Peter Batey, Peter Brown, University of Liverpool, Liverpool, United Kingdom
Alternative approaches to the spatial targeting of urban policy - a comparative analysis (assigned to theme
Area-based urban policy initiatives, involving the channelling of resources to selected neighbourhoods within cities, represent a major component of urban regeneration activity, in Britain and elsewhere. For more than thirty years spatial targeting methods have been used to define areas of greatest need. Such methods have ranged from the use of statistical measures derived from census statistics, including the so-called Z score Index of Deprivation (1980s) and the Signed Chi-square Index of Local Conditions (1990s), through to the latest use of government operational statistics (potentially capable of being updated more frequently than census data) to construct an Index of Multiple Deprivation (2004). Geodemographic systems, in which neighbourhoods are classified according to their demographic, social and economic characteristics, are another widely-used targeting method. The precision of the spatial targeting has varied too, from the use of relatively small census enumeration districts and output areas at one end of the scale, and larger electoral wards and super output areas at the other.The purpose of this paper is to present a comparative empirical analysis of the results obtained from different approaches to spatial targeting. Two methodological approaches are employed: the Index of Multiple Deprivation 2004 (IMD 2004) which makes use of super output areas, and a geodemographic system based on the 2001 Census - People and Places - which is available at both output area and super output area levels. The measure of spatial targeting performance is drawn from the 2001 Census and consists of a single variable measuring household deprivation in up to four dimensions. This measure has the advantage that it refers to individual households and is therefore not subject to the usual problems associated with the ecological fallacy encountered in using aggregate data. This measure is expressed as a proportion of all households in the area in question. Performance is compared in each case using a Lorenz curve and Gini coefficient. The best performance is defined by a ranking of individual output areas according to the performance measure, while the worst performance is given by a uniform distribution in which the level of household deprivation is constant from one area to another. The comparisons take several different forms but all of them use national data sets for England and Wales. First there is a direct comparison between the geodemographic and IMD 2004 systems, using each system at its finest spatial level, output area level for People and places and super output area level for the IMD 2004. Differences in the results of targeting here will be a product of three factors: methodological approach, choice of data, and level of spatial aggregation. In the second and third analyses an attempt is made to establish the relative importance of two of these factors; in one case a comparison is made between the results of IMD 2004 and People and Places at the the super output area level (here differences will be due to a combination of data and method), and in the other case the output area and super output area level results of using People and Places are compared (differences here will be due entirely to the effect of spatial aggregation). On the basis of these comparisons, it is possible to draw conclusions about the relative merits of each targeting approach, and to understand more fully the trade-offs between data, spatial aggregation and updateability. The paper compares these results with the output of related research conducted by the authors that is aimed at measuring success in targeting specific urban policy initiatives.
 | Conference organized through conf-vienna
(copyright Gunther Maier) |