This paper will describe research commissioned by DfT in 2004 to "gain a quantified understanding of business location, particularly the role of transport in this, and implement that understanding in a model that can be used to forecast the number of jobs in each area of the country". The core of the work was the development and application of a simulation model in two case study areas, Milton Keynes and the South & West Yorkshire Multi-Modal Study area, or SWYMMS. However preliminary work included reviews of recent relevant research, including business surveys. Typically in surveys businesses tend not to report transport as being among the top reasons for them choosing their current location, listing instead the availability of premises, the ability to recruit, and access to customers and suppliers, which may be other businesses. In fact transport contributes significantly to two of these - recruitment and access to customers and suppliers - because it provides access. This suggests that in order to predict how transport might affect business location decisions it is necessary to show how itfeeds through into these types of access, and how much weight they are given in the decisions made by businesses. Although less directly related totransport, businesses also report the importance of the availability of suitable premises. This might seem obvious, but of course once built, thesepremises provide a significant degree of inertia in the system, constraining and directing, by their physical nature, what types of business activity might be possible. Usually they are constructed and/or refurbished by developers in response to market conditions, and will take time to be built. There is thus a mix of delays - as developers and businesses assess conditions and take action - and inertia, as past decisions limit or shape what might be possible in subsequent years by virtue of the legacy of the physical infrastructure. The major part of the paper will be a description ofthe model and the results it generates. It is a simulation model, showinghow events are likely change incrementally over periods of years. It tracks an index of how attractive each zone in the modelled area is for business activity, in terms of the provision of premises, the ability to recruit, and access to other businesses. Businesses may move in or out of each zone at rates that may be constrained or stimulated in response to the availability of suitable premises, the ability to fill job vacancies, and access to other businesses. The brief did not request that household location was modelled explicitly (this is the subject of a separate commission) but household numbers, and thus the workforce, affect the ability of employersto recruit, and so when simulating long periods it is necessary to represent them. In this model household numbers in each zone over time were supplied exogenously. Transport is represented via conventional link-based network models, with hierarchical logit structures governing choice of mode and route for trips within and between zones. However the transport models are implemented in a dynamic framework that explicitly recognises how people take time to adapt and change behaviour over time. In particular, it generates a travel-to-work matrix that continually shifts over simulated time as the numbers of businesses and households in each zone change. Two case studies were set up to simulate how each area had evolved over periods of ten years. The model was set up first for Milton Keynes, and then applied to the SWYMMS area. In each case we initialised the model using data describing the numbers of households, businesses and jobs in each zone in 1991. The model generates its own travel to work matrix, and this was calibrated against Census data for that year. Technically, the model is innovative, providing an interesting mix of conventional transport modelling withinthe dynamic framework provided by System Dynamics modelling software. In particular, it is not an equilibrium model - although it can reach equilibrium states - but will demonstrate how events change, dynamically, over time. For the covering abstract see ITRD E135582.
Abstract