Is the TREAD Act Working?
It may still be too early to say for sure, but recent auto safety-related announcements have been filled with specific quantitative data about warranty claims and failure rates. Could it be that the TREAD Act's Early Warning Reporting system is really doing its job?
When Congress passed the Transportation Recall Enhancement, Accountability and Documentation Act back in 2000, it mandated the creation of an Early Warning Reporting system to detect patterns of automotive safety problems. The goal was to allow the National Highway Traffic Safety Administration to do its job faster and better by requiring automotive manufacturers to make quantitative reports.
Dare we say the system seems to be working?
Last year, there were two widely-held scenarios for TREAD Act reports. One was that the reports would be warehoused in some vast government repository where data goes to die. The other was that NHTSA would enforce compliance by singling out a few victims for a public flogging, thereby scaring all other manufacturers into announcing safety problems "voluntarily," before the government had a chance to flog them.
Critics said the government would receive more data than they could possibly analyze, so either it would be filed away without analysis or just a small subset would be analyzed as a deterrent to those who thought the government wasn't reading their reports. Comparisons were made to the tax code, where overall compliance is boosted by the feeling people have that they're going to get caught if they try to hide income, and the penalty for doing so will far outweigh the taxes they avoid.
Too Early to Tell?
It's still early days for TREAD reporting. While the Internal Revenue Service has had close to 90 years to spread its carrot-and-stick message of voluntary compliance and criminal deterrence, NHTSA has had barely 90 days to tweak its system. The first Early Warning Reports were due on Dec. 1 for the quarter ended Sept. 30, 2003. The historical reports covering the three years between July 1, 2000 and June 30, 2003 were due by Jan. 15, 2004. Reports for the fourth quarter of 2004 were due just a few weeks ago. So the system really is still new.
However, in the past month on at least five different occasions, NHTSA has made specific quantitative announcements of data regarding vehicle warranty claims, failure rates, complaints, crashes, injuries, and deaths. NHTSA spokesman Rae Tyson, who has turned up in more than half the recent articles about recalls and safety investigations, allows that this aspect -- the public release of specific quantitative data -- is something new for the agency.
What follows is a roundup of some of the articles, with multiple sources listed in cases where the newspaper archives may not remain online permanently. Many cover the same ground, but readers will notice that each article includes one if not several mentions of data extracted from Early Warning Reports.
- Ford Excursion SUV with Firestone Steeltex tires
Tires experience "rapid air loss."
NHTSA reported failure rate of 28 to 31 per million in 2002.
The Tennessean, Feb. 22, 2004
NHTSA reported six crashes with five fatalities and 20 injuries.
Newsday, Feb. 27, 2004
Washington Post, Feb. 27, 2004
NHTSA reported 1,000 complaints, 58 injuries and 13 deaths.
The Tennessean, Feb. 22, 2004
- Ford Windstar minivans
Door can slide open while the vehicle is moving.
NHTSA reported 56 complaints but no injuries.
USA Today, March 9, 2004
- GM Silverado and Sierra pickups
Failures in the cables that hold up the tailgate.
NHTSA reported 61,025 warranty claims.
Bloomberg News, Feb. 24, 2004
- Kia Sephia and Sportage
Seat belt latches are defective.
NHTSA reported 1,495 warranty claims, 35 crashes and 36 injuries.
Bloomberg, March 9, 2004
- Toyota Camry, Camry Solara and Lexus ES300 sedans
Vehicles surge forward unexpectedly.
NHTSA reported 30 crashes.
USA Today, March 9, 2004
Bloomberg, March 9, 2004
Reuters, March 9, 2004
The news here is that NHTSA is releasing specific nuggets of Early Warning Report data to support specific vehicle recalls. And by specific, we mean not an estimate of "over a thousand" warranty claims but exactly 1,495 faulty seat belts. Firestone wasn't recalling "hundreds of thousands" of tires. It was recalling 297,000 tires size LT265/75R16, load range D, manufactured in Joliette, Quebec, from March 1999 through December 2002. This data is so specific one might assume it came from an Excel spreadsheet, which in fact, it did.
There is no publicly released comparative analysis of the data, for instance a contrast in PowerPoint between failure rates in the third and fourth quarters, or a correlation of the new data to the baseline data computed from the one-time historical reports. However, by its very appearance in conjunction with a recall, NHTSA is indicating it has found a pattern or an anomaly in the data. In other words, 61,025 warranty claims for faulty tailgates must be unusual, even for GM, because NHTSA said it upgraded its probe into the matter. But 31 tire failures per million must not be a significant departure from the baseline, because NHTSA declined to reopen that investigation.
In the case of the Firestone Steeltex tires on the Ford SUVs, the stick portion of the carrot-and-stick approach seems to have factored into play. Even though the recall technically remains classified as voluntary by the manufacturer, the pattern revealed in the Early Warning data would probably have led to a NHTSA investigation. It's open to debate, but it seems as if Firestone detected the pattern before NHTSA could, and it acted upon the data before it could be publicly flogged.
An Associated Press article quoted Firestone spokesman Dan MacDonald as saying "We haven't identified a specific problem with the tires," but he added "the data suggested that there's an issue, and the decision was made that instead of having a long drawn-out investigation to figure out if there is a problem, let's just step up and replace the tires."
Read a different way, one could assume that without the knowledge that TREAD reports were going to the government, a manufacturer might be tempted to wait for conclusions more substantial than "the data suggested that there's an issue" before initiating a recall. What's all the more astonishing is that before the TREAD data began streaming in, NHTSA declined to reopen an investigation of the Steeltex tires. But when the "voluntary" recall was announced, it was restricted to one tire size from one factory.
"Certainly, the Early Warning piece of it is working," Tyson told Warranty Week. "I don't think there's any question that after the first Firestone investigation and recall it was pretty obvious that there was information that, had we been in possession of it, could have caused us to move much quicker," he said. "So Congress, realizing that, gave us these new tools. And I think so far, our expectations have certainly been realized."
Expert Setup Assistance
The initial setup of the computer system was performed for NHTSA by the John A. Volpe National Transportation Systems Center in Cambridge, Massachusetts. The Volpe Center, an R&D arm of the Dept. of Transportation, actually charges other government agencies for the work it does on their behalf. Of its $222 million annual budget, not one dime is appropriated directly by Congress. Instead, agencies such as NHTSA pay a fee for the tasks they assign to the center.
The actual processing and analysis of the TREAD Act data, however, is performed by NHTSA's own Office of Defects Investigation. And by all outwards appearances, they seem to be doing it well. "Although it's an incredible volume of information that comes in every quarter, there's no question that it gives us some additional tools to keep an eye on problems," Tyson said.
Back in Feb. 2002, NHTSA Administrator Jeffrey W. Runge told a committee in the U.S. House of Representatives about the agency's decision to subcontract the TREAD Act system's setup to the Volpe Center. "To improve ODI�s outdated information storage and management system and to handle the large volume of information that will be submitted under the early warning rule, we have contracted with the Volpe National Transportation Systems Center to design and implement a new state-of-the-art data warehouse. We have worked intensively with Volpe and its subcontractors to ensure that this system will address our needs, and we expect to have it on-line, on schedule and under budget, by the end of this year. When the new system becomes operational, we believe it will enable us to manage and effectively utilize the early warning reporting data."
Nick Shelness, chief technology officer for Entigo Corp. (and a sponsor of this newsletter), gave the TREAD reporting system a tentative thumbs up. "It's not quite clear to me how it was supposed to work," he said, "but I guess the government's premise was if it could get its hands on warranty incidence data and fatality data, then it could perform analysis that allowed it to discover problems before the manufacturers realized they had these problems -- or if they had realized, before they acted on them."
That seems to have happened. James King, vice president for TREAD Act solutions at Cypress Corp., also was somewhat tentative in his assessment that TREAD seems to be fulfilling its mandate. "It's starting to work," he said. "I don't know that they've got it all right yet." But they do seem to be putting their data warehouse to use, he said.
Shelness said the success or failure of TREAD was probably determined long ago when the Volpe Center was doing its initial setup. "A data warehouse per se is not a magic bullet," he said. "The trick is to ask the right questions of the data, and to structure the data in such a way that you can answer those questions."
Shelness said it's possible that NHTSA is now asking the right questions, and that it structured the data in just the right way to get those questions answered. But he's just not sure.
Still Gaps in the Data?
King cited a specific example that makes him wonder if perhaps the system isn't as good as its early reputation. He said he made a Freedom of Information Act request in December for a list of all manufacturers who requested Early Warning Report filing IDs. The list he got back had only 350 names on it. So either NHTSA hasn't responded completely, or else the number of companies filing TREAD reports so far is much smaller than was once anticipated.
"By NHTSA's own documents, there should be 1200 or 1300 companies," King said. Others put the potential count into the multiple thousands of companies. "To be just over 350 companies seems quite low." He said he's now repeating the request to see if perhaps the list has lengthened over time.
Still, the specificity of these recent recall announcements is somewhat surprising to him. Living near Detroit for many years, King said he became accustomed to hearing recall announcements in much the same way that Midwesterners get used to commodity reports or Southern Californians process the weekend box office totals. It's routine, and rarely is surprising. But these last few weeks have been different. In the past, "it was always rounded off to the nearest hundred thousand," he said. More recently, the data has become extremely specific.
King said he thinks this trend will actually be good for the TREAD reporting software market, which given his title as vice president for TREAD Act solutions, is not unexpected. If the government really is on top of its game, then manufacturers also have to be on top of their game.
"What we're seeing is the more NHTSA uses this tool, and comes back after the manufacturers, they're going to need software to automate the processes of responding," King said. "And that's where we come in. As NHTSA requests people to substantiate what they're doing, it is going to drive people to look out at their enterprise and ask, 'What am I doing with my documents? How do I get easier access to those documents?' Our software can eliminate all those file cabinets, and then with a simple query I could find every field report or every claim against the company given 15 different search criteria."
Shelness sees a similar positive outcome. If NHTSA and the Volpe Center can put together a system that can detect patterns and anomalies from TREAD reports, then so can the manufacturers themselves. If NHTSA can detect a spike in warranty claims for a specific component, so can the OEMs themselves. Therefore, at least theoretically, if they can prepare reports for NHTSA to analyze, they also could perform the analysis themselves.
"TREAD reporting is one aspect of it," Shelness said. "But actually doing failure analysis and product improvement analysis is maybe a more important part." In other words, it doesn't have to end with a determination that the defective tires were made in Quebec during certain years. Why did they fail, and why in only one size and on only one type of vehicle? Maybe it's not entirely the tire's fault?
Partnerships with Syncata
So while one would presume that all the TREAD reporting systems that will ever be needed were purchased and installed last year, there are reasons to believe that meeting the December and January deadlines was only the beginning. In recent months, Syncata Corp. signed a pair of partnership agreements with other TREAD Act reporting software vendors. In January, it signed with Cypress, and in February, it signed with Entigo.
Syncata and Cypress will together sell the TREAD Act Accountability System, a combination of their respective CoreTREAD Reporting Engine and Document Content Management software. Entigo will re-sell Syncata's CoreTREAD Reporting Engine in conjunction with Entigo Warranty.
Shelness said the Entigo-Syncata partnership is centered around the idea that compliance is just the beginning; that analysis is the next phase. "Entigo today has an operational system for gathering service data and claims against those services," he said. "But we don't have any reporting today. And certainly we have no analytical reporting. We team up with SAS, for example, to do analytical reporting for some of our customers."
"Our feeling is there is a section of the market that needs to do TREAD reporting, for whom the SAS or SPSS-type analysis would be too heavy-duty," Shelness added. "And it seems to us that Syncata has a suitably-priced product to address that market."
King had similar thoughts about the merits of the Cypress-Syncata partnership. "Our niche is the management of documents and content," he said. "We handle document content. The thing that we didn't have was the front end. And we're not a consulting firm." Meanwhile, Syncata has a great front end system but lacked a robust document management capability.
"They developed a great front end to fill out the Excel spreadsheet, and to apply rules set by the organization and by NHTSA," King said. "They do the front end and we do the document management. It seemed like a great marriage. We've pulled together a product that really gives you an end to end solution."