Deconstructing The “Massive Beat” in Employment Data- Corrected

Share!Tweet about this on TwitterShare on FacebookShare on LinkedInEmail this to someonePrint this page

Editor’s Note: The earlier version of this article included unadjusted numbers derived from a spreadsheet error. The error resulted in the not seasonally adjusted month to month employment declines being understated. The data has been corrected. The analysis and conclusions are unaffected. Thanks to rdkyote for bringing this to my attention!

The headlines are blaring of a massive surge in January employment that blew away analysts expectations. Frankly, I find it hard to believe that any analysts would not have expected this “news.” The real time Federal Withholding Tax daily data for January, which I dutifully cover each week in the Treasury updates, showed a massive surge beginning in late December. Since everybody didn’t get a 10% raise, the analysts might have inferred that more people were working. Whether that’s a sustainable trend or not is another question, but for January at least, there should have been no mystery.

I like to look behind the headlines at the real unadjusted, unmassaged, unmanipulated numbers to get some idea of what’s really going on. Here’s where things get strange. Total reported employment and full time employment plunged in January, as is normal for that month. So the Gummit survey data doesn’t square with the tax collections. Had we based our forecast for the headlines (which is the only thing that matters to the market in the short run) on the withholding data, we would have gotten it right, but for the wrong reasons. It’s a head scratcher that suggests that the Gummit’s employment numbers shouldn’t be trusted, which isn’t news. What we do know for sure is that there was a gigantic surge in withholding taxes from late December to mid January, and that surge disappeared completely in the last week.

Real Withholding Tax Collections Chart- Click to enlarge

Real Withholding Tax Collections Chart- Click to enlarge

So there’s no question that things were fantastic in January, although why and how that happened is a mystery. Last week’s action suggests that the good news may not persist in February. We also know that the big beat in the headline numbers was an accident. The seasonal adjustment fudge that the Gummit adds to the mix grossly overstated what the actual survey data showed. Here’s a picture. The red line is the actual survey numbers. The blue line is the fake seasonally adjusted number.

Nonfarm Payrolls Chart- Click to Enlarge

Nonfarm Payrolls Chart- Click to Enlarge

Remember: Red… actual. Blue… fake.

Just so you know your eyes aren’t playing tricks on you, let’s zoom in to just the past 13 months.

Nonfarm Payrolls Chart- Click to Enlarge

Nonfarm Payrolls Chart- Click to Enlarge

There you have it. The headline, fake, number was up by 243,000, purportedly the biggest increase since 2006. But what’s this? The actual survey number showed a decrease of 2.7 million jobs. In the world of seasonally adjusted government data, down can be up.

To be honest, that’s a good number for January. Last year the drop was 2.9 million. In 2010 it was also 2.9 million, and in 2009 it was 3.7 million. This year also compares well with the bubble years of 2005 (-2.7 million) and 2006 (-2.7 million). So looking at the top line, the bottom line is that it was a good report, just not the blowout positive number that the headlines reported. It wasn’t a gain, but it was a much smaller loss than in the worst years of the slump, and about as good as any non-recession year.

I like to look at full time employment. There again the seasonal fudging overstated the case. The numbers were not good. Full time jobs declined by 1.2 million in January. That’s worse than last year at 834,000 and 2010 at 1.1 million. But it is better than the 2.6 million drop in 2009 and the 1.7 million drop in 2008. Does that mean this year was just right?

The year to year gain was 1.5 million or just under 1.4%, which was a little less than last year’s 1.5%. Is that a good thing?

Full Time Employment Chart - Click to enlarge

Full Time Employment Chart - Click to enlarge

To put this in perspective, the actual survey data says that 111.9 million people had full time jobs in January. That compares with a peak level of 119.3 million in January 2008. They call that a “recovery?”

I wonder whether these numbers can be trusted at all, given the huge surge in withholding taxes in January. From that perspective, the BLS data would seem to understate the gain. But was the gain in taxes really about more jobs, or something else? What was behind that surge in tax collections is a mystery. Apparently, it may have had more to do with bonuses and sales commissions than a big increase in the number of jobs.

There may be a hint of that in the average weekly earnings report which showed a jump of 1.8% between December and January. Apparently some people got big paychecks during the period. I wouldn’t attribute it to a sudden increase in inflation, at least not yet.

In the end, it’s hard to give any of these reports much credence. The blowout headline numbers are misleading, although the tax withholding data showed that some people clearly enjoyed a windfall from late December through the latter part of January. But then that disappeared last week. The chances are that these employment numbers will be heavily revised, and if last week’s tax data is indicative of what’s ahead this month, the “good news” won’t be sustained.

Given the confusion inherent in these numbers, the proof of whether there’s any real improvement in the employment trend may not come until this summer when peak employment levels are normally reached. In 2009, 2010, and 2011, peak full time employment stalled at about the same level each July. This year and last year the seasonal lows have trended upward. So if the economy really is growing, given the running head start off the lows I would expect full time employment to leap past last year’s highs in May or June. If that does not happen, then we have gained nothing. The initial indications will come in the rate of growth in February and March. Those are numbers to watch. If the growth rate holds up, then the economy is growing, but if those growth rates slow, then we’re probably running in quicksand.

Meanwhile, the government’s own survey data show that 7.4 million fewer people have full time jobs today than was the case 4 years ago. Those 7 million jobs were the fake jobs spawned by the housing and credit bubbles. Those jobs were vaporized when the bubble economy collapsed. They are NEVER coming back. The “new normal” is just the old normal without the added froth. What we are left with is the bitter reality of fewer people carrying the tax load and more people needing government assistance. We have yet to see any real proof that the trends are improving enough to ameliorate those burdens on the economy.

Stay up to date with the machinations of the Fed, Treasury, Primary Dealers and foreign central banks in the US market, along with regular updates of the US housing market, in the Fed Report in the Professional Edition, Money Liquidity, and Real Estate Package. Try it risk free for 30 days. Don’t miss another day. Get the research and analysis you need to understand these critical forces. Be prepared. Stay ahead of the herd. Click this link and begin your risk free trial NOW!

Follow @Lee_Adler on Twitter!

  20 comments for “Deconstructing The “Massive Beat” in Employment Data- Corrected

  1. February 3, 2012 at 3:33 pm

    A note on this from David Stockman.

    “Let’s just call it seasonally adjusted noise. The January change over prior month not seasonally number for 2010 was -2,869k and this led to a -40k drop in the reported seasonally adjusted job number. For 2011, the unadjusted drop was virtually identical at -2,858k and that led a reported seasonally adjusted increase of 110k jobs. And this January the unadjusted drop was -2,689k and the reported seasonally adjusted gain was the headline 243k number. In sum, the change in the January unadjusted number between 2010 and 2012 was 180k but the seasonally adjusted gain was about 283k—-or by exactly the 100k “surprise” amount. The BLS seems to have been scrubbing its seasonals!”

    • Not2
      February 3, 2012 at 10:11 pm

      Come on, David Stockman. You of all people ought to know that you don’t use unadjusted numbers from month to month. You use them year-over-year

      Number of employed people, not seasonally adjusted (BLS Table A-13)

      Jan. 2010: 136,809,000
      Jan. 2011: 137,599,000
      Jan. 2012: 139,944,000

      In 2010-11 (January through January), the economy added a net of 790,000 jobs.

      In 2011-12 (January through January), the economy added another net 2,345,000 jobs.

      In 2010-12 (January through January), the economy added a net 3,135,000 jobs. The most recent year was significantly better than the year before.

      No one can really know exactly how many jobs got added between December and January, because those numbers are seasonally adjusted.

      In raw terms, the unemployment rate actually went up in January, because temporary holiday workers got laid off, just like they do every year. So the 200,000+ seasonally adjusted gain is to be taken with a grain of salt, just as all month-to-month gains ought to be.

      But then, you knew all that. No one has fiddled with the numbers except for you.

  2. Not2
    February 3, 2012 at 6:38 pm

    I realize that good economic news comes as a real blow to those who thought Obama would be an easy target this year, and that they’ll do everything they can to deflect or discredit the facts.

    But the facts, well, they’re still the facts.

    I’ve worked with government economic statistics for many years, including (but not limited to) the employment data. When people want to obfuscate, a favorite means is to misuse unadjusted numbers.

    The appropriate use for seasonally unadjusted data is to compare them to the same period a year earlier. Do that with the employment data, and you’ll see that there were 2,345,000 more jobs in January 2012 than in January 2011, and that the widely reported U3 unemployment rate fell by a full percentage point, from 9.8% to 8.8%.

    Seasonal adjustments aim to smooth out factors like holiday hiring and student hiring, to make monthly numbers comparable to each other. There’s a fair amount of noise in that data, so no one with any sense will draw much significance from a one-month change in either direction.

    But this is the third month in a row for the adjusted numbers, and it’s backed up in each month by the year-to-year comparisons of unadjusted employment and unemployment numbersm, including the broader U indices that include various categories of labor force dropouts.

    Bottom line: We’ve just seen a solid quarter of gains for the economy. I have no doubt that the partisans will come up with one or more of the usual Heinz 57 reasons to be gloomy, but the facts are the facts.

    It sucks when things get better, huh guys?

  3. Not2
    February 3, 2012 at 7:04 pm

    One other thing.

    The seasonally unadjusted unemployment rate for January 2012 was 8.8%, compared with the seasonally adjusted rate of 8.3%. Aha! Obama conspiracy! Must be!

    Um, but look at December 2011, another good month on the unemployment front. Seasonally unadjusted: 8.3%. Seasonally adjusted: 8.5%. And then there was November 2011, when the strengthening started to be noticed. Unadjusted: 8.2%. Adjusted: 8.7%

    Wow, who was hiding the light under the bushel! Why, a conspiracy by Obama to hide the good news until it would be more convenient for his election campaign, no doubt!

    Now, I don’t know if this site is run by Republican propagandists or not. But don’t quibble with the facts, or try to fuzz it up by comparing apples to oranges. The economy is getting better, and you know it.

  4. Heather
    February 3, 2012 at 9:34 pm

    I think it’s very discouraging that those long-time unemployed remain unemployed at the same rate, the underemployed/part-timers percentage hasn’t budged and that there was actually a drop in the labor participation rate, probably because people gave up. I don’t think the BLS report is a conspiracy but I also don’t think it means real progress. I am a moderate and voted for Obama in the last election, fyi. Not everything is political.

  5. Not2
    February 3, 2012 at 9:51 pm

    Heather, no one is trying to say that happy days are here again. It’s a strawman argument to dismiss the recent gains because those gains haven’t brought us back to the golden years of 1999.

    There are 2,345,000 more jobs than there were a year ago, and unemployment has dropped by 1% across all the BLS indicators, including the one that includes all of the “discouraged workers.”

    Long-term unemployment is down a little bit, but not enough. Sheesh, it took nearly 20 years for the damage done on the employment front in the 1970s to be undone, and you’re moaning because Obama hasn’t undone the effects of the biggest depression since the 1930s in three years?

    The BLS report factually represents real progress. Do we have a long way to go? You bet we do. Oh, and anyone who depicted the latest report as a “massive” anything is wrong. But there are still 2,345,000 more people working than there were a year ago, and I bet that a whole lot of them are glad to have a job.

  6. Cat from higher roof
    February 4, 2012 at 2:59 am

    Lee, I can give a clue where the tax withholding hike came from. I don’t believe any of those numbers, except may become reflection that people are giving up looking for good job and just taking any job. Any part time job still means no benefits, even 35 hours, on 8 bucks. It’s time. As unemployment benefit’s 92 weeks (where available) are exhausted, people realize that savings are not endless. The bad numbers in 2008 were necessary to scare congress and get stimulus. Fed was denying manipulation of market with simple explanation from supporters: nobody can manipulate it because it is too big. Later we found that drop of market is estimated 4 trillions, it means the subsequent miracle 50% recovery or 2 trillions is thanks to 4 trillions from the Fed (the 700 billions was just smoke for masses). It’s quite good efficiency for government (real branch),50%, it’s usually less then 30%, but 700 billions were really a smoke wall, some time ago Fed had admitted that it’s really 17 trillions. It explains a lot, including bonuses. Now, as I promised, here are clues regarding to tax withholding hike: two people in my circle lost full time job on December 30th of 2011. Does severance pay ring any bell to you? In addition one of people those still work full time (obviously, just very few very close people share salary information, it’s US) noticed that his paycheck dropped about 7% in January 2012, the explanation he has got is that some taxes were not withheld before. Coincident? It’s well known – statistic may be the biggest lie, just know what and when to present. Have we calculate everything as it was in 30th, and the picture will be really different. My best regards to Lee Adler, it nice to have at least somebody looking for real answers and giving real numbers. Have you noticed, yahoo does not give Failed Banks List” quick link on search “FDIC” any more, and it looks like the list does not give missing assets numbers any more?

  7. TDS4S
    February 4, 2012 at 11:42 am

    I doubt anyone will read this since the article is a day old now. But my dad is a military retiree. He got a note saying that congress had moved his retirement pay date from the last day of the month to the first. So he got December’s check on the 31st and then January’s on the 1st. If that happened to all government retirees, then that would account for a massive jump in with holding taxes without a massive jump in jobs.

  8. Heather
    February 4, 2012 at 12:23 pm

    Not2: Well, Gallup is reporting an increase in underemployment, unemployment and its latest weekly results contrast sharply with the government’s recent report that jobless claims fell to a three-year low during the week ending Dec 10. There are also a plethora of blogs questioning the suspicious seasonal adjustments. The only numbers that can be trusted as they have not been manipulated are the unemployment/underemployment numbers, which still show a very bad economy. The fact that almost half of the unemployed are long-term unemployed makes it all the more dire. http://www.gallup.com/poll/151601/Underemployment-Mid-December-Similar-Year-Ago.aspx

    • Not2
      February 4, 2012 at 6:18 pm

      The blogs that are questioning the BLS data are either perma-bears like the wackos at “ZeroHedge,” who either failed to grasp or purposely misrepresented the difference between seasonally adjusted and unadjusted numbers by mixing them, or are conservative sites that want to spin the data to make it look like the White House has somehow altered the BLS’s methodology.

      BLS statistics are perfectly suited to this sort of game-playing, because of how much data they report. There is a household survey and an employer survey of employment, and each is reported with both seasonal adjustments and in raw unadjusted terms.

      On top of that, the BLS also publishes a series of alternative unemployment measurements that take various categories of “discouraged workers” into account.

      There is a right way to read these numbers, and a wrong way. Those who want to spin them for political advantage have plenty of opportunities to compare apples and oranges. The “analysis” at this site is an excellent example of such spinning.

  9. Yes1
    February 4, 2012 at 4:06 pm

    Not2 is onto something in the sense that there is a cadre of conservative-leaning doommongers who want Obama to fail at every turn and are looking at everything to make him look bad.

    That being said, the participation rate have been declining all of Obama’s term and if it was constant in Jan 2009 terms all through his presidency, then we’d have U-3 unemployment at 11 %.

    Hardly a great improvement. So, yes, good news but 8.3 % is a monstrous lie and ‘Not2’ knows it and he’s just as partisan as the resident Republicans here.

    We need more independents.

    • Not2
      February 4, 2012 at 6:10 pm

      8.3% is not a “monstrous lie.” It’s the U-3 number. The BLS also issues other “U” numbers, and they’re freely available.

      The declining participation rate is definitely interesting. Part of it is that the baby boomers have started to retire, but part of it is also that the economy is in very rough shape. How big these parts are, I’m not sure. I’ll await something from the BLS that analyses it.

      The BLS is a non-partisan agency. They make certain avoid methodology changes within a particular administration. Changes are debated over a period of many years. None of the numbers recently reported are “Obama spin,” just as favorable numbers reported at certain times under Reagan and both Bushes were not “Republican spin.”

      Fact is, there’s been a quarter of reasonably good news on the employment front. In fact, 2011 in general was significantly better than 2010. Those are facts. If you want to deny them, then we don’t have much to discuss.

      p.s.: The sun will rise in the East tomorrow morning.

  10. goleris
    February 4, 2012 at 5:47 pm

    There are many reasons why the mid-Jan 2012 tax withholding saw a big jump compared to late-Dec 2011.

    1) For those making more than 106.8k in 2011, their last paycheck of 2011 may not have social security tax withholding since they had paid the max limit for that year already. But starting Jan 2012, they have to start paying social security tax again. The constant here is that the rate is still 4.2% (no change from Dec 2011) so far.

    2) January is the month of bonuses in many companies, as many have pointed out already.

    3) Severance packages for employees who got laid off in Dec 2011. Employees who are tax-savvy can get their employers to defer the payment to early Jan 2012.

    4) As somebody pointed out, retired military veterans got their Jan pay check on Jan 1 instead of Jan 31st in 2012.

    In fact, reason 4) is uniquely applicable to 2012 but not the past few years (2011, 2010 and so on) and could explain if and why this jump is more compared to Jan 2011.

    In all, it does not prove that unemployment is down as a percentage.

    Also note that the labor force increases every year (US population is increasing, people are retiring increasingly at an older age etc.) which means even if the number of employed people increases, that may not translate to a lower unemployment rate.

    The real proof of unemployment reduction may come only when the US national debt starts shrinking from its current 15 trillion USD. After all, if more people are employed and people’s wages are going up then govt will take in more taxes and spend less on unemployment benefits and that should bring down this monstrous debt. But that is assuming govt keeps taxes constant and does not stop paying unemployment benefit to the unemployed. We all know that assumption won’t be true since social security tax will go back to 6.2% (from 4.2% now) in March, 2012 and unemployment benefits will also run out soon for millions of people. In other words, you will soon see another “jump” in tax-withholding in March, 2012 but that will mean nothing in terms of economic progress for the vast majority of people in this country.

    Lastly, its the red line (real data) that matters, if at all. The fake blue fitted line which can be twisted and turned to mean anything should simply be ignored.

    • Not2
      February 4, 2012 at 6:12 pm

      The real proof of unemployment reduction may come only when the US national debt starts shrinking from its current 15 trillion USD.

      You mean like in the 1980s, when unemployment peaked in 1982 but the debt rose all the way into the mid-1990s?

      • goleris
        February 4, 2012 at 7:34 pm

        I don’t know how you got so confused. As I said earlier, if unemployment actually dropped and at the same time the debt also dropped then that would be the sign of real economic progress. Its not so hard to comprehend. Won’t you, as an individual, like your earnings to go up (or atleast not decline) and your debts to do down simultaneously?

        Regarding your comment about unemployment going down after 1982 while debt rising till mid-1990s, that is not an option anymore with the current debt already at 15 trillion and rising.

        Note that if unemployment reduction is being achieved via more debt generation, thats not much progress. You could always tell your children and grand-children and great grand-children “let me have my job so that you all can be debt-slaves all your lives. I still love you.”

      • Not2
        February 4, 2012 at 8:05 pm

        As I said earlier, if unemployment actually dropped and at the same time the debt also dropped then that would be the sign of real economic progress.

        That is not what you wrote. You wrote: “<he real proof of unemployment reduction may come only when the US national debt starts shrinking from its current 15 trillion USD.”

        As for all the rest, this thread is about the BLS employment release. The BLS didn’t issue an “economic progress” number. The issued employment numbers, and now we have partisans who are trying to distort them.

        Just for the record, I think the economy is in tough shape, just as it was pretty much throughout the 1980s and into the early 1990s. Back then, it took 15 years to get unemployment (U-3) down to the 4.x% range. The surpluses of the Clinton years were generated not just by a strong economy, but by tax policy choices.

        In any case, we might have a few things to agree with regarding the economy, but facts are facts, and the is nothing fake about the employment numbers. To quote Daniel Patrick Moynihan, everyone is entitled to his own opinion, but everyone is not entitled to his own facts.

      • goleris
        February 4, 2012 at 8:53 pm

        I am not sure what you are trying to argue about.

        To quote myself:

        “The real proof of unemployment reduction may come only when the US national debt starts shrinking from its current 15 trillion USD. After all, if more people are employed and people’s wages are going up then govt will take in more taxes and spend less on unemployment benefits and that should bring down this monstrous debt.”

        I see that you conveniently left out the last part of my above comment for whatever reason.

        Regarding your comment:

        “…the is nothing fake about the employment numbers. ”

        its clear that you are trying to create a straw man. I simply claimed I will only go by the real data (the red lines in the charts), not the adjusted ones (the blue lines). Going by the chart in http://wallstreetexaminer.com/uploads/graphic1441.png its clear that the unadjusted data indicates about 130.25 million non-farm jobs at end of Jan 2012. Thats the real number of jobs. The blue line indicates about 132.4 million but that is not useful in this context. Thats a fact you cannot deny, especially since you are not entitled to your own facts.

        Following this reason, it is easy to see that real U3 unemployment rate in Jan 2012 is 9.8% (I trust you can do the math). If (1-132.4/144.4) = 8.3% then (1-130.25/144.4) = 9.8%.

        It is evident from http://wallstreetexaminer.com/uploads/graphic1440.png that the actual number of non-farm jobs from Jan 2010 and Jan 2011 were worse compared to Jan 2012. If the national debt had dropped or even stayed the same during this period (instead of blowing up from 2T to 15T), then there may be justification in claiming that the economy has improved.

        I could go into what “Cat from higher roof” stated earlier about people taking lower-paying jobs just to stay alive but that would only expose the miserable state of the economy even more. I will leave it at that for now.

    • Peddler
      February 6, 2012 at 8:11 pm

      You are correct, the number of people claimed as dropping out of the labor force did not occur last month. What happened was an adjustment based on the 2010 census numbers and the BLS then applied their own statistical adjustments to the actual number of 1.7 million people who were not included until after the census information was accounted for.

      What the BLS did was to make adjustments to the number by reducing the people available in the work force in the age groups of 16-24 and 55 and older. The reason the BLS uses adjustments for these age demographic groups is that they traditionally had lower levels of labor force participation.

      With that statistical adjustment accounting for the good numbers, 8.3% unemployment, what it does not do is make a realistic adjustment to the larger of the two demographics, the 55 and older group. For the numbers to be accurate, the BLS should take into consideration the number of people 55 and older who have actually re-entered the work force due to loss of income, loss of retirement benefits, and all of the other factors relating to the deep, near depression recession we are still mired in.

      Another consideration, the 16-24 demographic are not as prone to not seek employment as they once were. Now, with families having so much stress on the income mom and dad once provided, they are out seeking work to assist, that is if work is available.

      So, the model used previously is now an outlier and not relevant to the current crisis we are facing.

      Statistics = lies, lies, and more lies.

      If you have information that counters what I have offered, please provide it.

Leave a Reply