Is There a Limit to Car Ownership Growth? An Exploration of Household Saturation Levels Using Two Novel Approaches
WHELAN G and WARDMAN M, ITS, University of Leeds, UK and DALY A, ITS, University of Leeds, UK and Hague Consulting Group, The Netherlands
Growth in car ownership has been one of the most intensely researched transport topics over many years since the eadiest work in the United States (de Wolff, 1938; Roos and von Szelski, 1939; Chow, 1957; Suits, 1958)and in Great Britain (Rudd, 1951; Smeed
Growth in car ownership has been one of the most intensely researched transport topics over many years since the eadiest work in the United States (de Wolff, 1938; Roos and von Szelski, 1939; Chow, 1957; Suits, 1958)and in Great Britain (Rudd, 1951; Smeed, 1951; Reynolds, 1954; Tanner, 1958). This importance is due to the dominant position that car travel has and will increasingly hold along with its infrastructure, demand management, energy, environmental, health, social and land use implications.
Car ownership forecasts are therefore of use to a wide range of organisations and they take many different forms~ However, a common theme running through much car ownership modelling is that ownership follows an S-shaped growth curve (for example; de Wolff, 1938; Tanner, 1958; Bates, Gunn and Roberts, 1978; DETR 1997; Dargay and Gately, 1999). The economic rationale used to support this practice is provided by product life cycle and diffusion theories whereby the take-up rate for new products is initially slow, then increases as the product becomes more established, and finally the rate of increase diminishes as the market becomes close to saturation (Duesenberry, 1949).
Looking at a simple plot of car ownership over time (Figure la), it is difficult to identify an S-shaped pattern in ownership, although such a pattern might be more recognisable in a 'longer time series. However, when we look at ownership patterns across different income groups (Figure lb) it becomes evident that S-shaped ownership curves with clearly identifiable saturation levels exist. Bearing in mind that the average annual household income in 1996 is income group 5 (£201663), it is clear that household ownership levels have yet to reach saturation, a further reason why the S-shaped curve is not apparent in Figure la.
By specifying car ownership models with an S-shaped functional form and a saturation level (either, implicit or explicit), forecasts of vehicle ownership will be curtailed as saturation is approached. This feature is likely to be of little significance to forecasts for emerging markets but is highly significant in more mature markets such as Great Britain.
In this paper we employ two recently developed approaches to estimating saturation levels. In the first instance, we use a nested Iogit structure to estimate saturation levels using disaggregate data. These models have the same functional form as the car ownership models that underpin the UK National Road Traffic Forecasts but they offer the potential to allow saturation levels for different household categories (e.g. single person households) and area types (e.g. London) to be determined directly within the model rather than be inputs into the calibration process. Secondly, we revisit the aggregate models of car ownership which in Great Britain were pioneered in the work of Tanner (1958, 1977). Here 'global' saturation levels are estimated directly using the non-linear least squares procedure with a further innovative feature being that we distinguish between household saturation levels for owning at least one car and owning two or more cars in contrast to previous work which has dealt with car ownership per head.
Association for European Transport