DETROIT -- For years, the U.S. government’s auto safety watchdog sent form letters to worried owners of the Chevrolet Cobalt and other General Motors small cars, saying it didn’t have enough information about problems with unexpected stalling to establish a trend or open an investigation.
The data tell a different story.
An Associated Press review of complaints to the National Highway Traffic Safety Admin-
istration shows that over a nine-year period, 164 drivers reported that their 2005-2007 Chevrolet Cobalts stalled without warning. That was far more than any of the car’s competitors from the same model years, except for the Toyota Corolla, which was recalled after a government investigation in 2010.
Stalling was one sign of the ignition switch failure that led GM last month to recall 1.6 million Cobalts and other compact cars, including the Saturn Ion, Pontiac G5 and Chevrolet HHR. GM has linked the problem to at least 12 deaths and dozens of crashes. The company says the switch can slip out of the "run" position, which causes the engine to stall. This knocks out the power steering and power-assisted brakes, making the car harder to maneuver. Power to the device that activates the air bags is also cut off.
GM has recently acknowledged it knew the switch was defective at least a decade ago, and the government started receiving complaints about the 2005 Cobalt just months after it went on sale.
Although the overall number of complaints represents only 0.02 percent of the nearly 625,000 Cobalts sold from 2005-2007 in the U.S., experts familiar with NHTSA say they were enough to warrant an investigation and recall. The Cobalt had about the same rate of complaints as the Corolla. And the agency knew of at least two fatalities in Cobalt crashes that involved a sudden stall when it first declined to investigate the cars in 2007.
Spotting trends in the tens of thousands of complaints NHTSA gets each year is a tough job, and this case may have been more complicated than most. The Cobalt had a litany of problems, including fuel leaks, and a power steering defect that the agency did investigate. GM may not have disclosed all the information it had on the switches. And the 2010 recall of millions of Toyotas for unintended acceleration claimed much of the government’s attention.
But several experts say NHTSA should have pressed for a recall sooner.
"They’re not connecting up the dots. That’s the generous explanation," says Clarence Ditlow, executive director of the Washington-based Center for Auto Safety, who has studied the government’s auto safety agency for decades. "The not-so-generous is that they did connect the dots but they just didn’t do anything."
U.S. Transportation Secretary Anthony Foxx, whose department oversees auto safety, has asked for an internal investigation into the GM issue. In a letter calling for the probe, Foxx said he is unaware of information that NHTSA "failed to properly carry out its safety mission based on the data available to it and the processes followed."
The safety agency, in a statement provided to the AP, said that during the past seven years, its investigations have brought 929 recalls of more than 55 million vehicles. "Each potential recall investigation is unique and dependent on the data gathered in each case," it said.
Foxx has said that GM didn’t give the government enough information on the defective switches. In papers submitted to the safety agency last month, GM says engineers proposed solutions to the problem in early 2005, but the company didn’t take action, developments unknown to the safety agency at the time. But the AP analysis makes clear that even without that information, NHTSA had evidence in 2005 that the switches were a problem.
That summer, the agency hired a contractor to look into a July 29, 2005, crash in Maryland that killed 16-year-old Amber Marie Rose. The report concluded that the air bags of Rose’s 2005 Cobalt did not inflate, and the ignition switch had moved from the run position to "accessory," which runs devices like the radio but not the engine. Alcohol and speed also were factors, the report said.
Rose’s birth mother, Laura Christian, said that after the crash, she studied the government’s complaint database and found multiple problems with engine stalling and power steering failures on other Cobalts. She tried to tell NHTSA, but the agency wasn’t interested, she said.
"Basically, it was ‘No, thank you,"’ Christian said. "NHTSA should have known, based on the information I have seen, certainly in 2006."
On Thursday, David Friedman, the agency’s acting administrator, told Christian that he was taking steps to improve NHTSA’s information-gathering processes, Christian says. They met at the agency’s offices.
Among the other evidence available to the agency:
n In December 2005, General Motors sent the safety agency and its dealers a service bulletin telling them that drivers could inadvertently turn off the ignition switch with minimal effort in Cobalts from the 2005 and 2006 model years. Dealers were told about repairs and to tell drivers reporting engine shutdowns to remove unnecessary items from their key chains.
n In October 2006, GM sent the agency and dealers another service bulletin, adding Cobalts from the 2007 model year.
n In 2007, the government commissioned a report on a 2006 Wisconsin crash that killed two teenage girls and injured another. In that report, Indiana University’s Trans-
portation Research Center found that the ignition in the 2005 Cobalt was in the "accessory" position and the air bags failed to inflate. Investigators told the agency that "inadvertent contact with the ignition switch or a key chain in the 2005 Chevrolet Cobalt can in fact result in engine shut-down and loss of power."
n In 2007 and later in 2010, NHTSA’s Office of Defects Investigations examined data on stalling incidents and air bag failures in GM cars. Yet the agency recently told House members it was unable to spot trends that were significant when compared with "peer vehicles" or the U.S. passenger car fleet, according to a letter released earlier this month by the House Energy and Commerce Committee.
But Ditlow says comparisons with peers are less important than simply watching the numbers and taking action when they get too high.