A model is proposed for national car ownership forecasting in which the level of ownership depends on income and motoring cost levels in previous years as well as in the current year. Such a model provides a method of explaining the "time" trend in previous models and of resolving inconsistencies between observed short-term and long-term elasticities. When fitted to recent data, the model indicates a strong element of persistence, with the long-term income elasticity being about 4 times the short-term one. When used to forecast ownership over the next 30 years for alternative assumptions about economic growth, it gives much the same result as an unlagged model with a time trend if an economic growth rate of about 1.5 per cent is assumed; over a range of assumptions it gives a much wider range of forecasts than the unlagged model.
Abstract