Constant Recalculation of Payroll Seasonal Factors Is Boosting 2011 Payrolls.  The Bureau of Labor Statistics’ (BLS) practice of using “concurrent” seasonal-factor adjustments artificially shifted previously reported jobs into October 2011 and contributed to boosting the reported monthly gain versus September.  Net of that and an upswing of 50,000 jobs in the adjusted monthly Birth-Death Model add-factor for October, monthly payroll change would have been on the downside of “unchanged,” instead of plus 80,000, but that still would have had a 95% confidence interval around it of plus-or-minus 129,000 jobs.

Heavily distorted seasonal factors, and the plus-or-minus 0.23% confidence interval around the 0.07% easing in the unemployment rate, left the rounded 0.1% headline drop in unemployment as meaningless as the reported 80,000 jobs gain in October.

PAYROLL SURVEY DETAIL.  The BLS reported today (November 4th) a statistically-insignificant, seasonally-adjusted October 2011 month-to-month jobs gain of 80,000 (a gain of 182,000 jobs before prior-period revisions) +/- 129,000 (95% confidence interval).  September payrolls showed a revised 158,000 gain (previously a gain of 103,000), while August’s monthly gain revised to 104,000, versus last month’s estimate of a 57,000 gain.  Monthly changes for July 2011 and before also were revised, but those revisions were not reported by the BLS so as to avoid confusing payroll employment data users.

In terms of year-to-year change, the unadjusted October 2011 growth rate slowed minimally to 1.15% versus the revised 1.23% (previously 1.12%) reported for September, and against 1.11% (previously 1.09%) annual growth reported for August.  Although the graphs of long-term year-to-year unadjusted payroll change had shown a rising trend in annual growth, which primarily reflected the still-protracted bottom-bouncing in the payroll series, that pattern has flattened out in recent months, as shown in the first graph following of the near-term detail in year-to-year change.  These numbers still reflect some short-lived year-to-year distortions as a result of the year-ago hiring surge and full layoffs of temporary census workers.

As shown in the next, longer-term graph (historical detail back to World War II), with the bottom-bouncing of recent years, current annual growth has recovered from the post-World War II record 5.06% decline in August 2009, which was the most severe annual contraction seen since the production shutdown at the end of World War II (a trough of a 7.59% annual contraction in September 1945).  Disallowing the post-war shutdown as a normal business cycle, the August 2009 annual decline remains the worst since the Great Depression, yet the current level of employment is far from any recovery.

The regular graph of seasonally-adjusted payroll levels is shown in the Opening Comments and Executive Summary section.

Concurrent Seasonal Factor Distortions.  As discussed in prior writings (see the Hyperinflation Special Report (2011), for example), seasonal-factor estimation for most economic series has been distorted severely by the extreme depth and duration of the economic contraction.  These distortions are exacerbated for payroll employment data based on the BLS’s monthly seasonal-factor re-estimations and lack of full reporting.

Where the BLS recalculates the monthly seasonal factors each month for payroll employment, going back a number of years, it only publishes revised data for the last two months of reporting (August and September 2011 with the October 2011 report).  Shown in the following graph, the latest “concurrent” seasonal factor changes upped October 2010 relative to September 2010, with implied greater adjusted gain for October 2011. 

Clearly seen in the most recent plot (heavy blue line) is a shifting of reported seasonally-adjusted jobs from 2010 into 2011.  With just two months of prior reporting shown as revised in the official BLS release, pre-August 2011 revisions were not published so as to avoid “confusing” people using the data.  

As discussed repeatedly in recent employment Commentaries, meaningful seasonal-adjustments tend to be stable over time, without wild fluctuations every time the seasonals are re-estimated.  This is true particularly for series like payroll employment and retail sales, where the seasonal factors are concurrent—recalculated each month for the current month’s raw data.  If the payroll seasonals were stable, the lines in the graph would be flat and coincident.  Instead, the variations intensify with each successive month.  The monthly recalculations of seasonally-adjusted payroll levels show irregular revisions, with monthly swings now of plus or minus 90,000 jobs shifting over time.  To the extent the numbers affect current reporting, the differences are enough potentially to alter financial-market perceptions and reactions. 

A further big issue remains that the month-to-month seasonally-adjusted payroll data have become increasingly worthless, with reporting errors likely now well beyond the official 95% confidence interval of +/- 129,000 jobs in the reported monthly payroll change.  Yet the media and the markets tout the data as meaningful, usually without question or qualification.

The inconsistency differences in the graph were calculated based on the raw unadjusted data and the seasonal-adjustment program available to the public on the BLS Web site.  Using the BLS data, we have calculated the seasonally-adjusted numbers as the BLS should be showing them, as of the current reporting, and the differences between official reporting and the consistent seasonally-adjusted series.

Payroll Benchmark Revision and Gimmicked Upside Biases.  Where, traditionally, the BLS does not adjust payroll reporting for the annual benchmark revisions before the January release of the following year, the Birth-Death Model appears to have been adjusted for the quarter beginning October 2011, so as to add an additional 50,000 jobs per month (600,000) per year to compensate for the pending 192,000 annual benchmark revision (see next section).  On September 29th, the BLS published a preliminary upside benchmark estimate for March 2011 unadjusted nonfarm payrolls.  The announced192,000 positive revision suggested there had been a monthly-average understatement of 16,000 jobs in the period from April 2010 to March 2011.  Official reporting history will not be revised until the February 2012 release of the January 2012 payrolls.

Birth-Death/Bias Factor Adjustment.  Despite the ongoing and regular overstatement of monthly payroll employment—as evidenced usually by regular and massive, annual downward benchmark revisions—the BLS nonetheless generally upped its monthly biases in post-benchmark reportings of recent years.  In the wake of the purported upside benchmark preliminary revision for March 2011 (see previous section), the monthly bias factor used in October 2011—the first month of fourth-quarter reporting—was revamped to a more-positive monthly add-factor of 102,000 than the 71,000 used in October 2010.  That followed a more-negative subtraction of of 43,000 jobs for September 2011, versus a 25,000 subtraction in September 2010.  The resulting monthly swing of 50,000 jobs—an upswing of 600,000 annual jobs—substantially over-compensates for the purported 192,000 upside benchmark revision.

The aggregated upside annual bias reflects an ongoing assumption of a net positive jobs creation by new companies versus those going out business.  Such becomes a self-fulfilling system, as the upside biases boost reporting for financial-market and political needs, with relatively good headline data, while often also setting downside benchmark revisions for the next year, which traditionally are ignored by the media and the politicians.  Where the BLS cannot measure the impact of jobs loss and jobs creation from employers starting up or going out of business, on a timely basis (within at least five years, if ever), such information is estimated by the addition of a bias-factor generated by the Birth-Death Model (a model of the effects of new business creation and old business bankruptcies).  The fundamental defects of the Birth-Death Model are discussed as usual in the ensuing paragraphs.

Positive assumptions—commonly built into government statistical reporting and modeling—can become self-fulfilling prophesies, with “stronger” economic data being reported as a result of happy guesstimates, or underlying assumptions of ongoing economic recovery.  Indeed, historically, the Birth-Death Model biases have tended to overstate payroll employment levels—to understate employment declines—during recessions.   There is a faulty underlying premise here that jobs created by start-up companies in this downturn have more than offset jobs lost by companies going out of business.  So, if a company fails to report its payrolls because it has gone out of business, the BLS assumes it still has its previously-reported employees and adjusts those numbers for the trend in the company’s industry. 

Further, the presumed net additional “surplus” jobs created by start-up firms, get added on to the payroll estimates each month as a special add-factor.  These add-factors have averaged 40,000 jobs per month over the last 12 months and appear now to have been upped to about 90,000 jobs per month.  With the economy continuing to falter, I expect a significant downside benchmark revision for next year (March 2012), given current details of the BLS’s happy estimates.

HOUSEHOLD SURVEY DETAILS.  The usually statistically-sounder household survey, which counts the number of people with jobs, as opposed to the payroll survey that counts the number of jobs (counting multiple job holders more than once), showed a October 2011 employment gain of 277,000 versus a 398,000 gain in September.  Severe issues with monthly seasonal factors still meaningfully cloud the significance of the reported monthly levels in the adjusted headline U.3 unemployment rate and other adjusted household-survey numbers.  Again, adjusted data have been shifted by highly unstable seasonal factors that are artifacts of the severe and extraordinarily protracted downturn in U.S. economic activity (as well as distortions created by last year’s census hiring and firing effects), not by the regular and stable seasonal patterns that were in place before the current economic crisis.

The reported October 2011 seasonally-adjusted headline (U.3) unemployment rate was little changed, with a statistically-insignificant 0.07 percentage point decline to 9.01% +/- 0.23% (95% confidence interval), from 9.08% in September.  Not-seasonally-adjusted, October’s U.3 unemployment rate was 8.5%, down from 8.8% in September.

Also heavily skewed by the seasonals, the October U.6 unemployment rate fell to a seasonally-adjusted 16.2% from 16.5% in September.  The unadjusted U.6 rate declined to 15.3% in October from 15.7% in September.  The broadest unemployment rate published by the BLS, U.6 includes accounting for those marginally attached to the labor force (including short-term discouraged workers) and those who are employed part-time for economic reasons (they cannot find a full-time job).

In 1994, during the Clinton Administration, “discouraged workers”—those who had given up looking for a job because there were no jobs to be had—were redefined so as to be counted only if they had been “discouraged” for less than a year.  This time qualification defined away the long-term discouraged workers.  The remaining short-term discouraged workers (less than one year) are included in U.6. 

Adding the SGS estimate of excluded long-term discouraged workers back into the total unemployed and labor force, unemployment—more in line with common experience as estimated by the SGS-Alternate Unemployment Measure—eased to 22.9% in October from 23.1% in September.  The SGS estimate generally is built on top of the official U.6 reporting, and tends to follow its relative monthly movements. Accordingly, it will suffer some of the current seasonal-adjustment woes afflicting the base series. 

Nonetheless, there has been a noticeable divergence in the SGS series versus U.6.  The reason for this is that U.6, again, only includes discouraged workers who have been discouraged for less than a year.  As the discouraged worker status ages, those that go beyond one year fall off the government counting, as new workers enter “discouraged” status.  Accordingly, with the continual rollover, the discouraged workers counted in U.6 continue are not changing much in aggregate, at present, but the long-term discouraged worker component in the SGS estimate continues to increase, as few of those that have dropped out of U.6 are gaining active employment.  See the Alternate Data tab for more detail.

As discussed in previous writings, an unemployment rate near 23% might raise questions in terms of a comparison with the purported peak unemployment in the Great Depression (1933) of 25%.  The SGS level likely is about as bad as the peak unemployment seen in the 1973 to 1975 recession.  The Great Depression unemployment rate was estimated well after the fact, with 27% of those employed working on farms.  Today, less that 2% work on farms.  Accordingly, for purposes of Great Depression comparison, I would look at the estimated peak nonfarm unemployment rate in 1933 of 34% to 35%.