JOHN WILLIAMS’ SHADOW GOVERNMENT STATISTICS

FLASH UPDATE

July 2, 2009

__________

June Jobs Loss Was 513,000 Net of Concurrent Seasonal Factor Bias
Likely Topped 700,000 with Birth-Death Machinations

Payroll Employment Growth Overstatement Could Top 2.5 Million
Per Year with Birth-Death Modeling

Annual Payroll Decline Deepened to 4.2%
Equal to 1958 Trough and Near 1949 Trough

SGS-Alternate Unemployment at 20.6%

__________

PLEASE NOTE: The next planned Flash Update is for Friday, July 10th, after the release of the May trade report. Any interim Flash Update or Alert would be published as dictated by developing economic or financial-market circumstances.

Happy 4th of July!  – John Williams

June Employment Reporting Showed Ongoing Economic Deterioration. The Bureau of Labor Statistics (BLS) released ongoing indications of deteriorating U.S. employment/unemployment conditions in June, with a worse-than-expected 467,000 drop in June payrolls, but a narrower-than-expected rise in unemployment to 9.5%. Net of the Concurrent Seasonal Factor Bias (discussed below) and net of distortions built into the reporting by the Birth-Death Model (discussed below), the June jobs loss likely exceeded 700,000.

The severe recession continues to deepen. My broad outlook has not changed; the worst of the financial and economic crises remain ahead of us. Before getting into the detail of the June report, a variety of special considerations are detailed, as directly related to the reporting of employment conditions as well as to broader implications for economic reporting in general.

Birth-Death Model Gives Faulty Upside Boost to Payroll Reporting During Recessions. The Birth-Death Model used by the BLS to adjust for employment changes due to jobs gained or lost by new business creations and failures is structured on a non-recession environment. As a result, it spuriously adds upside monthly biases into the payroll employment reports during economic downturns, enough, perhaps to spike current reported annual payroll growth by more than 2.5 million jobs. This system was introduced in the early 2000s to upgrade the "bias adjustment" system created in the mid-1980s. 

In the wake of underestimated payroll growth coming out of the 1980 to 1982 double-dip recession, the BLS introduced upside bias adjustments to its monthly payroll employment reporting. Blaming its inability to catch jobs created in the formation of new companies, the BLS began adding 120,000 to 160,000 jobs to each month’s seasonally-adjusted report, with the monthly bias level recast every quarter or so. Bad estimates made of payroll growth, in theory, would be corrected in the next year’s benchmark revisions.

The system was not designed to accommodate recessions, but the benchmark revisions tended to show a pattern of fairly consistent overstatement with the annual revisions, regardless of the business cycle. During the reporting cycle covering the 1990 to 1991 recession, a particularly large downward benchmark revision in previously reported payrolls levels was blamed partially on the BLS assuming that companies that had stopped reporting during the recession still were in business, with proportionate payroll employment attributed to them by the BLS. The problem was that much of the non-reporting reflected companies going out of business.

Amidst growing public criticism, the BLS moved to justify the bias factors with the introduction of a "Birth-Death Model." The BLS continued to assign proportionate employment and trends consistent within industry groups to non-reporting companies. It justified not counting jobs losses from business deaths, assuming that they were offset by gains from business creations. The net difference between business births and deaths and related employment changes were estimated based on an average sampling of the prior five years of data. 

The bulk of that modeling was based on periods of economic growth. In recessions, however, more jobs tend to be lost with failed companies than are gained with generally smaller start-up firms. As a result, the average net birth-death adjustment should be a contraction, during a recession, not a positive monthly bias factor. At present, the seasonally-adjusted payroll change starts with an average 75,000-plus upside bias — more than 900,000 per year — before any of the regular monthly sampling of payroll employment is added in. In the present economic downturn, a monthly net loss (with a requisite required net negative bias) of 75,000 jobs from the birth-death impact per month likely would be more appropriate. Such an estimate could be quite conservative, with the 1,800,000 annual swing suggested by the reversal of a 75,000 upside bias to a downside bias, most likely topping 2.5 million. No one can tell for sure, since the numbers available from the BLS just are not that meaningful. Future benchmark revisions eventually should show a more realistic picture.        

Although the published monthly biases are added to the unadjusted numbers, seasonal adjustments have been changed over time to move the monthly impact towards the average seasonally-adjusted 76,000 bias. Had this not happened, monthly changes would be more clearly skewed by the biases, as was evident during the early days of the Birth-Death Model’s use.

Impact of Protracted Economic Contraction on Employment Reporting. As the severe economic contraction rolls into multiple years, various areas of reporting have been or will be affected that could be misinterpreted as signs of economic improvement or that might not be showing the full extent of deteriorating conditions.

Discouraged Workers. Discouraged workers are those who meet all the qualifications for being unemployed, except they have not looked for work in the last four weeks, because there are no jobs to be had where they live. In 1994, this definition was amended to limit discouraged workers to only those who had looked for work in the last year (as opposed to no duration limitation in prior reporting). Accordingly, anyone who has been discouraged now for more than a year — since June 2008 or before — has disappeared from the rolls of the government’s "alternative measures of labor underutilization." The SGS-Alternate Unemployment measure adds in an estimate for those no longer included in the government’s broader reporting.

Ongoing Claims for Unemployment Insurance. Benefits eventually expire, and while the numbers may gyrate with changing government programs, a decline in ongoing unemployment benefits is not an economic positive in the current circumstance.

Furloughed Workers. As seen recently for many State of California employees, for example, an increasingly common circumstance is for employees to be forced to take off days without pay. Such does not impact payroll employment reporting, but it is measured in the broader unemployment measure, U.6, in the category of "total employed part time for economic reasons."    

Year-to-Year Change. As year-to-year numbers shift to assessment against collapsing economic numbers a year ago, year-to-year percent changes will start to show a pattern of plateauing at extremely negative levels. Such is not a sign of pending economic rebound so much as it is an artifact of protracted economic contraction.

Seasonal Adjustments. The extreme economic disruption has altered a variety of traditional seasonal patterns, with a resulting cautionary note that seasonally-adjusted data may be skewed unusually. Year-to-year comparisons help to mitigate such distortions.

June Reporting Showed Minimal Revisions. This morning’s (July 2nd) report showed minimal revisions, but it may have been subject to unusual seasonal-factor distortions, given the impact of the severe recession, the normal sharp swings in employment tied to the end of the school year, and the BLS’s inability to adequately adjust for same.

Payroll Survey. The BLS reported a statistically-significant, seasonally-adjusted jobs loss of 467,000 (down 459,000 net of revisions) +/- 129,000 (95% confidence interval) for June 2009, following a revised 322,000 (previously 345,000) jobs loss in May.

From peak-to-current (the peak month was December 2007; the current month of June also is the short-lived trough of the current cycle), payroll employment has declined by a seasonally-adjusted 6,460,000 jobs, or 4.7%. Year-to-year contraction (unadjusted) in total nonfarm payrolls continued to deepen, down 4.22% in June versus a revised 3.96% (was 4.00%) in May.

The unadjusted annual decline in June payrolls was the deepest since a similar decline at the trough of the 1958 recession, but still shy of the 4.9% trough seen in the 1949 downturn. When the 1949 annual low growth is broken, possibly next month, the annual percentage contraction in payrolls will be the most severe since the production shutdown following World War II (see the accompanying graph).  

Concurrent Seasonal Factor Bias. The pattern of impossible biases being built into the headline monthly payroll employment continued, with an upside bias of 46,000 jobs in June 2009 reporting (see the accompanying graph). Instead of the headline jobs loss of 467,000, consistent application of seasonal-adjustment factors — net of what I call the concurrent seasonal factor bias (CSFB) — would have shown a more-severe monthly jobs loss of about 513,000. This pattern has generated an upside reporting bias seen in 10 of the last 12 months, with a rolling 12-month total upside headline-number bias of 1,210,000. A worksheet on this is available upon request. (See SGS Newsletter No. 50, for further background.)

Birth-Death/Bias Factor Adjustment. As discussed in the prior Birth-Death Model section, the biases from this process tend to overstate monthly jobs gains. Never designed to handle the downside pressures from a recession, the model adds a fairly consistent upside bias to the payroll levels each year, currently averaging about 76,000 jobs per month. The unadjusted June 2009 bias was 185,000, up from 165,000 the year before, but down from 220,000 in May.

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 (including multiple job holders), showed June employment fell by 374,000, after falling by a reported 437,000 in May. At work here continue to be poor quality seasonal adjustments.

The June 2009 seasonally-adjusted U.3 unemployment rate showed a statistically-insignificant increase, to 9.51% +/- 0.23%, from 9.36% in May.  Unadjusted U.3 rose to 9.7% in June, from 9.1% in May.  The broader June U.6 unemployment rate rose to an adjusted 16.5% (16.8% unadjusted), from 16.4% (15.9% unadjusted) in May. The less-than-proportionate seasonally-adjusted increase in the U.6 measure, versus the U.3 measure, again reflected seasonal factor distortions.

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 bulk of the discouraged workers. Adding them back into the total unemployed — unemployment in line with common experience — as estimated by the SGS-Alternate Unemployment Measure, rose to about 20.6% in June, versus 20.5% in May. Updated graph and data have been posted to the Alternate Data tab at www.shadowstats.com.

Week Ahead. Trade Deficit: Consensus expectations are for minor deterioration in the May trade deficit versus April, per Briefing.com, due for release next Friday (July 10th). Chances remain fair for a worse-than-expected number, given a likely accelerating catch-up in rising oil price costs and the impact of same on imports.

_______