The Geography of Poverty and the Geography of Welfare Benefits

Data on welfare benefits are widely used in research and public administration to describe spatial variations in the prevalence of poverty in the UK. Many poor households, however, receive no benefits, and not all benefit recipients are income-poor. Are statistics on benefits receipts, then, really good proxies for describing the geography of poverty?

The following text is slightly adapted from an article that apppeared in the CASE 2012 Annual Report

Some current uses of benefits as poverty measures

Since the late 1990s, administrative data on the receipt of welfare benefits have become widely used as proxy measures of income poverty and deprivation. For example, the rates of receipt of Income Support are part of the elaborate formulae that allocate central financial support to local government districts. This is an adjustment for the level of “need” taken to arise from the varying proportion of low-income households in different council areas. Another example is the neighbourhood deprivation indices of the various countries within the UK. These are taken up by policy analysts and academic researchers as an authoritative statement of which neighbourhoods have the highest rates of deprivation. All the current official indices rely heavily on welfare benefits data to represent spatial differences in income poverty. Estimates of poverty based on benefits data are available more quickly and, most importantly, at smaller spatial scales (such as neighbourhoods, districts and cities) than traditional measures, which derive from large sample surveys such as the Family Resources Survey. As one part of the Social Policy in a Cold Climate programme, CASE has been investigating the implications of using welfare benefits data to characterise differences in rates of poverty and deprivation between parts of Britain.

Potential problems of benefits data as poverty proxies

The basic reasoning behind using the receipt of benefits as a poverty indicator is plain: recipients of means-tested benefits, such as Income Support, normally de facto have an income that leaves them income- poor by conventional definitions, which set thresholds relative to national median income. However, these uses invite the question of how closely, in fact, the spatial distribution of poverty is represented by that of benefit receipt. So, for a start, are recipients of various benefits always income-poor by standard definitions, based on a conventional poverty line of 60 per cent median income? What percentage of income-poor households receive no welfare benefits, because they are ineligible, unaware, or choose not to claim? Does this mean that benefits data are biased proxies – that is, they lead us systematically to over- or under-estimate the prevalence of poverty in different types of place? Is there a danger that the social problem of poverty becomes misunderstood as synonymous with the welfare benefits system? CASE’s work in 2012 has looked at some of the questions. We do this by first testing the validity and coverage of benefit receipt as poverty indicators within the best available source, the Family Resources Survey, and then by comparing survey estimates of poverty for different places with the rates of means-tested benefits receipt therein.

How valid are benefits as poverty proxies?

The validity of a poverty proxy (such as receiving a welfare benefit) is the proportion of households who are in fact poor. The coverage of a proxy is the percentage of all poor households who are identified by the proxy. The summary results (Table 1) are revealing.

As one might expect, a sizeable majority of working-age benefit claimants do have incomes so low that they are income-poor by conventional and international standards. However, people who report receiving such a benefit to the survey represent only a minority of all income-poor families in the UK, around 25 per cent. If we look at households receiving other means-tested transfers, such as Pension Credit or Working Tax Credits, the proportion rises, but a greater number of these families have low incomes that are marginally above the standard poverty line.

What proportion of the poor receive benefits?

These coverage figures are under-estimates, as receipt of welfare benefits is under- reported in sample surveys such as the FRS. However, they do show that a very large proportion of the income-poor in Britain receive no transfers from the state. The factors affecting the relationship between income transfers and poverty – such as housing costs, wages and eligibility for benefits – may vary from place to place, and so there is a risk that administrative data proxies may be biased estimates of the relative incidence of poverty in different parts of the country. This is borne out when we compare regional data on means-tested benefits to regional survey estimates of income poverty (Figure 1).

Region key: EE: East of England; EM: East Midlands; LO: London; NE: North East; NW: North West; SC: Scotland; SE: South East; SW: South West; WA: Wales; WM: West Midlands; YH: Yorkshire & Humber.

Notes: The “household benefit-receipt rate” is the count of claimants of the four major means-tested benefits, divided by the number of households in the region. The income poverty rates are single-year estimates from survey data, and the approximate 95 per cent confidence intervals are shown by the bars above and below the central estimate. The best-fit line is a linear regression weighted by the household count in each region.

Sources: Means-tested benefit counts from DWP, via NOMIS; Household estimates from DCLG (England), GROS (Scotland) and WAG (Wales); Income poverty rates are author’s calculations from Households Below Average Income.

The relationship between regional benefit rates and poverty rates is not one-to-one, and varies significantly between places. In London, for example, the “real” rate of poverty, once housing costs are taken into account, is consistently higher than the rate of means- tested benefit receipt would imply. We would expect such inconsistencies in the relationship between benefits and poverty to be even more marked for smaller spatial units, such as local authority districts.

Conclusions

We can draw several conclusions from these results. For public policy, they confirm that the political question of poverty is by no means reducible to the matter of state support, as a majority of the poor receive no benefits. Where benefits data are used to allocate funding, programmes should be clearer about the allocationary principle at play, and more cautious about how well administrative data correspond to income poverty. Similarly, academic researchers should be alert to potential biases that can arise from using such data to describe variation in poverty within and between cities and regions; there is an important role for modelled and simulated small-area poverty estimates. Nonetheless, benefits data remain a vital source for understanding the spatial distribution of poverty in Britain, and CASE will be conducting further analysis to 2014 to see how this distribution is affected by the Coalition’s social policies and the prolonged economic downturn.