June 13, 2009

AP Reporters Conned by Pew ‘Green Jobs’ Report (See Updates)

APlogoUpsideDownSometimes the numbers in a wire service report are so ridiculous, you just know that they’re bogus.

On Wednesday, June 11, a duo of Associated Press reporters, Chris Kahn and Sandy Shore, with an assist from Tali Arbel, reported on a study “green jobs” study released by the Pew Charitable Trusts. In “The Clean Energy Economy: Repowering Jobs, Businesses, and Investments Across America,” Pew made the growth in “clean energy” appear more impressive than it is by vastly understating job growth in the rest of the economy during the past decade — by a factor of three.

None of the three AP “journalists” involved, and none of the alleged layers of fact-checkers and editors at the wire service, had the intuitive sense to detect an error by Pew so pathetically obvious that anyone following the economy at all — and that includes the folks at Pew — should have known the figure involved was false.

Here are the first few paragraphs of the AP story (bold is mine):

The fledgling renewable energy industry has grown steadily over much of the past decade, adding jobs at more than twice the national rate, according to a Pew Charitable Trusts study released Wednesday.

Solar and wind-power companies, energy-efficient light bulb makers, environmental engineering firms and others expanded their work force by 9.1 percent from 1998 to 2007, the latest year available, according to Pew.

The average job growth in all industries was 3.7 percent during the same period.

The entire energy sector has experienced growth in recent years as well, according to the Bureau of Labor. Bureau data shows coal mining jobs jumped 16 percent from 2003 to 2009. Oil and gas extraction jobs jumped 28 percent.

So there is no misunderstanding, here are the exact quotes from Pew’s press release, and from its full report (a PDF whose link is on the third line of the press release; bold is mine):

(from press release) Pew found that jobs in the clean energy economy grew at a national rate of 9.1 percent, while traditional jobs grew by only 3.7 percent between 1998 and 2007.

(from full report) But Pew’s research shows that between 1998 and 2007, clean energy economy jobs—a mix of white-and blue-collar positions, from scientists and engineers to electricians, machinists and teachers—grew by 9.1 percent, while total jobs grew by only 3.7 percent.

So the AP reporters looked at some of the information readily available at Uncle Sam’s Bureau of Labor Statistics. Good for them.

Too bad they didn’t look at the following information at BLS, obtained by yours truly in all of three minutes (go to this BLS page and check the top four boxes, retrieve the data, and change the beginning year to 1997 to replicate), that totally debunks the Pew Report’s core claim that “clean energy” is a leading economic contributor.

You see, my research at BLS reveals the following about “total jobs” from the end of 1997 to the end of 2007:


Total jobs grew by 11.1% during the 10-year period in question. Total private sector jobs grew by 10.7%. These percentages are triple and almost triple, respectively, the 3.7% figure served up by Pew and swallowed whole by AP. They are also larger than the Pew report’s claimed “clean energy” job growth percentage of 9.1%.

Thus, the first sentence of AP’s report should have read:

The fledgling renewable energy industry has grown steadily over much of the past decade, but has added jobs at less than the national rate.

Sort of blows the whole reason to bother with Pew’s report, except perhaps to report how fundamentally flawed it is, doesn’t it?

Though Jim Taranto at the Wall Street Journal’s Best of the Web (second item at link) didn’t comment directly on the overall jobs disparity in Pew’s report vs. reality, he did make this observation about the AP’s likely outlook towards Pew’s report that led the wire service to be so lax in its fact-checking:

This misleading framing of the data did not originate with the AP; the Pew press release is titled “Pew Finds Clean Energy Economy Generates Significant Job Growth.” But this points to a way in which media bias often operates. Rarely do news reporters merely rewrite a press release touting a study that comes from a conservative or free-market group. Instead, they take a more critical approach in order to compensate for the group’s bias. There’s nothing wrong with that; the fault lies in journalists’ failing to apply a similar skepticism to liberal groups.

Taranto’s point is valid, but there’s a larger one.

The unfortunate AP reporters involved appear to have been sucker-punched by their media comrades’ “Clinton grandeur followed by recession followed by jobless recovery followed by flat performance followed by recession” rendition of the economy that was recited virtually without interruption during the Bush years. Never mind the reality that jobs on the ground increased at a double-digit rate during the decade reviewed; these reporters were so taken in by the false narrative that they didn’t even recognize how totally and pathetically bogus Pew’s 3.7% job-growth figure is.

This means that the falsehoods of the eight years preceding the ascension of Dear Leader seem to have fed on themselves to the point that the establishment media’s own reporters couldn’t flag bogus information that should never have passed the stench test, let alone the smell test — let alone getting into print and published.

Cross-posted at NewsBusters.org.


UPDATE: Commenter Bradley looked at the graphic in Pew’s Report (saved at my host) allegedly showing percentage state-by-state job growth in “clean energy” and in total. As expected, it has the same errors.

For Ohio, it claims a 2.2% contraction from 1998-2007. Not at all — BLS data shows that Buckeye State employment at the end of 2007 was 5.43 million (Table 5–Seasonally Adjusted Employees on Nonfarm Payrolls), vs. 5.38 million at the end of 1997 (Table 3, same title) — an almost 1% increase. That’s not huge, but it’s enough to count.

But the Buckeye State hasn’t grown much compared to the rest of the country. So let’s try Texas, from the same sources: 2007 – 10.39 million; 1997 – 8.58 million. That’s an increase of 21.1%, vs. Pew’s 6.7%. Zheesh.


UPDATE 2: The footnote at the graphic in Pew’s report says that its source for the national labor growth figure of 3.7% and those of individual states is the “National Establishment Time Series (NETS) Database and data from the Cleantech Group LLC.” NETS is not publicly available.

The specific methodology description on Page 14 of the report (Page 16 of the PDF) says the following (link within was added by me):

This report counts jobs, companies, patent registrations and venture capital investments that are part of the clean energy economy, as Pew defines it, across all 50 states and the District of Columbia. Because a perfect data set with which to count these jobs and businesses does not exist, and obtaining an accurate count of this emerging economic activity is difficult, Pew used data that provide detailed information on individual companies.

As a first step, Pew’s researchers identified companies receiving clean technology venture capital. Next, we used the National Establishment Time Series (NETS) database—a time series database of U.S. public and private establishments based on data from Dun & Bradstreet—to identify similar and related companies. This approach enabled us to capture the different sets of activities that result in products and services produced and supplied by the clean energy economy, creating the most comprehensive and accurate count of jobs yet available. For the purposes of this analysis, we studied jobs and businesses between 1998 and 2007.

Fine. NETS, based on linked description, appears to be a good way to track growth of disparate types of businesses that aren’t broken out in fine detail by BLS. NETS looks at existing businesses and how they did and didn’t grow.

But please note that the two paragraphs above don’t specifically inform us that Pew also relied on NETS for its information about job growth in the economy as a whole. We have to look at teeny-tiny footnotes in a few exhibits for that.

The very detailed explanation of Pew’s methodology at Exhibit B (Page 45-48 of report, Pages 47-50 of PDF) describes in detail the steps taken in using NETS to extract “clean energy” job estimates. I generally have no quarrel with that; in fact, it seems pretty clever. But again, Pew doesn’t specifically say that it also used NETS info to estimate job growth in the economy as a whole. The closest it got was to tells us why BLS and Census data weren’t considered useful for estimating “clean energy” jobs. It never told us why NETS info is better than BLS or Census data for reporting job growth in the entire economy — because it isn’t.

From all appearances, using NETS data as a proxy for job growth in the entire economy is an egregious misapplication of the database. How can it be defended when it obtains a result so obviously at variance with readily available data coming from people who have been at it for decades?

But even if Pew had put out the most wondrous explanation imaginable, the fact remains that the 3.7% result cited for “traditional (i.e. all non-clean energy) jobs” in the press release and “total jobs” in the report itself is wrong, deceptive, and led Pew to a totally incorrect conclusion. That it was blindly accepted by journalists who should know better but nevertheless accepted it on faith because it was coming from people they consider “good guys” is very sad, and all too typical.

Regardless of whether I’ve properly pegged the nature of Pew’s error, the fact remains that however it applied NETS to jobs in the economy as a whole, it came up with an egregiously and obviously wrong result. The BLS data prove it.


UPDATE 3: While we’re at it, Pew’s Exhibit 12 (also saved at host) is using “Total Jobs” numbers that are way, way off about the number of people actually working in the individual states; I believe they are really “total adults of working age.” Glutton for punishment that I am, I added up the 51 numbers presented and came up with over 157 million — 19 million higher than the nonfarm total of people working at the end of 2007 per BLS above.

Individual state totals confirm the error. Examples: Ohio’s 6.30 million per Pew’s Exhibit 12 is 16% higher than BLS’s 5.43 million; Texas’s 11.73 million per Pew is 13% higher.

Though this error hurts Pew’s “case” by resulting in a lower-than-actual percentage of people working in “clean energy,” the more important point is that Exhibit 12′s overstatements confirm the report’s general sloppiness.



  1. Of course its grown steadily. It’s subsided by the government, rushed into being by the trumped of fears of “global warming”, and showered with liberal (wordplay 100% intended) tax favors while less politically correct energy industries are left to fend for themselves, highly regulated, badgered, and sometimes outright threatened with extinction (for instance, coal.)

    Comment by zf — June 13, 2009 @ 11:59 pm

  2. Thank you for posting this. I checked what it said for California, and the same pattern held.

    It claimed total employment from 98 to 07 rose by 6.7 percent, while green jobs rose by 7.7 percent. But according to the California Employment Development Department

    , total employment rose from about 15.3 million in 1998 to 17.1 million in 2007. That is about 12 percent by my math.

    Did I make a mistake somewhere, or did Pew/AP?

    Comment by Bradley J. Fikes — June 14, 2009 @ 2:13 am

  3. You are exactly right, Bradley. Thanks for reminding me to link to a screenshot of their state by state line-up, which, as you noted is also predictably wrong.

    For CA in particular, the numbers at the place you linked represent the total labor force (working and not working). The CA numbers for those working at the links I used in my update are 15.29 million for 2007 and 13.31 milion for 1997, a 14.9% increase.

    Comment by TBlumer — June 14, 2009 @ 7:57 am

  4. The PEW report uses the NETS data and not the BLS data. The BLS data does not technically measure job growth within firms, the point of the PEW report. The NETS data, by way of contrast, is firm-level data and can be used to measure employment growth within existing firms.

    All of this is clearly explained in the PEW report.

    By the way, it is more conventional to speak of “average annual rates of growth” and using your table alone, that number is about 1.4% for the entire non-farm sector.


    I have no relationship to the PEW team. My comments above are based solely on my reading of the PEW report and the posting herein.

    Comment by Samuel Myers — June 14, 2009 @ 8:51 am

  5. #4, I am about to get to that in an update.

    I understand your point about conventional annualized measurement. In that case, Pew should have said 0.87% per year for clean energy and 1.06% for the economy as a whole, or 0.36%. But …. that miniscule a % might not have successfully gotten past the AP filter, and might have blown them up.

    Update: Now that I’ve updated, I disagree about the clarity of Pew’s explanations. They did fine explaining why using NETS is good for estimating clean energy jobs, but they just assumed we’d buy NETS as a good estimator of job growth for the economy as a whole, without defensibly telling us why.

    But they could tell us why until the cows come home and they would remain egregiously wrong.

    Comment by TBlumer — June 14, 2009 @ 9:21 am

  6. The NETS data, by way of contrast, is firm-level data and can be used to measure employment growth within existing firms.

    Then we need an explanation of why the total job growth numbers Pew cites vary so much from the official national and state numbers. Either Pew has identified a major flaw in the official statistics or it has made an egregious error.

    My guess is Pew screwed up. A national job growth rate of just 3.7 percent in a decade makes no sense, when the total US population increased by 11.4 percent during the same time, which encompassed two economic booms and one mild recession.

    If Pew just measured job growth in existing companies, that could have thrown off the numbers. That would have failed to capture the jobs created by new companies.

    Comment by Bradley J. Fikes — June 14, 2009 @ 11:14 am

  7. #6, it’s the latter, as best I can tell. As you noted, concentrating on “existing firms,” it may be ignoring new ones. But its other methods may have picked up new clean energy firms and missed the rest.

    As you intimated, there is no way the official stats are that wrong.

    Your comment also indicates that anyone with even a not very deep base of factual knowledge should have caught it. The AP didn’t. (/surprise)

    Comment by TBlumer — June 14, 2009 @ 11:18 am

  8. Tom,
    For CA in particular, the numbers at the place you linked represent the total labor force (working and not working)

    The place I cited also gives the total employed numbers. Here is a link to an Excel spreadsheet that gives the total labor force, along with total employed and total unemployed.

    Comment by Bradley J. Fikes — June 14, 2009 @ 11:21 am

  9. #8, hmmm. I wonder why BLS’s numbers are so much lower (and they are).

    Aha, here’s what they’re doing. They’re starting with the labor force and just forcing the number working to be the labor force minus those unemployed. That is wrong, because it forgets about subtracting marginally attached workers, discouraged workers, etc. that BLS picks up before arriving at those who are really working.

    That spreadsheet from CA is really weak and misleading. Obviously not your doing, of course.

    Comment by TBlumer — June 14, 2009 @ 11:37 am

  10. Thanks, Tom. I’m going to bring up this whole wretched AP stenography mess with my editors tomorrow.

    What good is a wire service if it lacks basic numerical literacy? Maybe we can get a deduction on our bill for every major error found in an AP story. Or maybe newspapers can organize a truth squad to fisk AP stories for such egregious blunders. We could run these stories to show readers we’re trying to improve accuracy, and to put some heat on AP.

    And your point about the Calif. is also disturbing and needs addressing.

    Call me Data Diogenes.

    Comment by Bradley J. Fikes — June 14, 2009 @ 11:58 am

  11. #10 Bradley, I’d be interested in seeing, and maybe even posting, about what results from editorial discussions, to the extent you’re permitted of course.

    Beyond what you mentioned, AP, at least in OH, was notorious for lifting locally-generated stories without recognizing where they came from. I say “was,” because I think OH papers have sort of cut AP out of the action during the past couple of years. You’d have to contact one of the OH papers to get a complete description of all that has happened.

    Comment by TBlumer — June 14, 2009 @ 12:51 pm

  12. Tom,
    It’s not revealing any state secrets to say my editors are intensely unhappy when they get bogus information from the wire services. I’ll ask my editors about how much we can share, and will direct them to your site. I also posted about your study on a blog: http://abriefhistory.org/?p=809

    Time is our main problem in dealing with such errors. We need a way to quickly flag suspect stories before they run. Perhaps a co-op of local newspapers could form such a truth squad, aided by people such as yourself. That would put pressure on AP to improve the factual accuracy and critical thinking of its reporters, instead of putting opinion in its news stories.

    Yes, I have read about the Ohio newspaper action. A journo friend in Columbus keeps me apprised of Buckeye news. In fact, I’ll be visiting Columbus in early July.

    Comment by Bradley J. Fikes — June 14, 2009 @ 2:46 pm

  13. Bradley, interesting news.

    I’ve read your post. It’s very good. Thanks for the nice words.

    Maybe your subscribing publication and others can force a correction out of them. They sure don’t listen to bloggers, though it’s been a long time since I’ve even tried to get a correction.

    Your monitoring idea is good. Breaking up the AP’s near-monopoly on original news reporting in many areas would be better.

    As long as AP has its near-monopoly, I think it needs a constant, ongoing monitor, instead of scattered bloggers sporadically catching stuff.

    Comment by TBlumer — June 14, 2009 @ 3:44 pm

  14. #10, Data Diogenes is a pretty clever name.

    Comment by TBlumer — June 14, 2009 @ 4:17 pm

  15. #14, Tom,

    Challenging the AP near-monopoly is a great idea. Again, I don’t think I’m going to be thrashed by my editors by saying that we’d like to see more hustle by AP reporters, as well as more critical thinking. Using three reporters to do little more than rewrite a press release is indefensible. At least one of the reporters should have been scrutinizing the data’s accuracy, instead of just running down anecdotes. A few rival reporting organizations could light a fire under them.

    While the AP under Ron Fournier is setting itself up as the Great Arbiter of Veracity for politicians, it’s taking stenography from think tanks and interest groups with their own agenda. Here is one more example.

    Comment by Bradley J. Fikes — June 14, 2009 @ 9:05 pm

  16. #15 Bradley, that is very good work on your part, and pathetic propaganda by the American Lung Association.

    Comment by TBlumer — June 14, 2009 @ 9:20 pm

  17. I did tell my editor about the story’s problems, and he has asked the AP to investigate. I also posted about this on Romenesko.

    Maybe that will light a fire under them.

    Comment by Bradley J. Fikes — June 17, 2009 @ 11:19 am

  18. Mr. Blumer:

    Your BizzyBlog.com post about The Pew Charitable Trusts’ report on the clean energy economy (“AP Reporters Conned by Pew ‘Green Jobs’ Report,” posted June 13, 2009 and updated) cites U.S. Bureau of Labor Statistics (BLS) data to incorrectly suggest that Pew misrepresented total job growth in the overall economy.

    Accurately classifying and counting emerging economic activity, like the clean energy sector, is difficult, if not impossible, with existing government databases.

    To ensure precision in classifying and counting each job, Pew used comprehensive, micro-level data from the private National Establishment Time Series (NETS) database, which in turn, uses data from Dun & Bradstreet. The depth and detail of the NETS data, which cover essentially all U.S. establishments, including small businesses and sole proprietors, allowed our researchers to characterize the businesses and count the job positions at particular companies in the clean energy economy. You acknowledged this yourself in a second update of your post, saying that NETS “appears to be a good way to track growth of disparate types of businesses that aren’t broken out in fine detail by BLS.”

    Using NETS data for one set of numbers and BLS for another would make for an inaccurate and methodologically incorrect comparison. When comparing jobs in the clean energy economy to jobs in the overall economy, Pew used NETS data for both—allowing for an apples-to-apples comparison that put trends in the clean energy economy in the proper context.

    That is why we are confident that our job numbers and growth figures are accurate.

    Kil Huh
    Project Director
    Pew Center on the States
    901 E Street, NW 10th Floor
    Washington, DC 20004
    Work: 202-552-2067
    Email: khuh@pewtrusts.org

    Comment by Kil Huh, Project Director, Pew Center on the States — June 18, 2009 @ 4:02 pm

  19. #18 M. Huh — Nice try, no sale.

    Your report falsely answers this core question:

    – How much did the number of people working in the entire economy increase between 12/31/1997 and 12/31/2007?

    The choices are:
    a) 3.7%
    b) either 10.7% or 11.1%.

    If you pick a), you’re wrong.

    You picked a) in your report. Therefore, you’re wrong.

    In the process, you fooled a few newbies at AP — reporters who at least went to BLS for some individual industry data, but who failed to see the big picture and should have known better.

    You DO know better. You don’t use another “comparative” data point if it is self-evidently false — not if you’re being intellectually honest.

    What I said is that NETS appears to be fine for micro work. But either it’s unsuited for macro work, or you used it improperly for that purpose — and you failed to apply a basic smell test, a virtually unquestioned long-term benchmark at BLS, to the result. Your macro job growth figures are NOT accurate; they are wrong. Pew DID misrepresent total job growth in the overall economy.

    You report’s core claim, that green or “clean energy” jobs grew by roughly 2-1/2 times the rate that jobs grew nationally, is demonstrably false, and you can’t explain it away. Shame on you for trying, and of course failing.

    The honorable thing to do would be to withdraw the report. Given your pathetic attempt at justification, I’m not expecting it.

    Comment by TBlumer — June 18, 2009 @ 10:49 pm

  20. Kil Huh,

    You failed to answer the question: How do you reconcile the NETS data for overall job growth with the vastly different numbers from the BLS and state agencies?

    If the NETS data about overall job growth are correct, then all the government measures of job growth are grossly in error. It would mean employment scarcely grew at all in the decade measured. An explanation of this discrepancy is needed; not just a restatement of what Pew had originally said.

    If the NETS data are not applicable to overall job growth, then a retraction is in order.

    Comment by Bradley J. Fikes — June 18, 2009 @ 10:56 pm

  21. Tom,
    Good timing! I had discussed this with my editor today. After this non-reply, I think he’ll pursue the matter further with AP.

    Unless Pew provides an explanation that fits the facts, I will warn my colleagues that this group’s information is unreliable.

    Comment by Bradley J. Fikes — June 18, 2009 @ 10:59 pm

  22. #21 Brad, I would like to see such a pursuit, and your warning constitutes good advice.

    Comment by TBlumer — June 18, 2009 @ 11:02 pm

  23. I’ve also posted about this on my Facebook and Twitter pages, linking to this latest round of comments.

    Pew simply doesn’t care about accuracy. I would say unbelievable, but by now it’s all too believable.

    Comment by Bradley J. Fikes — June 18, 2009 @ 11:21 pm

RSS feed for comments on this post.

Sorry, the comment form is closed at this time.