With the NFP payroll report out tomorrow morning, Bloomberg.com put together a timely infographic on one of our favorite pet peeves: the Birth Death Adjustment.

Overall, the piece is good (see charts below). One small quibble: They should have mentioned that int he beginning of the cycle, the B/D catches job creation that the usual methods miss; problem is that at the end of the business cycle, it misses job losses the usual methods catches.

Click for interactive charting:



As in so much of life, improvements in one part of the model cause problems in another . . .

Category: Data Analysis, Employment

Please use the comments to demonstrate your own ignorance, unfamiliarity with empirical data and lack of respect for scientific knowledge. Be sure to create straw men and argue against things I have neither said nor implied. If you could repeat previously discredited memes or steer the conversation into irrelevant, off topic discussions, it would be appreciated. Lastly, kindly forgo all civility in your discourse . . . you are, after all, anonymous.

12 Responses to “Bloomberg on the Birth Death Adjustment”

  1. investorinpa says:

    At this point, shouldn’t they just go with a payroll tax based model…or come up with a combo of multiple methods…I still think its stupid that job numbers could fall but unemployment numbers increase. I know you’ve explained how it is 2 separate surveys, but come on, we can do sooo much better.

  2. JesseLivermore says:

    Does this mean the unemployment numbers get revised upwards as well?

  3. Transor Z says:

    If there’s one thing we can all agree on it’s those damned incompetent Washington bureaucrats. The errors in BDA are a direct result of the CRA, subprime lending to minorities, and bailing out the banks. Any change to the BDA will contain hidden loopholes for lobbyists and more taxpayer bailouts to banks.

    Signed, Luntz

  4. MikeG says:

    BR, I’m seeing the GIANT FONT issue again as I scroll down the page…

  5. bsneath says:

    So the birth-death model clearly understates job losses at the beginning of a recession and likely understates job creation at the beginning of a recovery.

    Thus in using this model, BLS has rendered the monthly numbers to be inaccurate. However BLS will not change the methodology because overtime it averages out (after massive baseline adjustments of course).

    I would put forth that this is just another example of why those who have faith in the efficacy of government should reconsider.

  6. Transor Z says:

    So XYZ Corp. is correcting material misstatements in its financials going back three years.

    Thus in using this accounting model, XYZ has rendered the annual numbers to be inaccurate. However XYZ will not change the methodology because the misstatements were caused by alleged conspiracy and fraud between the CEO, CFO, and the accounting firm.

    I would put forth that this is just another example of why those who have faith in the efficacy of privatization should reconsider.

  7. torrie-amos says:

    well, it’s just one of those things, probably pretty good for 30 years or so, yet, there’s always an exception, to start a biz you need money and a market, folks laid if this round, well, they aint got neither, thus, ain’t happening

    does anyone know anyone starting there own biz now? in what? employ more than 1?

    i know of none, in 2000-2001 a half dozen, in early 90′s downturn again half a dozen, that’s good enough for me

  8. willid3 says:

    i am thinking the state and federal payroll tax data would make a better choice to judge employment, especially if it was done ever month, but even if it wasn’t it would give a better yardstick to judge employment but it would have to be a required action by employers, even if they went out of business to do. and why not judge the economy as a whole based on state sales taxes as much as others? even if they are based on the honor system?

  9. advocatusdiaboli says:

    I think the true test of the validity of a model isn’t just how often it works though that is important, but whether it works reliably when you are most dependent on its accuracy such as in recessions. Because this model falls down precisely when we need it most, it needs to be abandoned. But of course it won’t be: what government would want a more accuracy in reporting on their failure to stimulate jobs growth?

  10. Transor Z says:

    I checked into this last year. BLS and Census work together and use what’s supposed to be the most comprehensive database of businesses. The BDA modelling is based on what’s in the database. Supposedly the database is constantly benchmarked against monthly state UI reporting. But there’s a 5-year moving average in the mix, which IMO is the real problem with accurately tracking business cycles.

    So I called/emailed about a dozen state Secretary of State corporation divisions to see what kind of data they have. The Colorado office can’t even tell you how many total businesses are incorporated in Colorado. AYFKM? In an Access database it’s really hard to do that. You go to the bottom record, look at the row number, and subtract one for the header row. You would think somebody could tell you how many active records are in a database…

    Some states capture detailed info on annual incorporations and dissolutions, breaking it down by type: C-Corp, LLC, LLP, etc. But, at least with dissolutions, some businesses fold and don’t bother reporting it to the Corporations Division and eventually they got dissolved administratively. So the incorporation/dissolution data at the state level is sketchy, to say the least. But that should be the most direct way to get BDA data. But I guess that would be too easy; thus the need to model.

    The interesting thing I found was that, while the # of incorporations has remained flat or increased/decreased slightly YoY in the states I looked at, YoY change in # of dissolutions has steadily climbed for the past 10 years. So dissolutions as a % of incorporations has been markedly rising since the dot com bust. I just thought that was interesting.

  11. Chris says:

    I wonder if governments have any additional information respectively better models on the job market for internal use only or if they really base their decisions on their ‘public-relation’ models? Same goes for inflation or any other economically relevant statistic.

  12. foxmuldar says:

    What happened to those green shoots? I didn’t see any talk about them today during the bloodbath to 10,000. Instead of wondering about what numbers the government puts out, try asking yourself why analysts are always wrong. Why they keep calling for lower unemployment numbers and then getting it so wrong. Most analyst work for brokerage firms or banks. Its best for them to sound positive so that stocks move higher. Thus we hear about green shoots and jobs just around the corner. So much BS. Analyst and Economists get it right maybe 10% of the time and continue to stay employed. Why?