Two measures of car model malfunction probability, fraction of cars over 1% CO (y-axis) and average CO concentration of all cars (x-axis), demonstrate that five 1987-89 car models (14 year-model combinations) have a malfunction probability several times that of all other models.
When an automobile's emissions control system fails, it may be because that model is more prone to failure than others, according to a study conducted by the Center's Energy Analysis Program and Marc Ross of the University of Michigan. This finding goes against the conventional wisdom that improper maintenance or deliberate disabling of the emissions systems by car owners is the cause of "high-emitting" vehicles. The results may provide clean-air regulators with more cost-effective strategies to reduce air pollution from autos than the current system of individual automobile inspections. We presented these results at the Asilomar Transportation and Energy Conference last summer and at conferences held by the Society of Automotive Engineers and the Coordinating Research Council earlier this year.
Regulations established by the Environmental Protection Agency (EPA) have resulted in dramatic reductions in tailpipe emissions from new cars: on the order of 95% for carbon monoxide (CO) and hydrocarbons (HC) and 75% for oxides of nitrogen (NOx). However, increased vehicle use, as well as two loopholes in current vehicle emission regulations, have counteracted these emission reductions; if we don't close these loopholes and/or curb vehicle use, we may see increases in auto emissions and urban air quality problems in the future.
One of the loopholes is that the test procedures to certify compliance with emission standards do not reflect the high speeds and accelerations of current driving. The test procedures encourage manufacturers to design cars that merely pass the test, rather than minimize emissions over all types of driving. The second loophole is that there is very limited testing of emissions as cars age. Although manufacturers are now responsible for designing cars that meet the tailpipe standards up to 100,000 miles, in practice only cars that have been "properly maintained and used" are selected for compliance testing.
We analyzed a large dataset of car tailpipe emissions measured with a remote sensing system. The system uses an infrared beam to measure the CO, CO2, and HC emissions directly from the tailpipe as the car drives by, while a video camera records the car's license plate. Researchers from the University of Denver set up the system at sites in Southern and Northern California to gather data for the California Air Resources Board in 1991. The state Department of Motor Vehicles matched each car's license plate with its vehicle identification number, which provides detailed technical data on each vehicle.
We focused on cars that were two to five years old at the time of measurement. We expected that these young cars, which were within the manufacturers' warranty period for emission control components, would have a low probability of malfunction. About 10% of these cars showed CO emissions that were 50 times that of properly functioning cars--a very high rate for young cars.
Next, we grouped the vehicles by manufacturer and model to analyze the probability of malfunction for the most popular models. The malfunction rate varied widely by vehicle model; some models had only a few malfunctions or none at all, while others had average failure rates of up to 30%. The data did not show a correlation between manufacturer and probability of malfunction. In fact, they suggest that the manufacturers whose models had the highest failure rate also had very clean models. The five worst performers were low-priced models from Asian manufacturers: an average of 22% of the models in this category had malfunctioning emissions controls, compared to 6% for all the other models (see figure). At the same time, the mid- and high-priced models from the same manufacturers had very low failure rates. Most domestic models fell in the middle of the range of malfunction percentages. We also found that malfunction rates were higher in vehicles with certain technologies, such as carburetors (as opposed to fuel-injection systems).
If confirmed, these findings have important policy implications at both the state and federal level. Current policies, such as California's Smog Check program, are designed to identify individual vehicles for repair through regularly scheduled vehicle inspections. A controversy is currently raging between states and the EPA over what is the most cost-effective system to identify individual high-emitting vehicles: a network of remote sensors or more detailed laboratory tests. The results from this study may shift the focus of the debate.
Our analysis indicates that emission control problems in new cars are not necessarily the fault of drivers not properly maintaining their cars; certain models appear to have more problems than others. Policies that focus on improving the durability of emission controls on all models may therefore be more cost-effective in reducing emissions than the current or proposed approaches.
--Tom Wenzel
Tom Wenzel
Energy Analysis Program
(510) 486-5753; (510) 486-6996 fax