Deriving Values of Statistical Lives From Observations of Speed Limits and Driving Behaviour

This paper discusses different ways in which empirical estimates of the value of a statistical life (VSL) can be derived from observations of highway driving speeds and how these speeds are affected by speed limits and penalties for speeding. Assuming that drivers optimise with respect to driving speeds, we derive three alternative such concepts. The first two of these are based on construction of driver utility functions, and the last one on revealed government preferences as in a paper by Ashenfelter and Greenstone (2004a). The two last (but not the first) are based on observations of changed driving speeds when speed limits and penalties for speeding change. When drivers instead are law-abiding and adhere to speed limits, only the last approach can be used. While Ashenfelter and Greenstone's may be the only viable approach given current data availability, we also indicate some potential problems with their results, in putting great demand on government information about VSL, and in possibly overvaluing VSL since mortality risk is considered as the only social cost of increased highway speeds. © 2005 LSE and the University of Bath