GM recalls: The numbers tell a surprising story – MarketWatch
A MarketWatch analysis shows General Motors had an unusual run of statistical good luck before the recall crisis of 2014
General Motors has issued a torrent of recalls in 2014, for problems ranging from unsecured floor mats to the ignition-switch defect linked to at least 13 deaths. These 54 separate recall campaigns affected as many as 26 million vehicles in the U.S.
The recalls have turned into a public-relations nightmare for the company, raising questions about GM’s management culture and its safety protocols. But a deep dive into the numbers suggests that they may represent something simpler: The law of averages catching up to America’s biggest auto manufacturer.
MarketWatch analyzed recall data for vehicles and equipment collected by the National Highway Traffic Safety Administration through August 3, 2014. The 10 companies featured in our analysis are the top manufacturers by 2013 sales in the U.S., according to Motor Intelligence.
How do this year’s recalls compare to recent history? Here’s a look into recall data dating back to 2004, from the top 10 automakers based on U.S. vehicle sales.
Our analysis found that nearly 94% of all vehicles potentially affected by recalls in 2014 are manufactured by one of the top 10 manufacturers. These companies are also projected to have manufactured 92% of all vehicles in operation in 2014.
GM leads the industry in recalled vehicles for the year so far, by a large margin. But what’s surprising, from a statistical standpoint, is that its share of recalls has been so low over the previous years. GM accounts for 26% of all U.S. vehicles in operation (or VIO), according to data from Experian Automotive. That’s the most by far of any automaker — second and third place go to Ford, with 19%, and Toyota, with 13% — but it’s actually their lowest share of VIO since Experian started tracking the figure in 2008.
It’s commonplace for top manufacturers to have years where their share of recalls far exceeds their share of VIO: In fact, each of the top 10 auto manufacturers has had at least one such year since 2008, according to our analysis.
To put things simply: The more cars a company makes, the more of that company’s cars are likely to be recalled.
The more interesting question, then, is whether a company’s recall numbers are far out of proportion to the number of cars it produces. Among the top 10 automakers, the average ratio of share of vehicles in operation to share of cars recalled since 2008 is 1.2 (shown in the chart below by the black line), close to a one-to-one correlation. Each point on the chart indicates this ratio for one company in one year. Points outside the shaded areas are more than two standard deviations from the average, commonly a measure of a statistical outlier.
Though GM’s recalls in 2014 appear to stand out, the point falls inside a statistically normal range. Of course, just because the high number of recalls across the industry is “normal” doesn’t mean it’s acceptable from a consumer-safety standpoint. (GM is setting up a compensation fund for people whose deaths have been linked to defects in its autos; the company also says it has hired additional safety investigators, and CEO Mary Barra recently told a Senate hearing that she had “elevated safety decision-making to the highest levels of the company.”)
The number of potentially affected vehicles, or recalled vehicles, is not a simple count of the number of vehicles with recalled parts. When combining potentially affected vehicles over several recall programs, vehicles are counted each time they are recalled, so the same car may appear in the data more than once. That means while there have been 44 million potentially affected vehicles so far this year, the actual number of cars recalled in the U.S. is lower. You can read about the data and our exceptions at the end of this article.
Even without counting General Motors, the number of motor vehicles recalled in the United States is on pace to exceed a 10-year high of 20.4 million.
When you do add in GM, which has had a total of 26.1 million recalled vehicles so far in 2014 alone, the numbers across the industry shoot up to 44 million — a 15-year high, already, with nearly five months left to go in the year.
As we said, because GM has the largest number of active vehicles in the U.S., its high number of recalled vehicles may not be news on its own. However, coupled with the faulty ignition switches, it’s the change in the number of cars recalled from previous years — up more than 3,300% from 2013 and 140% from 2004, which had the highest value for GM in the past decade — that’s striking.
From 2004 to 2014, 24% of all vehicles potentially affected by recalls were manufactured by GM, which makes sense given the company’s 26% share of America’s vehicles in operation as of 2014. However, GM’s percentage of recalled vehicles this year is much higher, at 59%. From 2004 to 2013, GM represented a relatively low 16% of all recalled vehicles.
Industry-wide, automotive recalls often involve vehicles that are several years old when the recall is announced. The chart shows the distribution of GM’s 2014 recalls across model years. Had defects in its models from the mid-2000s come to light sooner, GM might not have had that long streak of years where its share of recalls was well below its share of VIO. But there’s nothing inherently unusual about such a lag time: The cluster on the chart of recalled model years around five years old is a common pattern for other top companies, according to our analysis.
For a related story, read: Why Recalled Cars Stay on the Road (WSJ)
Another interesting pattern that emerges from our analysis: Relatively small manufacturers account for a large share of the number of recall campaigns, and big manufacturers generate relatively few campaigns, though they involve a lot of vehicles.
Despite manufacturing more than half of the potentially affected vehicles, GM is only responsible for 14% of all recall campaigns this year, according to our research. That said, the company’s 54 recall campaigns are more than double last year’s number and nearly four times the manufacturer’s share of all recall campaigns in 2013.
If a recall campaign represents a single defect in a product, the top companies represent an average of 19% of defects each year. That has edged up to 32% in 2014 alone — again, largely due to GM’s woes — but it’s still well below their share of the market.
As mentioned earlier, potentially affected vehicles can’t be directly measured against a company’s vehicles in operation (VIO), because cars may be counted multiple times in the recalled total. But while there may be a finite number of fixes that have to be made, percentages can give us a better sense of scale. By that measure, the number of vehicles potentially affected by recalls this year is equivalent to 17.8% of all vehicles in operation in 2014. Last year, that number was 8.5%.
Looking at it that way, GM’s 26.1 million potentially affected vehicles in 2014 are equivalent to 40% of GM’s fleet and almost 11% of all vehicles in operation, using Experian Automotive data.
You can look up your vehicle online to see if it’s been recalled. In particular, 2014 Chevrolet Silverado and 2014 GMC Sierra car owners should take their cars to be serviced because those models alone are covered by six different recalls by General Motors this year.
About the data
The National Highway Traffic Safety Administration (NHTSA) oversees motor vehicle recalls in the United States. Manufacturers submit a Part 573 Defect and Noncompliance Report as a result of internal investigations (76.5% of recalls since 2004) or investigations by NHTSA’s Office of Defects Investigations (21%) or NHTSA’s Office of Vehicle Safety Compliance (2.5%). Information about recalls is filed by manufacturers of either the recalled item or the manufacturers of the motor vehicles with the recalled part installed.
Three fields in the data identify the manufacturers related to the recall: the make of the vehicle, the manufacturer of the recalled product and the company that filed the report. Because the forms are completed by various manufacturers and divisions of companies, neither of these fields alone are reliable identifiers for the manufacturer taking responsibility for the recall. After comparing reports to manufacturer data, we determined only records reporting the same company as the manufacturer of the recalled parts and filer of the report could be assigned to said company.
Data not fitting this pattern were excluded from each company’s numbers, but counted as part of total recalls. As a result, some of the numbers in this analysis may vary from other reports. You may have seen news reports showing the number of recalled vehicles for GM at 29 million. That number includes vehicles across North America and exports. In this report, we are highlighting U.S. data only.
Data for Chrysler LLC and all makes under the brand are not included in the data for the top companies before splitting from Daimlerchrysler in 2007.
Although Kia and Hyundai share many facilities in the United States and are often considered one company in reports, the recall and sales data lists them separately so our analysis does too.
The following makes are included in the recall analysis for each company:
- General Motors LLC: Buick, Cadillac, Chevrolet, GEO, GMC, Hummer, Oldsmobile, Opel, Pontiac, Saturn
- Ford Motor Company: Ford, Lincoln, Mercury, Merkur
- Toyota Motor Corp.: Toyota, Lexus, Scion, Daihatsu
- Chrysler LLC: Chrysler, Dodge, Ram, Jeep, Fiat, Plymouth, Eagle, Alfa Romeo, AMC
- American Honda Motor Co. Inc.: Honda, Acura
- Nissan North America Inc.: Nissan, Infiniti, Datsun
- Hyundai Motor America: Hyundai
- Kia Motors America: Kia
- Volkswagen of America Inc.: Volkswagen, Audi, Bentley, Lamborghini
- Subaru of America Inc.: Subaru
Whenever possible, missing data has been filled in by examining available documents. Some of the recent data not yet available in NHTSA’s database has been collected from the manufacturers’ press releases. Recalls for the top companies determined to be the responsibility of one company but including vehicles from another (Pontiac Vibes included in Toyota recalls and Isuzus in General Motors) are split into separate entries using the breakdown of potentially affected vehicles submitted in the Part 573. These vehicle separated out become part of the total recall data.
Recalls for equipment, identified by an invalid model year in the data, and campaigns missing or still waiting on the number of potentially affected vehicles are not included in the final analysis. All vehicle types are included.
Experian Automotive calculates vehicles in operation using title registrations provided by each state. The data for the article counts light vehicles at the end of the fourth quarter of each year. The numbers for 2014 are from Q1. Data was not available before 2008.
Since the statistical model used to compare the number of recalled vehicles to a company’s vehicles in operation is limited to a small set of data, we tried testing the results against earlier data from 2004 to 2007 using two methods to approximate vehicles in operation: holding the 2008 percentages constant and using the later year-over-year changes in combination with 2004 to 2008 new car sales. Neither approximation significantly changed the output of the model. In both cases, the range for normal values grew larger, making GM’s 2014 recalls move more securely into the normal range.