Journal of Social Policy,36(4).

Marlier, E., A.B. Atkinson, B. Cantillon and B. Nolan (2006) The EU and Social Inclusion: Facing the Challenges, The Policy Press, Bristol.

Platt, L. (2006) Ethnicity and Child Poverty, Report for the Ethnic Minority Employment Task Force.

Sutherland, H., T. Sefton and D. Piachaud (2003) Poverty in Britain: The Impact of Government Policy Since 1997, Joseph Rowntree Foundation, York Publishing Services Ltd, York.

UNICEF (2007) An Overview of Child Well-Being in Rich Countries, Innocenti Report Card No. 7, UNICEF Innocenti Research Centre, Florence.

Footnotes

¹ Acknowledgements
Thanks to Alari Paulus for assistance. This paper is based on a keynote address to the Social Policy, Research and Evaluation Conference, 3–5 April 2007, Wellington, New Zealand.
Correspondence
Professor Holly Sutherland, Institute for Social & Economic Research, University of Essex, Colchester, Essex, United Kingdom CO4 3SQ.

² "My policies are based not on some economics theory, but on things I and millions like me were brought up with: an honest day's work for an honest day's pay; live within your means; put by a nest egg for a rainy day; pay your bills on time; support the police" The News of the World (20 September 1981).

³ Here and throughout, unless otherwise stated, poverty is considered to be living in a household with equivalised income below 60% of the national median.

⁴ These are agreements between HM Treasury and the spending Departments in the UK.

⁵ DWP (2006) Tables B1 and 9D.

⁶ These figures should be treated with caution. Child poverty is measured at the EU level for children aged 15 and under whereas the UK national estimates also count children aged up to 19, if in full-time secondary education. The 1998 estimates make use of the European Community Household Panel data collected in 1999 for all EU15 countries whereas the 2003 estimates use EU-SILC data for 12 of the EU25 countries and various national sources for the remainder (including the UK). Thus differences across time or between country may be due to non-comparability of data sources as well as actual differences in child poverty rates.

⁷ And, indeed, making "Education, education, education" the top priority for the incoming Blair Government pre-dates the child poverty reduction commitments (Blair's 1996 Labour Party Conference speech).

⁸ Lone-parent employment rises from 56% (in 2005) to 67.5% and worklessness in couples with children falls from 4.9% to 4.5% (Gregg et al. 2006, Brewer, Browne and Sutherland 2006).

⁹ This section draws heavily on joint work with John Hills.

¹⁰ Potentially the adjustment could be by a price index specific to the group receiving the benefit. For instance, the aim may be to allow a pensioner to continue to purchase the same basket of goods, in which case the relevant price index is for that basket – or for a proxy, such as the "pensioner prices index".

¹¹ Brewer, Goodman et al. (2006) also point out that the recording of CTC receipt in the Family Resources Survey data – used for most of the analysis reported in this paper – appears to be subject to non-response: thus the child poverty rate may in fact be lower than that measured using these data directly.

¹² This is on the basis of measures using income After Housing Costs (AHC). Most international comparisons and the rest of the analysis in this paper use Before Housing Costs (BHC) income measures. In London, BHC poverty rates are artificially depressed because of the inclusion of Housing Benefit in the BHC measure. High rents result in high Housing Benefit entitlements, making London incomes also seem high relative to nationally. But the benefit is entirely spent on the higher rents. AHC measures of income deduct rent from income.

¹³ For more explanation of these calculations see Corak et al. (2005) and European Observatory on the Social Situation (2005).

¹⁴ However, note that tax concessions are not included in this measure and the effect of benefits is included gross of taxes. From Figure 3 we can see that taking account of taxes is likely to have a small effect on incomes in the bottom 20% of households only in Belgium, Italy, Spain and the Netherlands (positively) and Finland (negatively).

¹⁵ Another factor is the fact that private pensions – making up a relatively large proportion of Dutch pension income – are counted as primary income in this analysis.

¹⁶ See also Levy et al. (2007), who test this explicitly.

¹⁷ http://www.jrf.org.uk/child-poverty/