Select Committee on Work and Pensions Written Evidence

Memorandum submitted by the Institute of Fiscal Studies (CP 19)



  Using the Government's preferred definition; child poverty is at its lowest level since 1989, at 3.8 million children. But if the current rate of decline in child poverty persists for three more years, the Government will miss its child poverty target for 2004-05 by around 300,000 children. Our current estimates suggest that meeting this target through increases in the per child element of the child tax credit could cost up to £1.7 billion a year on top of the already substantial increases in spending on families with children enacted by the Labour government.

  This perhaps slower than anticipated decline in child poverty illustrates how difficult—or, at least, how costly—it is to reduce poverty when economic growth is strong, and incomes of households in the middle of the income distribution are growing quickly.

  It is important that a focus on a single poverty line does not obscure the changes in the overall distribution of income for households with children. Under the current Government, real incomes have risen at (almost) all points of the distribution for individuals in households with children.

  Less heartening, estimates of the poverty gap have increased since 1996-97. This is due in part to the significant proportion of the poorest individuals in the HBAI survey who are not receiving means-tested benefits. The incomes of these individuals have fallen further behind the poverty line as incomes across the rest of society have risen. Though in many cases this will reflect genuine need, some of these individuals may not be as poor as they seem, either because they are experiencing incomes that are very low only in a very transitory sense, or they because have their incomes measured with error by the FRS survey. Whatever the reason, their presence make it very unlikely that child poverty could ever be abolished on the Government's currently favoured definition.

  Achieving a definition of child poverty that is acceptable to government, accords well with popular opinion, and can be accurately and objectively quantified is probably impossible. There is a risk that the current high-profile target to reduce poverty defined by low incomes is skewing the policy response away from improved public services for children towards increased cash transfers to their parents. For this reason, we would prefer the Government not to set targets for reducing child poverty exclusively but to also set "floor" targets for what it regards as the most important outcomes for children.


  1.  This note focuses on what has happened to the incomes of households with children since 1996-97, what the Government needs to do to meet its child poverty target in 2004-05, and how the Government should measure child poverty. It summarises and builds on analysis in M. Brewer, T. Clark and A. Goodman (2003), "What Really Happened to Child Poverty in the UK Under Labour's First Term?", Economic Journal, 113 (June), pp F240-F257, and M Brewer and G Kaplan (2003), "What do the child poverty targets mean for the child tax credit?", in IFS Green Budget 2003,


  2.  The Government has a specific, quantified, target for 2004-05: "the target for 2004 will be monitored by reference to the number of children in low-income households by 2004-05. Low-income households are defined as households with income below 60% of the median, as reported in the Households Below Average Income (HBAI) statistics . . . Progress will be measured against the 1998-99 baseline figures and methodology."[219] It is, though, unclear whether the Government would prefer to measure income before housing costs (BHC), or after housing costs have been deducted (AHC)[220]. In practice, ministers' statements have tended to focus on progress on the AHC measure, and we focus on progress towards that particular target in this note.[221]

  3.  There were 4.2 million children in poverty in 1998-99 on this definition, so there will need to be fewer than 3.2 million children in poverty in 2004-05 for the Government to meet its target. The level in 2001-02 is 3.8 million, down 400,000 since 1998-99, continuing the downward trend that began in 1996-97 (see Table 1. Figures 1 and 2 show the proportion of children in poverty under these same measures since 1979.222).

  4.  It is possible to calculate whether the changes in child poverty are statistically significant. Almost all year-on-year changes are not statistically significant, meaning that they could merely have reflected sampling variation.223 But the cumulative decline in child poverty since 1996-97 and since 1998-99 is comfortably statistically different from zero.



  5.  Though not directly targeted by government, there have also been declines in the numbers below 50% and 70% of the median AHC income.

  6.  When poverty is measured BHC, the pattern of changes over the most recent years varies depending on the poverty line chosen: on the targeted measure (the number below 60% median BHC income) child poverty remained unchanged in the last year of the data, but has fallen by 500,000 since 1998-99 (a drop of around 15%) (see Table 2). Looking at the other two poverty measures shown in Table 2, child poverty fell in the last year of the data on one (the number below 50% of the median), and rose on another (the number below 70% median). All BHC measures, though, show falls in child poverty since both 1996-97 and 1998-99.

  7.  This general pattern of poverty changes suggests that Government policies have not just moved people from just below to just above an arbitrary poverty line; instead, the numbers below all the poverty lines monitored have been declining.

  8.  The available data now spans the first three years of the six-year period highlighted by the Government for its PSA target (1998-99 to 2004-05). The target is for a reduction of a quarter in child poverty, and so far, measuring income AHC, child poverty has fallen by 10%. This means that the Government needs to reduce child poverty by more in the second three years (2001-02 to 2004-05) than it has managed so far.

  9.  We can compare the observed decline in child poverty since 1998-99 to a hypothetical path to the target in which child poverty declines by an equal amount in every year: see Figure 3. This shows that the fall in child poverty to date has been slower than this hypothetical path, and so the Government is "behind schedule" in this sense. If the current rate of decline in child poverty persists for three more years, the Government will miss its child poverty targets for 2004-05 by around 300,000 children. However there is no reason why the Government should reduce child poverty equally in six consecutive years.

  10.  Looking across a number of poverty lines, the Government is having more success in reducing poverty for lower poverty lines. Although there is no explicit target for the number of children below 50% and 70% of the median, the Government is considerably more than halfway towards reducing the number of children below 50% median by one quarter (both AHC and BHC), but a long way off this in reducing the number whose incomes are below 70% (see Tables 1 and 2).[224]


  11.  Analysis carried out by researchers at the IFS in January 2003—before the HBAI data for 2001-02 was available—suggested that taking into account tax and benefit changes already announced, but yet to be reflected in the HBAI data, the Government would fall short of meeting its child poverty target measured AHC.[225]

  12.  That analysis has not yet been fully updated to reflect data from 2001-02, though we plan to do this soon. In the meantime we can reach some tentative conclusions. Based on data from 2000-01, our central estimate was that the Government would miss its target in 2004-05 by 200,000 children. Between 2000-01 and 2001-02, child poverty fell by just 100,000, and this means that the Government is now further behind schedule to meet its target in 2004-05. This means that the Government is likely to miss its target by a larger amount than our original research based on data from 2000-01. We now expect that it will miss its target by somewhere between 200,000 to 400,000 children.

  13.  What measures might be needed to help the government hit its child poverty target in 2004-05? In earlier work, we estimated the number of children that would be taken out of poverty by various increases in the per-child element of the child tax credit. We assumed these increases would take place in April 2004, and would be additional to the increase in line with earnings growth which has already been included in the public finance forecasts. We compared the use of the per child element of the child tax credit to child benefit or the family element of the child tax credit. The results are shown in Table 5.[226]

  14.  Our estimate that the Government will fall short by between 200,000 and 400,000 means that the Government is now likely to miss its target even if it raises the per child element of the child tax credit by £3 a week above average earnings growth. To be sure of meeting the target, we estimate that it could need to raise the per child element of the child tax credit by £4, or if the shortfall is at the upper end of the range, by £5 a week above average earnings growth. An increase in the per child element of the child tax credit of £5 a week would cost the exchequer £1.7 billion a year.

  15.  Reducing child poverty through increasing child benefit or the family element of the child tax credit could cost around twice as much because they would also benefit well-off families with children, and not just those on low-to-middle incomes.[227] Given the structure of the new tax credits, the child element of the child tax credit is one of the most efficient ways for the Government to hit its child poverty targets. Another reform which would offer a relatively efficient way to reduce child poverty is increasing the adult rates of income support and JSA (income-related). This change would benefit those lone parents and couples with children who are out of work and in poverty, but we think it unlikely that the Government would pursue this option because of the knock-on consequences for those of working age without children (and, perhaps, for those aged 60 or over). Indeed, the adult rate of income support and JSA(income-related) has not been increased in real terms since its introduction in 1988.

  16.  There may be factors, though, other than the cost to the Treasury and the need to reduce child poverty for the Government to consider when considering whether to increase the per child rate of the child tax credit. For example:

    —  has the Government got the balance between spending on children and spending on adults right? Government spending on transfers to parents has increased by over 50% in real terms since 1997, a rate of increase that is unprecedented historically.[228] The poorest families now receive more for their first child than a young job-seeking adult is entitled to receive for him or herself.[229]

    —  what will be the impact on work incentives? Increases in the child tax credit have negative effects on parents' incentives to work; these could be offset with changes to the working tax credit, but this would increase the cost.[230]

  17.  As we discuss below, though, by setting a target that defines a child as poor on the basis of its parents' incomes, there is a risk that the Government will simply increase spending on tax credits when the money might be more effectively spent on public services for children.


  18.  It is important that the focus on the Government's target for 2004-05 does not obscure the changes in the overall distribution of income for households with children.

  19.  IFS research has shown that the reduction in child poverty is unambiguous in the sense that there are no sensible poverty lines defined as fractions of the median by which poverty has increased.[231] In fact, real incomes have risen under the current Government at (almost) all points of the income distribution for households with children. This means that there have been genuine and substantial increases in incomes for almost all households with children, including many of the poor.

  20.  However, although incomes of households with children have risen across the distribution, and the number of children in poverty has fallen, the average poverty gap amongst households with children increased between 1996-97 and 2000-01 on a number of poverty lines (see Table 7)[232].[233]

  21.  One reason for this increase in the poverty gap is that the "richest of the poor"—ie those on means tested benefits—have seen the biggest increase in their incomes, with many moving above the poverty line. By contrast, income growth amongst the "poorest of the poor" has been relatively low. Part of this reason is that not all of the poorest children have been gaining from the very large increases in means-tested benefits and tax credits: less than 60% of children with incomes below 40% of median income are in households in receipt of a means-tested benefit. Instead, a relatively large proportion of these children are in households where at least one member is full-time self-employed, and many are in families where someone is in work.

  22.  If we could be sure that the FRS survey (the basis for the HBAI data-set) was recording incomes accurately, then we should be very concerned at this apparent problem of acute poverty associated with non-take-up of benefits. But this is not the case. Non-take-up of means-tested benefits certainly exists, and some children classified as poor by HBAI are in households awaiting the outcome of a claim for a means-tested benefits, suggesting that administrative delays make their contribution to child poverty. But there are a group of children in the HBAI dataset in households who appear poor and are not receiving means-tested benefits that may be less of a policy concern, for two reasons:

    —  The FRS records only snapshots of income, and some "rich" families may look "poor" purely on the basis of a reported weekly income which is low or even zero temporarily. Although this group may be entitled to claim benefits during their short period of low incomes, they are less of a public policy concern than those with more permanent low incomes.

    —  There may also be some in the HBAI data whose incomes are wrong simply because their incomes have been reported, or perhaps recorded inaccurately.

  23.  Whatever the cause (which we believe merits further research), there are two implications for the Government's child poverty target:

    —  there are 1.2 million children in households in measured poverty but not receiving the main means-tested benefits, and this places a real limit on the ability of tax and benefit increases to reduce child poverty significantly in the future;

    —  some of these children would probably not be counted objectively (should such a thing be possible) as poor.

  24.  Overcoming these problems would involve either changing the HBAI methodology, or further progress on finding a better measure of child poverty.


  25.  The Government has yet to specify how we might tell whether poverty has been abolished in 2020, and what measure of poverty is due to be halved by 2010. In the meantime, the authors have observed civil servants attempting to argue that it is not intellectually incoherent for the Government to be committed to halving something that it does not yet know how to measure. In our view, though, if the Government wishes to set meaningful, quantifiable targets, then it must be explicit on how it is going to measure progress.

  26.  We do not have views on what the measure of child poverty should be adopted by the Government: this is inherently a political decision on which we do not feel able to give advice. But we would like to make some observations on:

    —  the choice between relative and absolute income measures of poverty;

    —  measuring material deprivation;

    —  the distinction between defining child poverty and setting government targets.


  27.  Despite unprecedented increases in spending on transfers to households with children, child poverty remains high, on a historical basis, on the Government's preferred measure. This is an excellent reminder of how difficult—or, at least, how costly—it is to reduce poverty defined in relative terms when economic growth is strong, and the incomes of households in the middle of the income distribution are growing quickly (median household income grew by 15% in real terms between 1998-99 and 2001-02: see Table 6). If the Government had decided, as is the custom in the US, for example, that poverty is defined by failing to attain a standard of living that does not change over time (in other words, if it chosen a poverty line fixed in real terms, sometimes called an absolute poverty line), then it could have claimed far greater success in reducing child poverty. The number of children in households below 60% of 1996-97 median AHC income has fallen from 4.3 million to 2.5 million, a decline of 42% (see Table 3). However, if the Government continues to believe that poverty is a relative concept, then it should continue to focus on the measures of child poverty that are defined in relative terms. Recent Government press releases are focussing attention on progress measured by an absolute, rather than relative, poverty line, and this is, at best, inconsistent with its stated goals.[234]


  28.  The DWP's initial conclusions to the responses to its consultation on child poverty included stated that "there was a lot of support for including some measure of material deprivation. This is a technically complex area and we will carry out further methodological work and discussion with experts to try to reach agreement on the most appropriate methodology and some of the more subjective areas."

  29.  A definition of poverty that captures material deprivation is attractive in the sense that it accords more closely with the popular perception of poverty than a definition based on income alone, and for this reason it can be a powerful tool for capturing the public imagination. However, we do not think that it is appropriate for the government to monitor or target it as an official measure. In particular those measures that have attempted to measure material deprivation rely on subjective concepts such as self-reported inability to afford particular items; they are forced to make a number of arbitrary judgements, such the number of necessities one needs to lack to be considered poor; the methods for determining which items are necessities, and how this is up-graded over time as society changes can also be criticised. This means that these studies have been perhaps less transparent than the work that lies behind the creation of the HBAI data-set and report.[235]


  30.  In our view, part of the reason that the Government wants to tackle child poverty is that children in poor households experience a different sort of childhood to those not in poor households, and that this experience is overwhelmingly a negative one, with persistent impacts. Indeed, the fact that poverty makes you "different" was one of the messages that arose from the DWP's consultations and workshops with children.

  31.  The policy approach encapsulated by the Government's PSA target for 2004-05 is, effectively, to reduce the inequalities in the incomes of households with children, particularly at the bottom end of the income distribution, by direct redistribution and the promotion of paid work. An alternative approach, however, would be to move the focus away from one exclusively based on income, and towards one which also takes into account children's outcomes.

  32.  For example, the Government could set targets for the children's outcomes about which it is most concerned—such as those set out in Opportunity for All—rather than just for their parents' incomes. Ideally, such targets would be expressed so that they would be designed to reduce the inequality in children's outcomes, not simply the average, or the outcomes for the high-attaining minority.

  33.  Such an approach would reduce the incentive within government to direct money towards boosting parental incomes at the expense of improving services (including schools, health and housing) for poor families with children. Rather, the incentive would be to channel money through whichever means improves children's outcomes the most effectively. To choose a simplistic comparison of the by way of example, the cost of increasing the per child element of the child tax credit by £5 a week would allow the Sure Start scheme to be more than tripled from its coverage in 2003-04, but expanding Sure Start is unlikely to have any impact on the HBAI measure of child poverty (at least in the short-run).

  34.  Which policies improve children's outcomes most effectively is an important research question. At the moment we have little research evidence on the relative effectiveness of income supplementation compared to improving public services (eg reducing class sizes). However the body of research evidence from the US (taken as a whole) suggests that the direct impact of additional income paid to poor parents on children's outcomes such as educational attainment and health is relatively small[236].

  35.  An approach which sets targets across a number of outcomes would emphasise the role of the public services in performing a redistributive function—for example, by continuing to spend more money on schools in deprived areas, so that the education outcomes of poor children improve relative to those of rich children.

219   HM Treasury (2002), "Technical Note for HM Treasury's Public Service Agreement 2003-06, The median household is the one for which half the rest of the population has an income higher than it does and half has an income lower. Back

220   It does not specify whether self-employed households are to be included. Back

221   For example, HM Treasury (2001), "Budget 2001: Investing for the Long Term", p.87, Box 5.3. Back

222   Comparable figures from 1994-05 to 2000-01 can be found in M Brewer, T Clark and A Goodman (2002), The Government's Child Poverty Target: How Much Progress Has Been Made?, IFS Commentary 88, or M Brewer, T Clark and A Goodman (2003), "What Really Happened to Child Poverty in the UK in Labour's First Term?", Economic Journal, June 2003, vol 113 no 488, pp F240-F257. Back

223   See Brewer, Clark and Goodman (2002), op cit for full results. Bootstrapping involves recalculation of statistics for each of a series of random samples drawn from the original sample, as a way of approximating the distribution of statistics that would be calculated from different possible samples out of the underlying population. In this way, the range of effects of sampling variation is approximated. For more on this methodology, see Davison and Hinkley (1997), Bootstrap methods and their application. Cambridge: Cambridge University Press. Back

224   These poverty measures are monitored in Opportunity for All, the government's official audit of poverty and social exclusion: see Department for Work and Pensions (2002), Opportunity for All: Fourth Annual Report 2002, Cm 5598, London: HMSO. Back

225   This analysis estimated the impact of tax and benefit changes due to be implemented between 2001-02 and 2004-05. See M Brewer and G Kaplan (2003), "What do the child poverty targets mean for the child tax credit?", in IFS Green Budget 2003, That work also concluded that the government would hit its target exactly measuring incomes BHC, based on data from 2000-01, but child poverty measured BHC incomes did not change between 2000-01 and 2001-02, and this suggests that hitting the target in 2004-05 measured BHC is now a little less likely. Back

226   This analysis is based on the income distribution in 2000-01, up-rated to approximate the income distribution in 2004-05. More recent data is now available, but we do not think that it would change the results of the analysis substantially. Back

227   In approximate terms, the poorest half of families with children would gain from increases in the child element of the child tax credit. All families with children would gain from an increase in child benefit, and am increase in the family element of the child tax credit would benefit all but the richest half a million families with children. Back

228   See S Adam, M Brewer and H Reed, The Benefits of Parenting: Government Financial Support for Families with Children Since 1975, Commentary no 91, IFS, London, 2002 ( Back

229   The weekly jobseeker's allowance (JSA) for an adult under 25 will be £43.25 from April 2003, compared with £54.25 in child-related payments that will be received by a non-working family on income support or JSA for their first child. This gap will widen as the child element of the child tax credit increases in line with earnings growth. Back

230   There are two impacts: increases in any part of the child tax credit reduce the need for parents to work; and increases in the per-child element of the child tax credit widen the range of incomes over which parents' marginal effective tax rate increases by 37%, which reduces the cash gain from working more. Back

231   See Brewer, M Clark, T and Goodman, A (2003), op citBack

232   The poverty gap is difference between the poverty line and household income, where this is positive. The poverty gap can be scaled by dividing it by the poverty line multiplied by the number of poor households, and we call this the "poverty gap ratio". Back

233   See Brewer, M Clark, T and Goodman, A (2003), op citBack

234   See, for example, HM Treasury Press Release 101/03, and DWP Press Release entitled "Lower Incomes Rising Fastest", 13 March 2003. Back

235   We should declare a vested interest: for a number of years, IFS researchers have worked with officials from the DWP each year to check their processing of the HBAI data-set. Back

236   See Mayer, S (1997), What Money Can't Buy, Cambridge: Harvard University Press. Back

previous page contents next page

House of Commons home page Parliament home page House of Lords home page search page enquiries index

© Parliamentary copyright 2004
Prepared 22 January 2004