Abstract of Meeting Paper

The 1996 Annual Meeting of the Society for Risk Analysis-Europe

Incorporating Accident Risk and Disruption in Economic Models of Public Transport. Andrew W. Evans and Alan D. Morrison, University of London Centre for Transport Studies

This paper demonstrates how conventional economic models of public transport may be extended to incorporate the effects of accident risk to passengers and non-passengers, and nonscheduled delay to passengers. An extended model has been developed and applied to a hypothetical high-density surface railway system. which has the general economic characteristics of the London Underground system, and the safety characteristics of British Railways.

The results of the extended model should be regarded only as indicative, because the model is not detailed, and the system to which the model is applied is hypothetical, but they do indicate the nature of conclusions that are possible from this type of model. The main conclusions are the following.

1. The risks to passengers of rail travel are low: on the assumptions of the model, and with the commonly-used value of £2 million for a statistical railway fatality, risk represents 1.3% of the user cost of a railway journey.

2. However, the risks to non-passengers (that is, staff, third-parties, and trespassers) from railways are relatively more important, and have a significant effect upon the economics of risk reduction. There are about twice as many accidental non-passenger as passenger equivalent fatalities in Britain, even after discounting half the trespasser fatalities on the assumption that they are non-accidental suicides, though not recorded as such.

3. On the basis of a limited amount of London Underground data, measures to reduce non-scheduled delay are cost-beneficial; indeed, they are substantially self-financing, as a result of the additional patronage they would generate.

4. The reduction of railway risk below its present level is cost-beneficial when the value of a fatality is £2 million; however, it is marginal when the value is £O.75 million, assuming that the operator has a fixed level of subsidy. The £0.75 million is approximately the value currently used for road fatalities in Britain.

5. If the operator has a fixed level of subsidy, risk reduction leads to some redistribution of welfare from passengers to non-passengers. This is because passengers bear the costs of measures to reduce risk, through higher fares or reductions in other aspects of the quality of service, but non-passengers are the major beneficiaries.

6. Sudden increases in the perceived level of risk by passengers are possible following high-profile accidents. These cause a substantial loss in welfare, mainly because patronage falls and reduces the operator's revenue. If such changes in risk perception can be caused by major accidents but not smaller ones, they provide some justification for the widely held view that the prevention of major accidents should command proportionately more resources than the prevention of smaller accidents.

7. If the level of risk is permanently over-estimated by passengers, extra safety measures are justified. Without these extra measures, patronage would be inefficiently low.