JOHN WILLIAMS’ SHADOW GOVERNMENT STATISTICS

COMMENTARY NUMBER 321
August Employment and Unemployment

September 3, 2010

__________

August Unemployment: U.3 = 9.6%, U.6 = 16.7%, SGS = 22.0%

August Payrolls Fall 54,000, Gain 60,000 Ex-Census Workers

Better-Than-Expected Payroll Changes Were Not Statistically Meaningful

__________

PLEASE NOTE: The next regular Commentary is scheduled for Thursday, September 9th, following release of the July trade data. A Special Commentary, reviewing and updating the broad outlook on the U.S. economy, systemic stability and the financial markets, will be published separately in the week ahead.

– Best wishes to all for a most enjoyable Labor Day weekend, John Williams

 

More-Positive Payroll Indications, But Not Statistically Significant. Contrary to my forecasts, the August labor data met or exceeded market expectations. Although the government’s broadest unemployment rate rose sharply for the month, the notch higher in the headline unemployment rate came in at the consensus level. The payroll employment data generally were stronger than expected, and included upside revisions to prior-period reporting. As cautioned in yesterday’s comments (Commentary No. 320), however, none of the headline numbers were statistically meaningful. 

Sharp deterioration in the U.S. labor market conditions should be seen in next month’s reporting. An early indication of such was reported today — subsequent to the Bureau of Labor Statistics (BLS) reporting — where the purchasing managers nonmanufacturing (services) survey for August showed a plunge in its employment component. A diffusion index, a reading above 50.0 indicates expansion. The employment index fell to 48.2 (indicating outright jobs contraction) in August, from 50.9 in July. The August index leads the September payroll reporting.

The broad outlook is unchanged, and a review and update of the specific economic, systemic-stability and financial-market outlooks will be published next week in a Special Commentary.

Separately, unusual variations in and revisions to seasonal factors, an unsustainable pattern of increases in the monthly birth-death model upside biases and some masking of the deteriorating broader unemployment conditions were at work in generating today’s employment and unemployment results. 

There are two types of political manipulation of data. The first and more infrequent case is that of direct intervention (externally or internally) to produce a desired number. The second and more common case is the use of gimmicked methodological changes over-time that tend to build-in regular positive biases to economic data and negative biases into inflation numbers. The handling of discouraged workers (no longer tracking long-term discouraged workers) and the handling of the birth-death model are examples of the second case. The birth-death biases are discussed in the Birth-Death/Bias Factor Adjustment section below.   

As to direct manipulation, seasonal factor gimmicks and simply the use of rigged numbers have been utilized at different times to alter reported data for political purposes. Rigged results usually have been tied to pending elections and/or financial market needs. There is no direct evidence of this type of manipulation going on at present, although there was some unusual seasonal-factor activity in the latest jobs report.

The Wonders of Seasonal Adjustment. Consider that the not-seasonally-adjusted data last month showed June payrolls at 131,345,000. With updated information, the unadjusted June payrolls gained 1,000, revising to 131,346,000 in August’s reporting. The seasonally-adjusted June number, however, revised from 130,373,000 to 130,419,000, a gain of 46,000, with 45,000 being created out of thin air by the monthly revision to seasonal factors. There was a time when seasonal factors were set for a whole year in advance. 

The ability to play monthly games with seasonals, the nature of assumptions in the handling of hard data and revisions to same, and a 95% confidence interval of +/- 129,000 jobs around the reported payroll number change, provide significant reporting leeway should someone choose to target payroll reporting in the context say of consensus expectations tied to the financial markets, or of related media hype that could impact public political perceptions.         

Long-Term Unemployed Rolling into Discouraged Category, versus Labor Force Participation. The Bureau of Labor Statistics (BLS) noted for August that "the number of long-term unemployed (those jobless for 27 weeks and over) declined by 323,000 over the month…" That number was seasonally-adjusted, a practice that seems questionable given the duration of the measure. Unadjusted, the ranks of the long-term unemployed were reported down by 268,000. Most likely, the bulk of those individuals rolled into the government’s tallying of the short-term discouraged worker category (used in unemployment rates U.4, U.5 and U.6), which was reported to have declined, too, by 75,000 in the month (unadjusted).   Balancing out a not-so-happy picture would be a portion of the short-term discouraged workers also rolling into the long-term discouraged worker category (included in the SGS-Alternate Unemployment measure), which is not tracked by the BLS. 

I have been working with the historical data in terms of the employed and labor force numbers versus the working-age population. Indeed, as has been suggested by others, the declining ratios of employed to population, and labor force participation rates, appear to reflect significant movement of individuals into the discouraged-worker categories (discouraged workers are not counted as part of the official labor force). If one were to assume that all the ratio changes were due to this factor (as opposed to changes in retirement patterns, etc.), holding the participation rate constant across the population would generate an implied broad unemployment of about 22%, where the SGS-Alternate Unemployment measure stands at present. A separate article on this will be published shortly.

Payroll Survey. The BLS reported today (September 3rd) a statistically-insignificant, seasonally-adjusted August 2010 jobs loss of 54,000 (a gain of 69,000 before prior-period revisions) +/- 129,000 (95% confidence interval). Net of the included layoffs of 114,000 intermittent and temporary census hires, August payrolls gained a statistically-insignificant 60,000 jobs. The reported gain in private-sector payrolls was 67,000 for the month, which more than covered a 7,000 decline in other government jobs that brought the total ex-census gain to the 60,000 figure. 

Against August’s total jobs loss of 54,000, a jobs gain net of census impact of 60,000, and a jobs gain of 67,000 in the private sector, July showed a revised monthly total jobs loss of 54,000 (previously a loss of 131,000), a jobs gain net of census impact of 89,000 (previously a gain of 12,000), and a jobs gain in the private sector of 107,000 (previously a 71,000 gain). 

As of the August payroll survey, 84,000 temporary census workers still faced lay-offs, with the bulk of the remaining impact on payroll reporting likely in September reporting.

From peak-to-trough (the peak month was December 2007; December 2009 is current, but likely short-lived, trough of the current cycle), payroll employment declined by a seasonally-adjusted 8,363,000 jobs, or 6.1%. As of August 2010 reporting, net of temporary census hiring, payrolls have gained 0.5% or 639,000 jobs since the December 2009 trough.

In terms of annual change, year-to-year growth in August nonfarm payrolls was 0.21% (0.15% net of surviving census hires), up from July’s revised 0.08% gain (previously a 0.01% contraction), and a revised 0.07% (previously 0.16%) annual contraction for July net of census effects.

Thanks to recent, protracted bottom-bouncing in the payroll series, current annual growth has recovered from the post-World War II record 4.96% decline in July 2009. The July 2009 decline was the most severe annual contraction seen since the production shutdown at the end of World War II, which reflected a trough of a 7.59% annual contraction in September 1945. Disallowing the post-war shutdown as a normal business cycle, the current annual decline remains the worst since the Great Depression. 

The long-term graph of year-to-year payroll change reflects the numbers as reported, with no adjustment for census hiring variations. The second graph, however, shows the year-to-year detail both with and without the census hires.

 

Birth-Death/Bias Factor Adjustment.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 five years), such information is estimated by the Birth-Death Model. Unusual birth-death activity continued in August 2010, with the monthly birth-death adjustment adding 17,000 more jobs than it did in August 2009, a pattern that continues contrary to what would be suggested by 2009 BLS reporting difficulties that understated declines in payroll employment, and by recent indications of a another downside benchmark revision for 2010 data, as discussed in yesterday’s Commentary.

Based now on the "assumption" of economic recovery, the bias factors used in the last five months of payroll reporting have been 25.9% ahead of what they were reset to after the most recent and disastrous benchmark revision. 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.

Historically, the Birth-Death Model biases have tended to overstate payroll employment levels — to understate employment declines — during recessions. These flaws were confirmed by the nature of the BLS’s massive downside benchmark revision to 2009 data published with the January 2010 jobs report, where the BLS had indicated that underlying assumptions to the Birth-Death Model were missing significant jobs losses due to business failures that the BLS could not measure. 

Although the upside bias had been scaled down some as a result of 2009’s reporting errors, the Birth-Death Model survives and remains a major distorting factor in monthly payroll reporting. The unsupportable premise that jobs created by start-up companies in this downturn have more than offset jobs lost by companies going out of business, continues. 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, presumed 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 are running at about 50,000 per month (seasonally-adjusted), at present. I estimate this monthly bias should be negative by 200,000 or so, on average. Since it is not, the BLS continues regularly to overestimate monthly growth in payroll employment by roughly 250,000 jobs. Such misreporting, however, will not be corrected until the next benchmark revision is published in February 2011, well after the November elections. Again, as mentioned yesterday, that revision likely will be negative, and its magnitude, as of the benchmark month of March 2010, should be estimated along with next month’s payroll reporting.

That said, the unadjusted August 2010 bias was a monthly addition of 115,000 jobs, versus an addition of 98,000 jobs in August 2009, and against a monthly addition of 6,000 jobs in July 2010.

Household Survey.  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 seasonally-adjusted monthly employment gain of 290,000 in August, following a reported decline of 159,000 in July. 

As with the last several months’ data, the series appears to be experiencing distortions from inappropriate seasonal-adjustments for temporary census worker employment.   In the payroll survey, the census impact is quantified and properly is not seasonally adjusted. Since some portion of the part-time census workers, however, already would be counted as employed due to other part-time or full-time employment, accurately quantifying the impact of census employment variations on unemployment is not possible. As a result, no separate accounting is made, and the census-worker impact has been seasonally adjusted as though the census hiring and firing patterns are annual events.

Where I have expected that some portion of the census layoffs would have pushed the unemployment rate minimally higher in the last three months, such has not happened, despite the original hires having had noticeable impact on reducing the reported headline unemployment rate. Nonetheless, sharp upside movements in the unemployment rate remain likely in the months ahead as the seasonal pattern distortions eventually reverse out.

Also, as discussed earlier, increasing numbers of longer-term unemployed previously counted in U.3 unemployment appear to be rolling into the short-term discouraged worker category, with a portion of the short-term discouraged workers rolling into the long-term discouraged worker category.

Just 4,000 Shy of Rounding to 9.7% Unemployment. The August 2010 seasonally-adjusted headline (U.3) unemployment rate rose by a statistically-insignificant 0.13 percentage point to 9.64% +/- 0.23% (95% confidence interval), from 9.51% in July. The U.3 unemployment count was just 4,000 shy of rounding to 9.7%. With upside pressure on the unemployment rate in September, such makes an apparent 0.2 percentage point jump in next month’s rounded U.3 rate an easy increase. Not seasonally adjusted, August’s U.3 unemployment rate declined to 9.5% from 9.7% in July. 

August U.6 unemployment rose to a seasonally-adjusted 16.7% (fell to 16.4% unadjusted) versus 16.5% (16.8% unadjusted) in July.

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 my estimate of the excluded long-term discouraged workers back into the total unemployed, unemployment — more in line with common experience as estimated by the SGS-Alternate Unemployment Measure — rose in August to about 22.0% from 21.7% in July. Developing trends here were discussed in the opening comments. The SGS estimate generally is built on top of the official U.6 reporting and tends to follow its relative monthly movements. See the Alternate Data tab for a graph and more detail.

As discussed previously, while 22.0% unemployment 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 a Great Depression comparison, I would look at the estimated peak nonfarm unemployment rate in 1933 of 34% to 35%.

Updated Real M3 Signal for Intensifying Downturn. As discussed in frequent Commentaries this year(see Commentary No. 316 for example), declining year-to-year change in real (inflation-adjusted) M3 signals a pending economic downturn or pending intensification of an existing economic contraction, with contracting broad liquidity invariably constraining broad economic activity. The signal is generated when real M3 first turns negative year-to-year, which occurred in December 2009 in the current economic cycle. The signal is not generated by, nor dependent on, either the length or the depth of the M3 downturn. The downside shift in business activity usually follows within six to nine months, which encompasses the current period. 

The following updated graph plots annual real M3 growth versus periods of recession formally recognized by the NBER. It includes approximate annual real contraction in the SGS Ongoing-M3 Estimate as of August 2010. The August M3 estimate used here is an annual real contraction of roughly 5.6% versus a 6.6% contraction in July. Details will be discussed in next week’s Special Commentary, which will include a review of the broad liquidity outlook and the money supply circumstance. The formal preliminary estimate for the SGS Ongoing M3 Measure for August will be posted this weekend of September 4th on the Alternate Data page for M3.

At present, the pace of annual contraction in nominal August M3 is on track to hit 4.3%, versus 5.4% in July, reflecting a flattening/upturn in month-to-month change, with annual comparison against a pace of declining monthly growth in the year-ago period. 

 

 

Week Ahead. Given the unfolding reality of a weaker economy (or re-intensifying downturn) and more serious inflation problems than generally are expected by the financial markets, risks to reporting will tend towards higher-than-expected inflation and weaker-than-expected economic reporting in the months ahead. Increasingly, such is being seen in economic reporting net of prior-period revisions.

Trade Balance (July 2010). Due for release on Thursday, September 9th, reporting of the July trade deficit will help to set the tone for the first reporting of third-quarter GDP at the end of October. Even a monthly shortfall close to the same amount seen in June — as likely will be the consensus — would be an early signal favoring quarterly contraction. Ongoing deterioration in the patterns of both the nominal and real (unadjusted and adjusted for inflation) monthly trade deficits should become fairly regular and are a fair bet for the upcoming report.

 __________