Sunday, February 7, 2010

lies, damn lies, & statistics

a lot of changes came down in last Fridays unemployment report, and for the most part, the media only reported the headline number, maybe breaking out part time/temporary or hours per week or similar stats, usually with an occasional anecdote or example, without digging below the surface of the government reported numbers…fortunately, a number of bloggers have each tackled the jobs the MSM journalists should have been doing, so ill try to gather the more salient points and present them here…

first, ill send you to CalculatedRisk, where he tackles the fact that the both the payroll jobs number and the unemployed rate declined, which even confused krugman…more details there, but the easy explanation is that this is because the data comes from two separate surveys; the establishment survey (employers) showed a loss of 20,000 payroll jobs in January, but the much smaller household sampling showed an increase in the employment level of 541,000.

next, ed harrison explains the effect of the seasonal adjustments: Unemployment number decline is all about seasonal adjustmentsA lot of people are questioning the unemployment rate of 9.7% in the face of a –20,000 non-farm payroll. How could we be losing jobs and have the unemployment rate drop? It would seem people are dropping out of the labor force.  However, I have now parsed the household survey data and most of the data seems reasonable. The labor force participation rate actually ticked up slightly (both seasonally adjusted and unadjusted) – as did the number of people not in the labor force who wanted a job (unadjusted only). This is what we would expect. What sticks out is the seasonal adjustment for the number of persons employed and unemployed. In December 2009, there were 15.267 million people unemployed on a seasonally-adjusted basis.  This ticked down to 14.837 in January 2010, a fairly large drop of 430,000. Meanwhile the unadjusted numbers go the other direction – massively.  In December 2009, the number of unemployed persons was 14.740 million. This rose 1.4 million in January to 16.147 million.     … the above link has detailed tables…and brad delong charts & then points out that this “is the fifth year in a row in which the year start has been accompanied by a drop in seasonally-adjusted UI claims that is then immediately reversed--suggesting that there might perhaps be a small bug in the seasonal adjustment process.”

then, kid dynamite and others have explained that the annual Unemployment: Birth/Death Adjustments Overstated Jobs by 902,000 - Leading up to the release of today's BLS Employment report, there was a lot of talk about the birth/death adjustment.  The b/d model can be a bit tricky to explain - I think Barry Ritholtz's old piece here does a good job of summarizing what happens.  Basically, the BLS estimates job creation or destruction based on the filings of new company incorporations - because companies which are too new or too small don't get counted in the CES (current employment statistics) survey.  …so obviously, comparison from december to january arent even on the same playing field, let alone a level one…Bloomberg also has some interactive charts to show how this adjustment works in an article titled Birth Death Model Insights….

Jesse @ Le Cafe Americain explains what this means:  Non Farm Payrolls Benchmark Revision and the Unemployment Rate as Cruel Farce - As you may have heard, the Bureau of Labor Statistics did a benchmark revision. This is Washington speak for 'revised the numbers as far back as anyone might care to remember to give ourselves more wiggle room.'  The benchmark is a product of the Bernays Factor, that measure of public gullibility which permits obviously contrived government statistics to be taken seriously. Did you react to the positive jobs trend initially announced in September - October 2009? Oops, it was really a greater loss than expected, and not a gain at all. One can only suspect that in a few years this whole recovery could be revised away without so much as a bureaucratic blush. Here is a picture comparing the old and new headline numbers.. The change is pervasive. One item of note is the taking of more job losses in the earlier years, setting up a stable base for potential job gains in the present, without embarrassing oneself by getting out of synchronization with the actual growth of the civilian population. There will be more 'truing up' of the numbers in the future.

there are other unemployment reports which may be more reliable…

TrimTabs: Here's Why The Real Jobs Loss Number Was 5x Worse Than What The BLS Reported - TrimTabs employment analysis, which uses real-time daily income tax deposits from all U.S. taxpayers to compute employment growth, estimated that the U.S. economy shed 104,000 jobs in January.  Meanwhile, the Bureau of Labor Statistics (BLS) reported the U.S. economy lost 20,000 jobs.  We believe the BLS has underestimated January’s results due to problems inherent in their survey techniques. In addition to their regular report, the BLS published benchmark revisions to their employment estimates derived from an actual payroll count for March 2009.  As a result, job losses from April 2008 through March 2009 were revised up a whopping 930,000, or 23% from their earlier revisions.  In addition, the BLS revised their job loss estimates for 2009 up 617,000, or 14.8%. While the BLS originally reported job losses of 4.2 million in 2009, TrimTabs reported 5.3 million, a difference of more than a million lost jobs.  We consistently reported that based on real-time tax data, job losses were much higher than the BLS was reporting.  This past January, the BLS revised their job loss estimate to 4.8 million, an increase of almost 600,000 lost jobs.  The new total brought the BLS’ revised estimates much closer to TrimTabs’ original estimate based on real-time tax data.

and there’s also john williams at shadow stats, who produces an alternate unemployment report (with chart), which shows U-6 still above 20%…

in addition, something noone mentioned, but im sure is a factor in what will be a temporary bump in the employment numbers, is the 1.2 million or more census workers now being hired…

i have previously posted an unemployment chart comparing job losses in this recession to others: here’s a link to a chart  showing the employment-population ratio; this is the ratio of employed Americans to the adult population….
and just so i havent left any thing out, here’s a FRED chart illustrating the real problem: how long people have been out of work, compared to other periods in our post-war history…


  1. Very informative. Thanks.
    Dec. usually has a much higher seasonal xmas shoopping temp workers. Not this year. The unemployment numbers are bogus beacause the govt. does not want to scare the people more than they already are. Not enough attention is being paid to jobs, jobs, jobs.
    They made us into a service, and consumer society. Now there is no one to service, and there are no consumers. In my business every one is down 30-50%. They are running out of time. Drastic times call for drastic measures. I am afraid the peeeps will not like the drastic measures that are coming.

  2. in the comments section on kid dynamite's post, linked above, Transor Z points to his musings on what the unemployment rate would be if so many people hadnt dropped out of the labor force during this recession, and then later links to his just completed paper, with charts, which concludes that the U3 would be 11.5%, instead of 9.7%, if those dropouts were included...

  3. Thanks for the mention, rjs. I won't put a link for my piece here to encourage folks to go check out Kid Dynamite's nice post on the BDA.


  4. Excellent post, RJ. I've been aware of massaging the data for some time, but this aspect is new to me. Thanks.