Creation of a System of Functional Areas for England and Wales and for Scotland
O Feldman, D Simmonds, David Simmonds Consultancy, UK; N Troll, MVA Ltd., UK
The Functional Areas analysis provides a catalogue of possible zoning systems based on identifying areas of relatively high self-containment in terms of the travel-to-work patterns in 2001.
Geographers and planners today place an emphasis on spatial organisation that includes the notion of functional regions ? areas defined by business and economic activities rather than by administrative boundaries. We define the functional regions as regions which have more interaction with each other in terms of commuting than with other regions.
In June 2004 the Department for Transport (DfT), UK, commissioned a short project to use 2001 Census journey-to-work data in two streams of work: creation of Balancing Areas for use in the next version of the National Trip End Model (NTEM); and creation of a system of Functional Areas/Regions to form the basis for future research on household and business location choice modelling. This paper presents the methods which were used to obtain a system of Functional Areas/Regions and the results of the project. The project has been carried out by a consortium of MVA Ltd and David Simmonds Consultancy.
The main data used for this work has been the census journey to work matrix data. The other input to this project was boundary files for England and Wales and Scotland separately. The data for England and Wales gave the resident Output Area (OA), the work OA and then the number of people making those journeys as all people, full time student, part time student and, whether they work mainly from home or the mode of transport used for the journey. Analysis was carried out for the data for the ?all people? category. The data for Scotland is subtly different as the census question was the mode of travel to main job or course of study. The output tables are therefore split into the following groups as well as by mode. These are: all people; people aged 16-74 in employment; full time student; people aged 16-74 in employment; not full time student; and other persons. The assumption was made that the full time students in employment would give their student address, and efforts were concentrated on the people aged 16-74 in employment; not full time student. For this project it has been important to display the data using Mapinfo software to be able to assess contiguity of the areas and to make any manual changes to the definitions.
The best-established basis for a functional approach to area grouping is to identify boundaries across which few people commute. Although commuting distances have lengthened over time to become very long for a minority of people, distance still has a significant deterrence effect and, as a result, for almost all areas predominant flows are to and from nearby areas. There are numerous boundaries among areas such as parishes or wards which are the common ?building block? areas for which the data needed for regionalisation analyses can be obtained, and these boundaries will restrict the options as to which areas can be grouped at every step of an analysis which is explicitly contiguity constrained.
To create the potential functional areas for England and Wales and for Scotland we carried out some analysis using a hierarchical clustering algorithm called Intramax procedure which is incorporated in the Flowmap software (van der Zwan et al., 2003). The Functional Areas analysis provides, in effect, a catalogue of possible zoning systems based on identifying areas of relatively high self-containment in terms of the travel-to-work patterns in 2001 (or in 1991). What it does not provide is a justification for using any particular set of Functional Areas as the basis for any particular model, either at a national level or for any particular region or sub-region. Any such justification must depend on other characteristics, such as ? we suggest ? the perception of what constitute the alternatives for different types of locational decisions. These perceptions may well be very different for different types of actor ? for example, the set of areas perceived as possible locations by a small, regional firm may be different from the set perceived by a large national firm. Moreover, some decisions may well be hierarchical in nature, even before getting down to the choice of specific locations or properties within the lowest-level area. For example, a multinational firm seeking to set up an operation in England may make a choice at the regional level before making a choice ? on different criteria ? at the area level. Somewhat similarly, different categories of households may perceive areas differently ? for example, a high-income household with two specialised, professional workers might perceive the North-West as a small number of large areas, whilst a household consisting of unskilled workers might perceive the same region as a larger number of small areas (with perhaps only the closer ones being clearly perceived).
In addition to emphasising the difficulties of choosing one set of Functional Areas as the basis for the higher-level zone systems in any model, these arguments bring out two further points. First, it is not obvious that any one zone system is in fact appropriate for all purposes, even though that is the conventional approach to modelling and the present project has proceeded very much on that basis. Secondly, although the contrast between 1991 and 2001 is muddied by the differences in methods for sampling and protection of potentially sensitive data, it is likely that any set of functional units based on travel-to-work (or any other patterns of interaction) will change over time; it is therefore not strictly appropriate to use zones based on data for one point in time as the fixed units for long-term forecasting. Ideally, perhaps, the definitions of areas other than the smallest and most basic units would emerge as outputs of the forecasting approach rather than being fixed constraints upon it.
Association for European Transport