By on November 1, 2011

Someone asked yesterday: “And what exactly is the difference between journalism and blog anyway?” Let me tell you a story:

Last Wednesday, I walked down the shop floor of Nissan’s humungous factory in Smyrna, Tennessee. A monitor pronounced that Volkswagen would end the year as the world’s largest automaker. My stomach knotted.

Two days before, Bloomberg had made the same proclamation: “VW Likely to Overtake Toyota as Top Carmaker in 2011, GM to Remain Second.”

I immediately warned the world that this is nonsense. But it didn’t stop the story from going viral. After all, who are you going to believe, a blogger at TTAC, or the professional journalists at Bloomberg?

The following day, I received a surprising email from someone very high up the food chain at Bloomberg … He confirmed that the story was wrong and that it should have never gone out “with the data the way it was presented.”  The author of the email requested and receives anonymity. Professional courtesy.

The big mistake

Bloomberg had written correctly that analysts of J.D. Power, IHS Automotive and PwC Autofacts see Volkswagen’s sales rise 13 percent, they estimate that sales of GM would gain about 8 percent, while Toyota’s sales should drop 9 percent. 10 months into the year, that’s an easy call.

Then, however, a serious mistake was made. The estimates were “used to calculate the average projections.” By Bloomberg. Bloomberg calculated Volkswagen’s year-end sales as 8.1 million, those of GM as 7.55 million, and those of Toyota at 7.27 million.

You need to know only one number to immediately notice the error: GM’s global sales in 2010 were 8.39 million. If you don’t know that number off the top of your head, you can google it. Add 8 percent growth, and you end up with more than 9 million. Even if you can’t figure a tip without a pocket calculator, it should be evident that 8.39 million plus 8 percent can’t possibly be 7.55 million.

Due diligence

But miracles happen. Allegedly. So as a precaution, I picked up the phone and called all three automakers. After making the requisite noises that they cannot comment on results before the results are in, all three told me that the Bloomberg story is bunk. A knowledgeable contact at GM assured me that I wouldn’t look stupid if I would guess that GM would end the year a little less than a million units ahead of the number 2 contender, Volkswagen.

Bloomberg could have done the same. They didn’t.

They blew the chance to google the right number. They blew the chance to pick up the phone. Instead, they picked the wrong number, or as my emailing contact at Bloomberg put it, “the estimators didn’t start from the 2010 base as reported.” The disaster was in the making.

What is really shocking is that the rest of the media did not notice the mistake either.

Going viral

Instead of landing on Snopes, the story went viral. Everybody, from Autoblog to the Zimbabwe Metro happily told its readers that Volkswagen will be the world’s biggest carmaker this year, half a million cars ahead of GM, and 800,000 units ahead of Toyota. Seasoned experts like Paul Eisenstein of the Detroit Bureau swallowed hook, line, and sinker “that sales will rise 13% this year for Volkswagen – which was the number-three maker for 2010 – to 8.1 million. General Motors will gain 8%, to reach 7.5 million. “

Formerly respected papers like the New York Times, the New York Daily News, The Guardian in the UK, the Chosun Ilbo in South Korea , and the Economic Times in India reprinted the admitted garbage. Solid business papers such as the Financial Times, even the world’s oldest car magazine, Autocar in London, fell for the numerical nonsense. Other wire services like Agence France Presse picked up the gobbledygook. Even Automotive News [sub], who really should know better, stepped into Bloomberg’s trap.

Untold numbers of blogs duplicated the drivel.

It is only a matter of minutes until Wikipedia will crown Volkswagen the world’s largest automaker. After all, the threshold for inclusion in Wikipedia is “verifiability, not truth.” And we now have hundreds of reliable sources with the requisite editorial oversight. Or not.

That oversight failed worldwide, except at The Toronto Star. It printed the story, but wisely hedged its bets by commenting that “meanwhile, contrasting analysis at TheTruthAboutCars.com has GM finishing on top for 2011, with VW in second and Toyota in third.”

Twice burned, not shy at all

Seasoned auto and business writers should have been more careful, because they had been burned before. In November 2009, IHS Global Insight, using math bordering on lunacy, crowned Volkswagen as the world’s largest automaker to wide echo in the media. A month later, Volkswagen ended the year in place 3, around 1.2 million behind number one Toyota. IHS, which should have been relegated to counting the moose population of Nome, AK, was one of the experts queried by Bloomberg for this humdinger. To be fair, this time, the error was not that of IHS.

While I scoured the web for someone who has the guts to concede that 8 percent added to 8.39 million can’t possibly be 7.55 million (I finally found a lone voice,) my contact at Bloomberg tried to track down where the error was. Could the Wuling minivans be missing? Could either Daihatsu, or Hino, or both have been ignored at Toyota? After crunching the data looking for faulty logic, my newfound Bloomberg buddy gave up: “I can’t get the number to match exactly.” And he came to the conclusion:

“This looks like a garbage in, garbage out story.”

The B&B chime in

What personally made me sad was that most of our Best & Brightest who did comment on the initial story were likewise duped. They just could not believe that a reputable wire service like Bloomberg can make a simple, but glaring mistake. A lot of the comments were about Bloomberg using sales and silly me using production numbers. Well, first of all, the number to use is production (more on that below.) And second, it does not matter. Whether you use sales or production numbers as a base, in this case, the ranking remains the same.

Sales Production Proj Growth Proj Sales Proj Prod. Rank
GM
8,390,000
8,476,192 8% 9,061,200 9,154,287 1
Volkswagen
7,203,094
7,341,065 13% 8,139,496 8,295,403 2
Toyota
8,438,000
8,557,351 -9% 7,678,580 7,787,189 3

Whether you use the production numbers reported to OICA (hyperlinked), or the sales numbers announced in the official financial reports (hyperlinked in the table above), you will end up with GM a little less than a million ahead of second placed Volkswagen. The actual numbers don’t really matter at this stage of the game. This is a simple 1,2,3 ranking, and no baseball stats.

Congratulations to commenter KitaIkki: He came to the same conclusions as the Bloomberg honcho: “Garbage in, garbage out.”

Car Stats 101

Following recent incidents, here is, for all business and motor journalists, a little primer on car counting. The small lecture on the use and abuse of auto industry statistics has been given before, but some may have skipped class.

The number that decides the final ranking is the number reported to OICA, the umbrella organization of the worlds automobile manufacturer organizations. Whether you agree with the methodology of OICA or not, use their rules. OICA crowns the king of the hill, you don’t. Esteemed members of the media, dear colleagues of the future telling profession: If you want to predict the future king, then you want to replicate exactly how the king is chosen.

  • The reported number is production, not sales. Why? Because only production is halfway comparable around the world. “Sales” is a term of high elasticity and subject to intense interpretation. If Toyota Japan sells a car to Toyota Motor Sales in the U.S.A., then this can be recognized as a sale. On the other end of the spectrum are hard registrations as reported by governmental agencies. Counting units produced is hard enough. There are CKD units floating around the globe. And it is not unusual that there are big differences between the year-end production number and the number reported to OICA for their table that is issued by the middle of the year. Despite the inevitable surprises, always use production numbers.
  • The reported number is on the group level, not by brand. A Toyota for instance includes Daihatsu and Hino. Who is part of the group and who is not is decided by the whims of the group’s rulers. Porsche is not counted as part of Volkswagen – never mind that Porsche owns half of Volkswagen and v.v. If Suzuki and Volkswagen would be counted as one, then the whole largest automaker business would be settled for good. But they aren’t counted as one, never mind that Volkswagen thinks of Suzuki as an associate. Hyundai and Kia are reported as one, despite the fact that Hyundai has fewer shares in Kia than when they were counted separate. Even if Renault and Nissan send out the occasional joint press release, Nissan and Renault are not reported as a group to OICA. Go figure. Attempts on logic will fail. Leave it to the people who run the group. Use what each group uses. It is what it is.
  • The reported number includes all automobiles from cars to heavy trucks and buses, but no motorcycles. Attempts on arriving at a light vehicle count outside of the U.S. and Canada will end in utter disaster. Don’t do it. Careful: J.D. Power often calculates light vehicle numbers outside of the U.S. Mistaking passenger vehicles for the total is a common mistake. (I am looking at you, Associated Press.) Make that mistake in the U.S., and you suddenly end up with 2.7 million units instead of 7.8 million produced. Nobody would do that in the U.S. – but they do it all the time elsewhere. Japan reports regular vehicles and minivehicles separately. Wade into that morass at your own peril. Instead, stick with the total of all automobiles, and you won’t go wrong. Don’t even bother taking out heavy trucks and buses. They amount to less than 6 percent of the world total, and nobody knows for sure where a minivan ends and a bus starts. Use total automobiles.
  • And finally, joint ventures. All cars made by a joint venture are usually recognized as made by the joint venture partner that provides the brand. This may make sense for a Buick or Chevrolet made in China. But why a Wuling? Because GM has a contract that allows GM to count all Wulings as GM’s. This is a matter of intense debate and controversy. You can and should disagree with it, but you should never adjust the numbers. Use the published number, as whacko as it may be.

Any deviation from the above, as sensible as it may be, is the beginning of a debacle. Once you start adjusting the numbers, your base is way off, and you’ll have the dreaded garbage in, garbage out situation that led Bloomberg and the world media astray.

And now for my favorite argument: “Numbers don’t matter!” Sure, everybody (except Volkswagen) says that officially. Sure, officially at the Olympics, it’s the taking part that counts. Both baloney.

Aftermath

Bloomberg killed all mentions of the erroneous story. Easier said than done: The story remains immortalized in the Google cache, and in thousands of Google hits around the universe. These days, when stories race around the globe at the speed of light, and remain accessible for eons without a library card, they should be written with more care and responsibility than ever. Instead, the fact checkers we used to have when I was a cub reporter appear to be among the many unemployed.

And where is the difference between a blogger and a journalist? Frankly, I have no idea. Kris Kristofferson has a song that describes both.

Get the latest TTAC e-Newsletter!

22 Comments on “Partly Truth And Partly Fiction: How Bloomberg Crowned The Wrong King Of Carmakers. And Why They Don’t Teach Math At J-School...”


  • avatar
    chug

    Many journalists, both reporters and editors, are not numerate.

  • avatar
    findude

    Thank you, Mr. Schmitt, for being a journalist. You made calls to original sources and you redid the arithmetic yourself.

    Bloggers (from web log) simply pass on things they have encountered on the web and comment about what they found.

    The journalist actually rolls up sleeves and gets dirty. It’s sort of like the difference between a sports commentator and an athlete.

  • avatar

    I don’t know which was better, the story or the botched headlines. Seriously though, I hope folks stop and think a second. If so many media channels can get a simple story dependent upon mathematical facts wrong, what else are they “getting” wrong when the truth is a bit more elusive.

    • 0 avatar
      racebeer

      I think that USA Today is probably the worst when it comes to screwing up numbers, statistics and graphs. When I taught Six Sigma a number of years ago, it was always easy to look at any given edition of that rag and find an example of poor math and botched statistical analysis. It was fun to use them as a homework assignment for the class!!

      • 0 avatar
        sean362880

        That’s something that bugged me about this analysis the first time around. Take the VW production number: 7,341,065 units. 7 significant digits. That’s an extraordinary level of precision for OICA to claim, especially for something which is so slippery to measure.

        Also, projecting 13% growth (1.13) implies 3 significant digits, not 7. So the projected production number should be 8.30 x 10^6, or 8,300,000 units, not 8,295,403.

        Not that any of this matters for ranking, about which Bertel is certainly correct.

      • 0 avatar

        If 7,341,065 units are reported to OICA, then OICA will report 7,341,065 units. The production number should (SHOULD) not be slippery at all.

        As far as the rounding goes, I’d rather let people do their own rounding. It does not matter. The final numbers will be different anyway.

    • 0 avatar
      chuckrs

      If you told me the concrete in my house foundation had a strength of 2011.737psi, I’d wonder what other screwy ideas you had. Lots of uncertainty there. However, for discrete countable objects, like this, one number is as good as another.

      • 0 avatar
        sean362880

        I’ll concede the point on OICA – maybe the previous year count really is that good. However, regarding future projections:

        “The final numbers will be different anyway.”

        Exactly! So one number isn’t as good as the other. In fact, one number is more wrong than the other. I can give you 10,000:1 odds that VW’s actual sales for 2011 WON’T be 8,295,403 units. I couldn’t say the same for 8.30 x 10^6.

        We all know that this kind of error is seen everywhere, and one could argue that it’s generally understood that the projection of 8.295403 x 10^6 isn’t meant to be taken precisely, but that doesn’t make it any less incorrect.

  • avatar
    GoFaster58

    If Toyota had not been a victim of the earthquake and floods this may have been a different story. I’m not defending Toyota as I believe them to be highly overrated because there are other vehicles just as good if not better that Toyota. And better looking too. However, I believe Volkswagen will overtake Toyota in the future minus the problems of mother nature.

    • 0 avatar

      This is a GM story and not a Toyota story.

      Before the beginning of the year Toyota had budgeted for a rather flat 2011. At that time, it was already clear that GM would pass them by in 2011. The tsunami killed all the forecasts. In June Toyota forecasted only 6.8 million units for calendar 2011. In August, when the situation looked a bit clearer, Toyota predicted 8 million. I am not aware of new forecasts, but the Thai situation should necessitate another revision. All this is and was known.

      As far as Volkswagen goes, they will have a feisty opponent in Toyota next year, possibly even in the first quarter of 2012.

      Both of them however are far away from #1 GM.

    • 0 avatar
      carbiz

      If Toyota did not have a hostage, closed market in the homeland where it can count upon 1.5 to 2 million annual sales, then it would not even be 3rd. ‘Nuff said.

      • 0 avatar

        Sure. There are seven other Japanese automakers and sundry importers who are required by law not to compete with Toyota.

        What’s today’s Kool-Aid flavor? If it’s blueberry, then I’ll have some.

      • 0 avatar
        wsn

        “If Toyota did not have a hostage, closed market in the homeland where it can count upon 1.5 to 2 million annual sales, then it would not even be 3rd. ‘Nuff said.”

        If GM did not have a UAW president, and the bailout, it would become a mere page in Wikipedia in 2008.

  • avatar
    DC Bruce

    Just a theory of course, but I have noticed gross errors and mis-statements of fact (not talking about politics here) from easily obtainable government statistics in a number of reputable news outlets (not blogs). My hunch is that there are lot of inexperienced, poorly paid folks doing this work. Not that that’s new in the business. I was a poorly paid rookie reporter on the Houston Chronicle 38 years ago. But the difference was — and this is my hunch — that the layers of review between my typewritten copy and ink in the paper — rewrite man, ass’t City Editor, City Editor — have been mostly, if not entirely eliminated. Since I was doing police news, with the high risk of a libel suit if I incorrectly reported that some schmuck had been charged with murder when he really only had been arrested on “suspicion of” murder, these layers of review pounded on me pretty thoroughly to make sure I had the basic facts correct. And, just to prove the point, during my tenure, a police beat reporter for the other paper in town (the now-defunct Houston Post) got one of those basic facts wrong . . . and his paper had to pay a large settlement of the resulting libel suit.

    Saves money!

  • avatar
    carbiz

    The Toyota, er, Toronto Star got it right? Wow. I’m impressed! After more than 20 years as a subscriber, the Toronto Star and I parted company about 8 years ago when they published a series of total BS stories about GM’s supposed gas tank troubles on GM’s newer products. The Star chose to put a silly story about some dingbat who ran out of gas on the 400 (a major freeway between Toronto and northern Ontario) in her Cavalier. Of course, that was on the front page of the Saturday paper (circulation of over 1 million at the time), while news that it was, in fact, PetroCanada’s fault (a problem at a refinery that sent a bad batch of gas through Southern Ontario that summer) and that some Nissans and Fords had also been affected were buried in the Business section later.
    It’s still a rag, but I’m glad to see that someone there (probably NOT Mark, however) has the brains to check alternate sources before running total crapola pieces.

  • avatar
    APaGttH

    The silence from the B&B who were insisting the Bloomberg numbers were right, or who were using endless strawman arguments is deafening.

  • avatar
    MarkP

    Back during the Y2K scare, the Atlanta Journal-Constitution ran a lot of stories with questionable “facts” (li,e “airliners will fall out of the sky!”). Some time later, as a sort of apology I guess, the chief reporter did a little piece where she said that reporters were not really trained in anything but how to read public records and how to ask questions. OK, fine, but that means they can’t even recognize the actual facts when they see them. Asking them to do basic arithmetic is an exercise in utter futility.

  • avatar
    th009

    Just a small nit, Bertel: VW no longer claims Suzuki is an associaye:
    http://www.autonews.com/apps/pbcs.dll/article?AID=/20111028/ANE/111029903/1317

  • avatar
    1996MEdition

    “What exactly is the difference between journalism and blog?”

    Proofreading.

  • avatar
    Kendahl

    I know little about most of the topics covered in the news media. Where I do have some competence, I often find errors, some of them as egregious as this one. Makes we wonder how much they screw up that I’m not smart enough to see.

  • avatar
    wsn

    Numbers doesn’t matter. I would very much prefer a Porsche 911 over a GM WuLing.


Back to TopLeave a Reply

You must be logged in to post a comment.

Subscribe without commenting

Recent Comments

New Car Research

Get a Free Dealer Quote

Staff

  • Authors

  • Brendan McAleer, Canada
  • Marcelo De Vasconcellos, Brazil
  • Matthias Gasnier, Australia
  • W. Christian 'Mental' Ward, Abu Dhabi
  • Mark Stevenson, Canada
  • Faisal Ali Khan, India