Maybe someone can make proper sense of this. If you like big data
bases, understand complicated analytics and have a big computer read
on. If not, let me suggest you click out now.
I have watched the Non Farms Payroll numbers for most of my life. I
would guess that there are a few hundred economists who spend a week a
month trying to make an educated guess at the headline. The MSM talks
about the numbers for days before and after the release. The financial
blogs are even worse. The fascination with the numbers is that they are
damn near impossible to predict, and they always move the markets.
The fact is the Government publishes the critical data necessary to
evaluate the current month employment and unemployment numbers weeks
before the actual release of the NFP report. The February information
is out, here it is:
The Social Security Trust Fund releases the February Payroll tax
receipt numbers early. Obviously this can’t be a completely accurate
number. February is not even over, so clearly this is an estimate. I
have followed these numbers for some time and have found the
preliminary numbers to match up pretty well with the adjusted count.
In looking at the numbers I have concluded:
-Over time there is a 100% correlation to rising employment and rising
payroll tax receipts. Similarly the correlation exists where in periods
of recession the rate of increase falls, or as in the case of 2008-10
the income goes negative. The severity of the recession determines the
actual outcome.
-In the 1/2008-1/2010 periods employment fell by 8.24mm (revised BLS
number). The Jan.+Feb. 2008 FICA total payroll tax was $98.7b. In 2010
it was 93.0b a drop of 5.7%. Multiply that times 135mm workers and you
come up with a drop in payrolls of 8.27mm. That is a 30k difference
over 24 months. That is a very high correlation. That this lines up so
well is not a coincidence.
-The month-to-month correlations are not reliable. There are lags in
the reported data. For example, if a person were fired it would show up
in the BLS numbers the next month. But if the person got a severance
that income would continue to be subject to SS tax. There would also be
a distortion from people losing regular work and taking a job as a
waiter at $5 per hour. This person would not show up as unemployed, but
SS revenue would decline.
To look at this I think it would be necessary to look at the data both
from SS and the BLS over an extended period of time. Twenty years of
monthly data. It is all available. The question is can the seasonals
and the leads and lags be evaluated and a methodology developed that
would translate the SS payroll tax number into a predictor of the BLS
NFP number?
I have two concerns with this:
A) Assume (100o to 1 chance?) that someone says, “Eureka! I have the
number +/ 25,000 as a result of the peek at the tax data. (A very big
deal) The minute that happens the Trust Fund will reschedule the
release date of the data. The punch bowl will be gone before the party
starts.
B) My worst fear is that there is some predictive value to
looking at the tax numbers. And that someone figured this out four or
five years ago and has been getting rich once a month ever since. I am
not the only one who looks at these silly things.
Notes:
*Consider the February 2009-2010 numbers. (I use only the monthly FICA
projected # as this is 95% of the data) 2010=47.5B, 2009=$51.5B. We
lost $4b YoY. That is a very big drop. It comes to 8%. There are 160mm
covered workers paying into SS. They range from an exec. to the
paperboy. Based on this math 13.3mm are no longer getting a check and
contributing to the system. This suggests that the NFP number is
greatly understating the drop in employment. There would appear to be
some very bad things happening outside of the BLS numbers.
*I don’t understand the leads and lags of this. But I have played with
and watched the numbers. It is hard for me to accept the current wisdom
that there will be a large increase in the February NFP given the drop
in tax receipts.
*For SS to be seeing a decline of 8% in revenues in February is
disturbing. Some additional months of data in 2010 are necessary to
prove this trend. But should this decline hold, it is a very poor sign
for the Fund.
Forget US-centric. I'd like to see a voter referendum for Tata Consulting Services in India to audit the IRS: Your social security number, salary in 1999 compared to 2009. Any decrease is the unemployment rate. Alan Greenspan and dimwit US Labor Secretary Elaine Chao have both chirped "health sciences" as the only jobs left in the US, and transitioning to a health sciences career is a long, long, 3-5 year process.
Okay Bruce, here is the flaw in your methodology.
Your premise is BLS is arriving at the number through competence.
What if these guys pick the first digit based on previous months...
In this case 4x,xxx and randomly generate the rest?
1. Economists and statisticians can never figure out the BLS models, no matter how much time they spend.
2. The BLS look smarter than economists (how hard is that anyway) and perpetuate the myth that they know what they are doing...
Not to mention, random generation takes 5 minutes and 8 government employees swearing secrecy to each other then leave for lunch at 9:30 and do not coming back until Thursday.
Sometimes it is just easier to invent things...
After all, Paulson and Mini-me, Neel "Down Biatch" Kashkari pulled the number $700 billion together out of one of their asses...
You have to use the majical numbers spin machine that gives the false numbers that are reported as the real numbers then revised later to some less false numbers.
Considering the seasonal adjustments to the raw data to achieve the headline number the SS data is a input to what is essentially an estimate written in sand, that will be revised repeatedly after it is released.
There was a graph on clusterstock maybe yesterday plotting income levels vs. unemployment figs. I shouldn't have been surprised, but I was. The lowest ranks are getting slaughtered, close to genocide, and it's coming our way quickly. You read about it what, 10 times a day?
That would affect your figs if for no other reason than SS is regressive whereas mc is flat (as someone pointed out above).
The more interesting question posed by BK is whether someone already has the advance NFP fig? And as early as what, 2005? Why not 2001? But at this point you've got double trouble due to potential infection from Heisenberg or moral hazard or feedback or what have you. Best of luck.
A company can show an increase in the average wages paid simply by laying off all the lowest paid people and not giving anybody else a raise.
Perhaps you should combine the information with state sales tax reciepts, which would be accurate. Those in combination with the income reciepts would perhaps shed light on the picture. Example, person loses job and get severence, they will not spend. Employee hears rumor that there will be layoffs, he/she does not spend. People who feel secure including self employed, will spend and therefore pay sales tax. Another thing you are forgeting, is that self employed/small business people do have the capacity to control their outflows of withholding taxes, unlike big corporations. you get a couple million small businesses (for cash flow reasons) not paying witholding or shifting income into different quarters, that can change those numbers.
Bruce,
Select HI only from the menu. FICA doesn't capture every dollar in wages and bonuses. The difference really shows up from June forward. I also watch the <a href="http://www.ssa.gov/OACT/ProgData/icp.html" target="blank">Benefits Paid by Type of Beneficiary</a> to pick up the rate of change in number of beneficiaries. In 12/07 SS recipients were being added at an annual rate of 553K, at 12/08 it was 748K, at 12/09 it was 1,240K. Those last two numbers exceed BLS/Census estimates by a considerable amount. It alos obviously impacts the total number of people employed.
The 2.9% HI contribution ties in pretty well to the BEA wages and salaries line on their Personal Income report. I would think that a bit of backcasting would allow you to use FICA, HI and the BLS average hourly wage to formulate a simple algorithm. I haven't tried it as yet but the data is certainly available.
(sorry if the HTML doesn't work)
I played around with that "simple algorithm" you speak of. I did not find it. It may not be findable. I don't think it is simple.
I look at this as if it were a gigantic pile of horse shit. Most people would say that it stinks. I say, "With all this shit there must be a pony someplace!".
So I am looking for the pony.
Goldman Sachs, Greece Didn’t Disclose Swap, Investors ‘Fooled’
Feb. 17 (Bloomberg) -- Goldman Sachs Group Inc. managed $15 billion of bond sales for Greece after arranging a currency swap that allowed the government to hide the extent of its deficit.
No mention was made of the swap in sales documents for the securities in at least six of the 10 sales the bank arranged for Greece since the transaction, according to a review of the prospectuses by Bloomberg. The New York-based firm helped Greece raise $1 billion of off-balance-sheet funding in 2002 through the swap, which European Union regulators said they knew nothing about until recent days.
Failing to disclose the swap may have allowed Goldman, a co-lead manager on many of the sales, other underwriters and Greece to get a better price for the securities, said Bill Blain, co-head of fixed income at Matrix Corporate Capital LLP, a London-based broker and fund manager.
“The price of bonds should reflect the reality of Greece’s finances,” Blain said. “If a bank was selling them to investors on the basis of publicly available information, and they were aware that information was incorrect, then investors have been fooled.”
Michael DuVally, a spokesman at Goldman Sachs in New York, declined to comment.
HOW can Goldman Sachs still be operating??? How much more in your face corruption and fraud can citizens and countries TAKE?
Don't forget the monthly SS revenue will decline as people meet the maximum. That would make it hard to use this field as a predictor.
Thu, Feb 11 2010, 11:09 GMT
by Economic and Strategy Team
National Bank of Canada
The second big winter storm in less than a week comes at an inopportune time for the upcoming employment report. Indeed, the blizzard is occurring during the reference period used for the calculation of the February jobs report (i.e., the pay period that includes the 12th of the month). Back in January 1996 when a blizzard shut down the eastern coast of the United States, roughly 200,000 jobs were reported lost on the month, a significant deviation from the average monthly creation of 150,000 observed prior to the blizzard. As today’s Hot Chart shows, jobs rebounded very strongly in February (by over 600,000) before reverting to trend the following month. Given the timing of the current storm, we think that the current snowfalls will bury payrolls in February. As shown, we believe that a loss of around 200,000 is probably the most likely scenario at this juncture. The first sizeable increase in headcounts will most likely occur in March 2010.
http://mediaserver.fxstreet.com/Reports/e154f1ab-84c6-4d85-808a-00d87044ac6d/099f1787-35b6-4c6c-a314-909409bbb219.pdf
Very interesting and very much worth following. Unfortunately I don't possess the analytics to deal with it nor the computing power.
I think you can look forward but there are many variables that could impact the numbers and what those figures are we may not know. The first part of this would be to generate a collectively exhaustive list of what would impact the numbers and deal with that variable in some capacity.
I agree with both you and Biderman, the numbers don't add up. There has to be a way to use the data that is available to project what NFP should be, problem is that no one trusts the BLS data and although you may be spot on the BLS will report BS as is typical.
I anticipate much good will come from this type of exercise although I would guess that many have / are working on something like this right now!
First, there are only about 130 million workers currently paying into the SS system.
Second, the stimulus provides for a small reductio in employee paid fica through the end of 2010-an amount of about $25 a month or so. I don't get the fica rollback or rebate or whatever you prefer to call it, because like about one quarter of workers I make too much.
You are wrong about the number paying into SS. The 130mm number you cite is the BLS Non Farm number. SS covers much more than that. From the 2009 SS Trustee report:
During 2008, an estimated 162 million people had earnings covered by Social Security and paid payroll taxes; for Medicare the corresponding figure was 166 million.
These numbers have come down. But not to 130mm. If that were to be the case you would have 30mm people unemployed. Not the case.
Perhaps if you weighted jobs by income, the unemployment rate would not be so high. The little guy is the one fired more often. He earns less, and consequently, a small drop in payroll tax receipts might foretell a large rise in unemployment. Similarly, the medicare tax is paid into infinity, so it is weighted even higher toward the high earners who probably still have their jobs.
If you could break up the payroll tax collections by wage level or hourly rate, and throw out the high earners, I think you would have a better shot.
"For SS to be seeing a decline of 8% in revenues in February is disturbing. Some additional months of data in 2010 are necessary to prove this trend. But should this decline hold, it is a very poor sign for the Fund."
The SS fund is, of course, toast, and with it our US economy. I understand that you're trying to make short term predictions in order to gain a supernormal ROI, but it's unfortunate that the financial media (hello, CNBC) and market manipulators (hello plunge protection team) allow that trader thinking to dominate the conversation. Because we are at the very crossroads of deciding whether we want a prolonged period of cold gray depression, or a single moment of white hot anarchy. If we choose the former there is a possibility of coming out the other side intact. If we choose the latter, we get to experience the event horizon of a black hole.
I believe this is the exact information parsed by TrimTabs, and they have been making the case for quite some time that the BLS numbers are drastically underrepresenting the true extent of the employment sink hole.
I think they look at the T.recepts. See my comment below. This is an advance read for the month. You get this number before the month is over. TTabs is looking backwards and saying, "It doesn't add up".
I agree, It doesn't.
But my question is: Can you look forward, not backward? I want to know tomorrow's headline today.
Dailyjobcuts.com gives store closings and layoffs to come or now. Hope it helps. But the trend appears to be further contraction.
Cities and shools have been anouncing layoffs recent days.
Disclaimer, jobless, not looking,no income ,fixen to pan for Gold.
Rassmussen polling is not only echoing this it is also revealing that the zombies and dumb money are gaining consciousness.
Markets like gridlock. The tsunami is not something man has the capacity to impact. Once the earthquake strikes there is no stopping the release of the energy. No stopping this one.
Be like Wayne....be where the puck will be.
Benford's Law....
thank you for the intriguing reference.
Why look at a magical forecast, the Treasury reports the daily withholdings
http://fms.treas.gov/dts/index.html
But tax withholding has crashed faster than payroll what with the stimulus withholding tables and underemployment fewer hours for those still with jobs.
What is real interesting to me is that Medicare taxes only fell a little over 3 percent in fiscal year 2008/2009 while jobs fell over 8 percent.
What is that? Fewer jobs, but average wages are up?
For another clue, check the monthly Treasury report on Federal unemployment tax receipts.
The estimate comes out in the second week. So it is an advance read of the daily monthly total. But the actual data would be a good back check on the estimates.
The Fund organizes info in a way that you can massage their data base. Monthly, quarterly, annually,fiscal, calendar,by FICA/SECA. Also the prior month/quarter carryforward is part of the riddle.
There are a number of odd things in the numbers. The SECA (small business) number is up yoy up. (38 vs 39b) This is a rounding error. But it still surprises me it is up not down. Guys losing their day jobs go into some biz out of the house?
How about this? The 08 total receipts=675b, for 09=670. That is less than a 1% drop. Go figure? It is in the lags someplace. I noticed the change picking up speed in October. Subject to a revision of the Feb estimate it means that one sector of the system is slowing very fast right now. Actual wages being payed, severance or not would seem to falling quite quickly.
how does medicare compare to FICA? FICA drops off at 100k wages, where as Medicare taxes all wages...more rich wage earners compared to FICA earners?
Both medicare and FICA should trend together for early months of each year -- but when high-wage individuals hit the FICA limit, the correlation stops. So if the FICA limit were 100K -- just to pick a number -- a 200K/yr worker would show point-by-point correlation only through June.
This is very interesting stuff and I will keep an eye on it.
Funny, though, I cannot help but think if we have at least 101 people employed in the US---50 on Wall Street trading desks, 50 as lobbyists in Washington, and one incompetent at BLS---the market will continue higher.