Thursday, January 8, 2015

"It's the Claims Data, Stupid" - why UER is about to plunge towards 5%

Claims data closes the year at astounding lows which can only reflect a white hot economy. This is not news as first "showed" when successively less and less workers were furloughed during the July auto shutdown for refits, starting in 2012.  This is apparent from a simple year over year of the non-seasonally adjusted data.

 Most get their ideas of claims data taking uncritically the seasonally adjusted data which shows a rather sedate changes that do not portray the jaw dropping drama that is occurring.  I don't think it is because of anything nefarious, it is that the worst of the crisis of 2009 and then a second wave in 2010 hit the seasonal surge from Christmas to the first few weeks of January.  CES/BLS is well aware of this problem and it seems to promote debate internally on the seasonal adjustment machine X-13 ARIMA-SEATS - which is just a fancy smoothing machine seeking auto-regression. This is typical of the tons of solid, bit dull, papers on the problems of the recession.  The bottom line is the memory in the 13A model is 5 years, and the lions share of the crisis hit in December and January 2009 so we are at the end of the out of the norm impact on seasonals.  I do not understand how very seasoned economists blithely carried on reporting claims uncritically accepting this major blemish.  To my read of the data it is impossible to make any sense of claims data unless you look at the nonseasonal data year over year.  The greatest insight is from observing the four key seasonal points.  The most important is the turn and the hit in the first week or two of January.  Above you can see how massive the crisis was with the spike towards almost 1 million claims in a one week hit.  Assuming 500,000 was the "norm", the hit beyond 500,000 was incorrectly being applied to data as a seasonal hit, so the seasonal data then was adjusted too low and did not reflect the severity of the crisis.  This was then carried on through the year and the seasonal applied to periods where employment was improving was reported as not all that good a time.  The error in seasonals was also applied to the next key period, the July auto refit, not reflecting how severe the hit was.  In July 2009 autos were not shutting down just for refit, they were closing down permanently, as the crisis of the July 2009 surge to excess of 600,000 shows.  In time the "miracle" Steve Rattner pulled off in almost single handed saving of the USA auto industry is obvious in the data given how swiftly the seasonal auto refit in July became just that - seasonal.  But the non seasonal data did not provide insight into how dire the environment was - there is no doubt given the non seasonal claims data that the nation was on the verge or had entered a crisis every bit as severe as the Great Depression.  To my read it is as if bathers at  a Miami South Shore beach for a year were frolicking not knowing hundreds of great white sharks were just past the surf line.  The country was a hairs breadth from ruin.  Which, if it had occurred would  have very likely led to world war as all those chronic imbalances from mercantilism currency or oil cartel pricing were settled in only a few months and given a context of a lot of pissed off Americans.

A view of initial claims that makes it easier - at least to me - for several intuitive "aha!" moments is to take the year over year and make them into a heat map. Years from 1992 are the x axis and weeks are the y axis.  Color scale is given on the right and has rich green as "good times" with low claims and white as "tough times".  Keep in mind this is not adjusted by population growth and is, again, non seasonally adjusted levels week by week and year by year.  By moving from left to right one can see an entire decade at a certain week go by, and if one reads up and down one can see the progression in any one year,

Week 28, reading across shows the move to light green to yellow of the rise in the claims from the July auto refit.  The year end turn surge in claims is obvious where the lone pure white spot showing on week 2 in 2009.  And of course the read from top to down on the 2009 column shows  how extraordinary the crisis was and when remedy started to become effective - the most significant was the Rattner auto rescue that was in place by the 2010 July period.  Further more, once the quick death of a severe solvency crisis was alleviated by the Fed via QE I - there is little sign QE ongoing, the majority of QE I and then QE II and QE III had any impact on employment.  Of course the immediate shock alleviation with QE I was critical, but thereafter - once solvency was restored to the system - there is no impact.  One can also see how a "normal" recession of 2007 on become a severe crisis as solvency became the problem starting with  Lehman and then during the surge in LIBOR to Treasury Bills in November and the "jump ball" from the election. 

Most seem to think that they can glibly dismiss claims data, either because they uncritically accept the seasonals, or as the obvious massive discrepancy - the "separate worlds" of claims data to the monthly reported employment data and the side cars that supposedly qualify unemployment like employment to population (EP) ratio or (does not seem chic now) the labor force participation rate.  I think there are several unique features to this crisis versus the only comparable comparison, the monetary tightening induced Reagan recession of trough of 1983 to 1986.  First the Reagan 1983 recession hit in the QI 1983 and was somewhat controlled as it came not by solvency cusp like hit, but was induced by Volcker doubling down in tightening Fed policy.  1983 was induced from monetary policy, not inexplicable exogenous shocks of the 2008 - 2009 crisis.  The hit of 2009 came screaming out of the blue, the recession of 1983 was induced.

This meant that seasonals were not a problem nor made a "looking through a glass darkly" problem in 1983.  The relationship between monthly sample data and the census weekly data was linear and reported honestly the volatility and swings that occurred.  This is not the case in 2009 as the strike was violent and as massive as 1983, the control panel was giving out such "noise" or such a dire status that I dont think the severity was believed, or appreciated, as it was in 1983.  This ended with the monthly data being smoothed and swiftly became out of synch with the weekly claims data.  Claims data relationship to monthly employment data peaked for both at the same time in the 1983 recession.  This is not the case in 2009 where claims data peaked in March 2009 and the monthly data slowly smoothed in and "bled" in how dire the situation was not peaking until June 2010.  At first I think this was an honest error in the fog of war of the crisis, that surely it just was not that bad, but then later I think the White House started to lean on BLS, and with perhaps complicit BLS economists deliberately misreported the monthly UER.  I think that UER peaked in March 2009 towards 11 1/2% and then both claims and monthly data paused, but in a coordinated fashion and then after the Fed assured liquidity and after Rattner's industrial rescue of America, the improvement and mending was swift and very large.  Instead the BLS was now use to smoothing and while it is hard to put the finger on when exactly occurred, BLS threw their lot in with the "Odyessian Fed" and from then on the monthly employment is contrived and is adjusted down only to the point it has  some connection to the accurate, impossible to manipulate, claims data.  Another point should be made in that the disconnect that occurred by March 2009 between claims data and monthly data was so large and so obvious with hindsight, and then maintained, that it clearly dismisses those who feel claims data is irrelevant as it is some sort of different world of  large corporate layoffs.  Clearly the data makes mince meat of that view.  Claims data and monthly data are one and the same as far as reflecting the economy in general.

Where does that leave us?  There is so few comparables to the 2009 crisis, not even the 1983 recession the only recession as "large" as the 2009  downturn, but is not of the same nature of 2009 being a monetary induced recession versus the 2009 solvency crisis.

There can be no debate that claims data is an accurate depiction of the current status of the USA economy.  The reasons that the monthly data is being presented  in such a doctored and strangely negative basis are over.  Perhaps part of the reason the Democrats took such a drubbing in the election and lost now both the House and Senate to the GOP is that the deception the BLS was maintaining had unintended consequences.  While it did cover up the manipulation of the coverup on how bad the situation was in 2009, and while it did become an important supportive factor for the Fed maintaining "forward guidance" and avoid returning to a data driven rules based policy  and thereby maintain the new Fed extraordinary powers, it also "worked" and convinced the American public that the Democrats were doing a terrible job in managing the crisis.  That is rich irony as the claims data - and also almost all other economic data (but for diffusion data of "pop pols" of ISM guys) - indicate the Democrats did a pretty good job and should have been doing a victory lap into the November election.

And that might be the important point.  Unless the Democrats wish to be eviscerated in the 2016 elections, and really those elections are starting, now is the time accurate and positive (if possible) reporting of employment data occurs.  Tomorrow may be the second month of this process, I suspect it started with the surge in payrolls last month, and we may see a 500,000 or so payroll number along with a large drop in UER to under 5 1/2%.  But perhaps the Fed is still leaning against this reporting and it will not happen.  But what is clear is that within 500,000 or so the monthly employment data will be whatever BLS, along with the clique they travel with or want to travel with, want it to be.  But I do think they are quickly losing freedom to do this and the day of reckoning is upon them.  The administration  will start to pressure the Fed, and they will no longer go along with the Fed with their dour "unemployment is really huge, huge I say!" and secret sauce indicators - that while that allowed the Fed to stay loose while Obama went on a 6 year campaign to centralize Democratic power with patronage and realpolitik austerity,  This strange partnership is no longer required and is now harmful to the administration, so Obama is likely or has already given notice that he is no longer supportive no more will abide by the Fed keeping this "arm waving" emergency status.

Claims data is the current status of the USA,  we are through NAIRU,  and we are now white hot - will we are "verdant" green.