For decades, the image of affluent suburbs and deprived city centres has dominated the popular impression of the modern city. Understanding this pattern – and household location choice more generally – is essential for urban planners, especially for addressing such issues as inequality and segregation within a city.
Urban economists have suggested that the pattern arises because housing prices are lower in the suburbs and richer households prefer and can afford bigger homes (Alonso 1964, Mills 1967, Muth 1969). Assuming most people work in the centre (sometimes called the central business district, or CBD), house prices per square metre have to be lower in the suburbs in order to compensate for longer commuting costs. Richer households therefore choose to locate in the suburbs because they can get bigger homes at a lower price and they push the poor into the centre by outbidding them.
But recent figures show a growing number of poor are locating in the suburbs.1 In some cities, such as Paris, the centre is dominated by high-income households. It is clear that economic factors alone cannot explain where household choose to locate within a city.
In a new paper (Cuberes et al. 2019), we explore how multiple factors influence household distance from the CBD in the eight largest English cities outside London (Birmingham, Bristol, Leeds, Liverpool, Manchester, Newcastle, Nottingham and Sheffield) from 2011 to 2016 by combining data from a number of sources. We use data from the Ordnance Survey, UK council tax records, UK Crime Stats, the 2011 UK Census and the UK Household Longitudinal Study (UKHLS), a large nationally representative survey of around 40,000 households.
Ours is the first UK study to bring together such detailed information including both individual/household characteristics (such as age, gender, level of education, immigration status, car/house ownership, and income) and neighbourhood characteristics (such as amenities, crime rates, council tax rates, and supply of social housing). The dataset contains detailed geographical information that allows us to make a close approximation of household distance from the CBD and match individual household characteristics to neighbourhood characteristics.
Figure 1 shows mean income by distance from the CBD (in quintiles, and with the corresponding fitted quadratic line) for each city. This reveals that distance and household income are strongly correlated without taking account of cities’ amenities and household characteristics. The figure also shows that there are substantial differences across cities – for example, there is a very positive pattern in Birmingham and a strongly negative one in Sheffield.
Figure 1 Spatial distribution of mean household income, by city, with fitted quadratic line
The majority of our amenity measures come from Ordnance Survey Point of Interest data. These include (i) public transport access points, such as bus and tram stops, and train stations; (ii) public services, such as schools and hospitals; (iii) historical and cultural attractions, such as historic buildings and museums; (iv) retail services, such as shops and department stores; (v) facilities for eating out, such as restaurants, cafes and public houses; (vi) sport facilities, such as leisure centres and gymnasiums; (vii) outdoor recreational facilities, such as commons, parks and playgrounds; public transport, retail outlets, and the incidence of property crime; (viii) the age of housing stock; (ix) the amount of social housing; and (x) the household’s Council Tax Band.
Figure 2 shows plots of these amenities by distance quintiles and clear spatial patterns emerge. The main takeaway from this figure is that most amenities are more prevalent closer to the CBD. Public transport access (graph 1) is highest in the CBD and sharply decreases with distance. All of our other positive amenities (graphs 2-7) are concentrated in the CBD, though there are signs of increased numbers at the periphery compared with the middle quintiles.
The share of old housing declines with distance, but with some increase in the furthest quintile (graph 8). Property crime is highest closest to the CBD (graph 9), and the share of social housing (graph 10) also declines steadily with distance. The average Council Tax band displays the least clear pattern since the lowest bands are in quantiles 2 and 4 with the highest in the furthest quintile.
Figure 2 Spatial distribution of amenities, with fitted quadratic line
To analyse household location more systematically, we estimate regression models at the household level, of distance to the CBD on the set of household characteristics explained above and the set of neighbourhood amenities, while controlling for city-specific differences (like geographical size and the specific physical geography of each city).
We define the CBD of a city by the location of its main railway station. But we run numerous checks with different definitions and obtain very similar results. Moreover, we also allow for the possibility of multiple areas of town centre activity within a city and our results are also robust to this specification.
Our results can be summarised as follows: there is a positive association between household income and distance from the CBD but this is reduced substantially, in both size and significance, when we introduce household characteristics and amenities into the model. The inclusion of household characteristics alone reduces the income coefficient by around two-thirds (from 0.169 to 0.055). When we add all the controls simultaneously, the income coefficient is small and in fact is not statistically significantly different to zero, meaning that no systematic relationship between income and distance remains once we take account of household characteristics and neighbourhood amenities.
All household characteristics (except owner-occupier status) are associated with distance from the CBD. Households with a male head, older people, and car owners tend to live further from the CBD, while those with higher education, single person households, and migrants live closer. The effect of migrant status is particularly large, suggesting that on average migrants live 25% closer to the CBD than non-migrants; this compares, for example, with 10% closer for household heads with higher education.
The association between amenities and households’ distance to the CBD is largely negative, reflecting the fact that amenities tend to be more available closer to the centre. For example, the negative relationship between public transport and distance to the CBD is a similar finding to Glaeser et al. (2008), who show that in the US reliance on public transit generally declines sharply with distance from the city centre.
Our results suggest that an increase in ten public transport access points (such as bus or tram stops) is associated with a 2% reduction in distance from the CBD. The exception to the negative relationship are the amenities provided by the home itself (proxied here by the Council Tax band) as larger homes are more prevalent in the suburbs.
Our results also reveal some important differences from the US evidence that has dominated this literature. Migrant status is important in England, and, in all of our cities except Bristol, migrants live much closer to the CBD than non-migrants (although this tendency is attenuated the higher their income), but race per se is not important to household location in England.
In addition, it appears that in England only the employed (and those above the poverty line) are influenced by the availability of public transport; the location of the not-employed (and the poor) does not seem to depend on public transport access. This is in direct contrast to the US evidence of Glaeser et al. (2008) and Pathak et al. (2017). This is an important finding and may suggest that the unemployed in England are less likely to be involved in active job search either due to better welfare benefits in the UK than in the US, or because they are constrained geographically.
Authors’ note: The UK Household Longitudinal Study (UKHLS) is an initiative by the Economic and Social Research Council, with scientific leadership by the Institute for Social and Economic Research, University of Essex, and survey delivery by the National Centre for Social Research. National Grid Reference information for UKHLS was accessed via the Secure Data Service. Neither the original data creators, depositors or funders bear responsibility for the further analysis or interpretation of the data presented in this study. University of Essex, 2016. Understanding Society: Waves 1-7, 2009-2016 8th Edition. Institute for Social and Economic Research, NatCen Social Research, Kantar Public. UK Data Service. SN: 6614.
Alonso, W (1964), Location and Land Use: Toward a General Theory of Land Rent, Harvard University Press.
Cuberes, D, J Roberts, and C Sechel (2019), “Household Location in English Cities”, Regional Science and Urban Economics 75: 120-135.
Glaeser, E L, M E Kahn and J Rappaport (2008), ‘Why Do the Poor Live in Cities? The Role of Public Transportation’, Journal of Urban Economics 63(1): 1-24.
Mills, E S (1967), “An aggregative model of resource allocation in a metropolitan area”, American Economic Review 57(2): 197-210.
Muth, R F (1969), Cities and Housing. The Spatial Pattern of Urban Residential Land Use, University of Chicago Press.
Pathak, R, C K Wyczalkowski and X Huang (2017), “Public Transit Access and the Changing Spatial Distribution of Poverty”, Regional Science and Urban Economics 66: 198-212.
 See https://theconversation.com/poverty-is-moving-to-the-suburbs-the-question-is-what-to-do-about-it-35986.