Local child poverty estimates are difficult, but essential to expose the stark realities of geographic inequality
Which of the following statements tells you more?
Around 4 million of Britain’s 14 million children live in households classified as in poverty because they have below 60% of median income after housing costs.
Among the 2,200 children who live in the Notting Barns area of Kensington, site of Grenfell Tower, nearly a thousand are in families with very low incomes. Just over half a mile away, among the 2,200 children living in three wards around Kensington High Street and Cromwell Road, only 150 are in this situation.
In fact each statement is useful: the first shows the overall extent of child poverty and the second what it looks like on the ground.
For the past 15 years, I have helped produce maps estimating where child poverty is most concentrated in the UK. These local child poverty figures are not just designed to shock, although they regularly do. They also show local authorities and others in which locations children face the double disadvantage of family poverty and area poverty. These children live in places where a lack of material resources and opportunities can worsen the effects of growing up in a socially and economically disadvantaged household.
value of these figures is clear – they could help ensure services are targeted where they’re most needed, for example. However, capturing the extent of income poverty in local areas is a highly imperfect process. To produce our local figures, we use two different data sets in combination to get the best local estimates. In response to our most recent figures, which reflected the clear evidence that child poverty is getting worse, the government has produced figures claiming to show that in fact child poverty is falling. But it has done so using raw data from a single source which is highly unsuitable for tracking changes in child poverty rates over time.
There are several reasons why local statistics are difficult to produce. National surveys asking people in detail about their incomes are based on samples that are far too small to be able to say anything about incomes at a local level. The best indicators we have of local child poverty come from data held by public authorities on the number of people claiming out-of-work benefits, and the number who claim working tax credits whose reported family income falls below the poverty line.
This so-called “administrative data”, reported regularly by HMRC, certainly gives you a good idea about where the worst-off wards and constituencies are, and some measure of the concentration of low income in these places. However, particularly when tracked over time, they need to be used carefully, in conjunction with other evidence.
One difficulty is that HMRC data assumes that everyone who is out of work is in poverty. In reality, the relationship between being out of work and being in poverty changes over time. The child of a non-working lone parent had around 85% chance of being in poverty 20 years ago, a risk that fell to 58% by 2013, but then rose sharply to 68% by 2015.
Another problem is that HMRC is quite good at assessing family income for tax credit purposes, but not so good at identifying who is in a low-income household (which takes account of a wider range of income than tax credit assessments, including income generated by household members not in the child’s nuclear family). HMRC’s local child poverty statistics only manage to count a third as many people on low working incomes as the full income surveys. This matters a lot, because working poverty has been rising, and out-of-work poverty falling. Two in three children in poverty now have at least one working parent.
An indicator that overcounts out-of-work poverty and undercounts in-work poverty is bound to show current trends (when the latter is falling and the former rising) in a favourable light. This is especially true in London, where in-work poverty, measured after deducting housing costs, is particularly high because of high rents.
And governments, even those like the present one that has underplayed the importance of income poverty, cannot resist highlighting such rose-tinted statistics.
This explains why, after the publication of our latest child poverty map, junior DWP minister and former London deputy mayor Kit Malthouse felt the need to parade data in the House of Commons that seemed to show child poverty in London falling rapidly, and to argue that our data showing the opposite must be mistaken.
But the data he was using, the raw figures produced by HMRC, is flawed in multiple ways. In addition to the limitations I have already pointed to, it has some pretty bizarre aspects – pointed out by HMRC itself in its latest commentary. One is that the measure of median income to which poverty is being compared by HMRC is actually going down, whereas all the DWP’s income distribution analysis shows median income rising. “Falling” median income creates a falling poverty line, and hence a lower child poverty count. Another feature is that HMRC doesn’t at the moment count out-of-work families on Universal Credit as being in poverty, even though it did so when similar families were on tax credits. This directly brings the child poverty count down.
The local data that we produce corrects for these quirks in the HMRC data by calibrating the results with the Households Below Average Income (HBAI) survey results. It uses the national differences between the HBAI and HMRC results, for in-work and out-of-work poverty respectively, as the basis for an adjustment to each of the local results. While this can only be seen as an estimate of what the correct figures actually are at the local level, it is a far more meaningful estimate than using the flawed HMRC figures without adjustment.
The two graphs below show why. In recent years, HMRC figures on their own have shown steady falls in poverty rates, particularly in London, which are not borne out by the HBAI data that measures income more accurately. The data produced for End Child Poverty by my colleague Laura Valadez produces results much more in line with the HBAI data, which show that child poverty is now increasing (except in London before housing costs, where the rate is steady).
Figure 1 – Percentage point change in UK child poverty rate using different indicators
Acronym decoder: Her Majesty’s Revenue and Customs; Households Below Average Income measure, Before Housing Costs/After Housing Costs; End Child Poverty
Figure 2 – Percentage point change in London child poverty rate using different indicators (three-year averages; labels show the middle year)
The Institute for Fiscal Studies forecasts that child poverty will continue to rise at an alarming rate, more than wiping out the considerable falls that took place in the 2000s. This will have very real impacts at the local level, which we will continue to estimate. A government that simultaneously publishes these figures but boasts about its progress on another measure purporting to show the opposite has its head stuck very firmly in the sand.