Integrated Computation of Time, Morbidity and Mortality Values from Revealed Preference Data
PENIN R and RIERA P, Universitat Autonoma de Barcelona, Spain
Morbidity and mortality values have been calculated in many ways, with a very large dispersion of results. There are philosophical and methodological reasons for the variety of ways used in estimating such values. Different approaches must yield different
Morbidity and mortality values have been calculated in many ways, with a very large dispersion of results. There are philosophical and methodological reasons for the variety of ways used in estimating such values. Different approaches must yield different values. However, some variations in the estimated values arise from methodological misspecifications. This paper proposes, and applies to a case study, an approach that overcomes some of those misspecifications and seems to make more sense if values are to be used in cost-benefit analysis of road investments. The application is to drivers facing alternative roads with different costs, time and accident rates.
The computational approach applied does not differ from the classical stated preference probabilistic models used for time value estimation, but introduces an integrated approach, which to our knowledge constitutes a novelty. First of all, the model is calibrated with time and other covariates. Second, time is substituted for fatal accident risk. Finally, a model including both variables and covariates is estimated. The results show that the integrated approach yields far more conservative (and one may say reasonable) values. Time values lower from 59 pesetashinute to 35 pesetas/minute and vehicle. The value of life presents an even more dramatic change, since the difference is of more than three times. This would imply that most time or accident valuations contain errors that tend to overstate those values. In the integrated model a test of value of life in time units is conducted. The test shows that the value of life expressed in time (according to the internal value of time estimated) corresponds to a behaviour equivalent to a subjective mean expectancy of life of 90 years, which appears to be quite plausible, and is statically not different from the actual live expectancy. The paper stresses the econometric advantages and limitations of the models used in this and other authorsÕ estimations.
Association for European Transport