1 Historical determinism and backcasting in future studies Paper presented at the conference on Urban Transport Systems in Lund, June 7-8 1999, preliminary version Mattias Höjer and Lars-Göran Mattsson 1. Introduction There are a number of different approaches for analysing what will, could, or hopefully should happen in the future (e.g. scenario technique, forecasting, backcasting, Delphi- studies, modelling). One approach does not necessarily have to be opposed to another, but there are differences between them. Evidently, they have different roles and cannot do the same job. In principle two different aims of a future study can be identified. The first is a desire to know what the future will be like, so that adjustments can be made when there still is time. This can be the motive for a planning authority just as well as for a business firm. The second reason for studying the future is a belief that planning can change development. With this thought, the interest in future development may be driven by a willingness to change the development path, rather than to adjust to a path that is more or less given. This can be the driving force for planning authorities and for big corporations, but hardly for smaller firms. The aim of this paper is to point at some risks with certain applications of future studies and also to give a brief presentation of the backcasting approach that in some situations may be more fruitful. We will also try to clarify the relationship between some of the different approaches, especially between forecasting, backcasting and mathematical modelling. We begin by looking at three common approaches in future studies, all of them taking historical data as a starting point. Some risks and problems with these approaches are discussed. In the succeeding section, we argue that backcasting can be a relevant alternative or complementary approach when development seems to take an unwanted path, and when no solutions to the identified problems can be found with the help of other approaches. Backcasting, and its relation to other approaches in future studies, are presented in some detail, in the section preceding the concluding discussion.