Using Safety Labels to Make Cars Safer
Providing buyers with an overall assessment of a vehicle’s crashworthiness would prod manufacturers to make better cars.
What is the best way to make cars safer? As in the case of reducing environmental risks, the traditional strategy has been government regulation. Design standards have been used to require certain features that are implemented in certain ways, such as seat belts and air bags. Performance standards have been used to specify how a car must perform under test conditions such as a frontal crash. But there is a quite different strategy that would complement and extend these traditional approaches; one endorsed by a committee of the Transportation Research Board of the National Research Council (NRC). The committee, which I chaired, proposed a relatively simple approach: Give customers clear summary information on the safety of all new vehicles, make the underlying details available to all who want them, set up a research program to ensure that the information will improve over time, and then step back and let the competitive pressures of the marketplace force manufacturers to produce safer cars. The committee’s recommendations are just as relevant today as they were when they were issued in 1996, perhaps even more so.
To be sure, a lot of safety-related information is available today from manufacturers, the insurance industry, organizations such as Consumers Union (CU), and the federal government. But much of this information is in the wrong form or addresses the wrong issues. For example, it is easy to obtain a list of a car’s specific safety features but almost impossible to obtain an informed estimate of how, if at all, these features actually contribute to making the car safer. Similarly, government frontal-crash test data (reported using a rating system from zero to five stars) basically indicate how a vehicle can be expected to perform in a frontal collision with another car in the same weight class. However, one could not tell from these data that the occupants of a very small five-star-rated car may fare much worse than the occupants of a less-well-rated large car when the two collide. Even the most careful and diligent consumer will find it difficult to put all the bits and pieces of available information together to come up with an informed overall assessment of vehicle crashworthiness.
In a report titled Shopping for Safety: Providing Consumer Automotive Safety Information, the NRC committee proposed a three-level information program. First, the committee urged the use of safety labels for all new cars sold in the United States. The label would carry an overall vehicle crashworthiness rating as well as a checklist of a vehicle’s crash-avoidance features. In addition, a safety brochure would be provided with each new vehicle, providing more details and comparisons with other vehicles. Finally, a safety handbook available in libraries and electronically via the Internet would provide greater detail and comparative information.
Combining data and expert judgment
At our initial meeting, many of the committee’s auto safety experts argued that it would be impossible to combine all the available data into a single summary measure. To test this assertion, we posed the following hypothetical situation: Suppose your 20-year-old daughter or son has just moved to another planet that is like Earth except that the auto companies are different. When asked for advice by your child on which car to buy, wouldn’t you want to see the available crash test and similar data? The experts agreed that, yes, they would. What would they do with the data? They replied that they would combine it with their years of experience in auto design and crash analysis and make a judgment about which cars are likely to be safest. This response told us that although there may be no simple objective formula that will allow a handful of test data to be combined into a reliable summary measure, such data, when combined with expert judgment, could be used to provide a useful, albeit imperfect, summary estimate.
The committee proposed that a process to do this be set up and run cooperatively by the National Highway Traffic Safety Administration (NHTSA), motor vehicle manufacturers selling in the United States, the insurance industry, and consumer groups. Information for the labeling program would be produced by an independent group of safety experts who would operate almost as a jury. These experts would begin with information about the relationship between crashworthiness and vehicle size and weight and then use analysis combined with professional judgment to incorporate results from crash tests, highway crash statistics, and a variety of other factors, including the presence or absence of specific design features. The committee didn’t believe that it would be possible to produce a similar summary measure of crash avoidance, so it suggested a simple checklist of crash-avoidance features.
Any summary measure of safety that could be provided today would be somewhat limited. The committee argued that such a measure would improve markedly in the future if the process of producing it were combined with a research program that resulted in better safety information for consumers as well as better data, tests, and analysis tools for vehicle designers. The point of this experimental program is not to make cars that do a good job passing a few specified tests. Rather, it is to make cars that have a high degree of crashworthiness in the complex environment of U.S. highways.
Could we afford such a program? The committee estimated the initial cost at about $20 million a year, not much money when one considers that about 40,000 people die in motor vehicle accidents each year in the United States. The value of even a small decline in net fatalities would be considerable and should easily exceed the cost of supplying better information to consumers and vehicle designers. If one bears in mind the current estimates of the public’s willingness to pay to reduce the risk of death in motor vehicle crashes, a $20-million-per-year research program would need to achieve a net mortality reduction of only about 10 deaths per year to justify program expenditures.
The initial reaction to our recommendations was generally negative. The auto industry opposed the strategy, which was not surprising because explicit summary estimates of vehicle crash performance would, at least in the short term, disadvantage some models and manufacturers. Surprisingly, a number of consumer groups also reacted negatively. Some of these groups were entirely preoccupied with their own short-term agendas, which included a requirement for a specific test. Because the committee’s report had not endorsed their agenda of the moment, they were unhappy.
But slowly over time, things began to change. NHTSA published a paper with similar conclusions in the Federal Register and sought public comments on the approach. After reviewing the recommendations more closely, representatives of a couple major auto manufacturers not only determined that the committee’s approach was feasible but sketched out how they thought it might best be done. NHTSA did not act immediately. During the past few years, the agency has been busy implementing additional performance standards. But now that it has implemented two types of tests (front and side impact) and has a third in the process of development (rollover), it is beginning to once again give serious consideration to the possibility of combining these data to produce a summary measure.
For such a system to be created, NHTSA probably must make the next move, even though it should be possible for a newly created nonprofit organization or an existing group such as CU, Underwriters Laboratories, or the insurance industry to take the lead. However, the necessary resources and leadership do not seem to be present.
Signs are promising for NHTSA action in the near future. When that happens, the system will be most effective if four things occur: 1) the effort is mounted as a cooperative venture in which relevant private-sector groups are included in a collaborative capacity; 2) a committee of independent experts (including academic engineers and statisticians as well as retired auto and insurance company engineers) is constituted to supervise the needed analysis and make the needed subjective judgments; 3) the process is linked to research programs in auto crash safety in a way that allows us to learn and improve over time; and 4) existing crash test procedures are not viewed as immutable but rather can be periodically considered for modification as our knowledge improves. This last item is likely to pose the greatest difficulty for NHTSA. However, to be truly effective, a system of crash test summary information must be adaptive. In the longer run, we should work at developing comprehensive design software that allows designers and regulators to subject proposed new designs to hundreds of different crash scenarios. Calibrating such codes will require a wider range of data from a variety of well-designed tests.
Traditional regulation moved the United States a long way toward achieving a clean, safe environment. More recently, market-based strategies based on providing good summary information have been making significant additional contributions to reducing some environmental risks. We can learn from this experience and apply it to auto crashworthiness. Traditional regulation has already made our cars far safer than they were 50 years ago. If we want to make them still safer in the coming decades while minimizing the heavy hand of government regulation, it is time to start providing customers with clear summary estimates of crashworthiness. The many consumers who take advantage of these will provide the market push to make all of our cars safer.