No. 299: Employment Report Outlook and Some Updates
JOHN WILLIAMS’ SHADOW GOVERNMENT STATISTICS
COMMENTARY NUMBER 299
Employment Report Outlook and Some Updates
June 3, 2010
420,000 May Census Hires Assure "Strong" Jobs Report
Consensus Estimates May Be Disappointed
Continued Data Distortions Generated by
Severity of the Economic Downturn
PLEASE NOTE: The next regular Commentary is scheduled for tomorrow, Friday, June 4th, following the release of the May payroll and unemployment reporting. Today’s Commentary updates the outlook for that reporting, along with some details of recent homes sales reporting and the benchmark revision to new orders for durable goods. In addition to the May employment/unemployment assessment, tomorrow’s Commentary will review why M3 is in contraction and the implications of same for the economy and inflation.
– Best wishes to all, John Williams
Keep in mind that the 95% confidence interval around the monthly change in payrolls (the census jobs are a separate issue) is +/- 129,000, and the Bureau of Labor Statistics (BLS) has the ability to bring in any given monthly report as it desires. Nonetheless, with the headline jobs number dominated by the census hires, President Obama’s expectations of "another strong" jobs report tomorrow likely will be met, probably exceeding the 290,000 jobs gain reported initially for April. What is not being discussed in all the happy media hype is that the census layoffs should show up heavily in the June payroll survey, with a large monthly payroll contraction almost assured for next month’s reporting.
On the unemployment front, a number of the census hires have other part-time employment and already are counted as "employed" in the household survey, so a fully proportionate drop in the unemployment rate for the census hires is not likely. If all 420,000 census hires came from the unemployed, such would reduce the unemployment rate from 9.9% in April, to 9.6% in May, by itself (consensus is 9.8% for May per Briefing.com). To the extent the unemployment rate is moderated by census hires, such, too, would reverse in the months ahead, beginning with a sharp reversal in June. In April, despite 66,000 census hires, the unemployment rate rose to 9.9% from 9.8% in March. Due to catch-up in poor-quality seasonal factors, the May unemployment rate could notch higher, still, despite the census hiring.
Where the payroll survey counts the number of jobs (someone with a regular part-time job plus short-term census work is counted as two jobs), the household survey, which generates the unemployment rate, counts the number of people with jobs (each person is counted only once, irrespective of the number of jobs worked).
The Census Bureau was kind enough yesterday to publish its weekly payroll data of temporary census employment, which corresponds closely to the census jobs published by the BLS (the weekly payroll involved is the one including the 12th of the month). Keep in mind that the monthly levels of census employment have to be subtracted from the subsequent month’s levels in order to estimate month-to-month change.
In terms of other indicators of employment activity, help-wanted advertising was mixed in recent reporting, with flat to positive indications. In the Conference Board measures, the newspaper advertising index held at 10 in April, for the sixth month, following a revised March reading of 10 (previously 9). The online index, after having shown a 5.7% monthly gain in April, was flat in May for total ads, which include ongoing ads from prior months plus new ads. I believe the new ads number will prove to be a the better indicator for this series. New ads were up monthly by 2.7% in May, following a 1.0% contraction in April. The seasonally adjusted data here are highly suspect for such a new series, and the short history makes it difficult to gauge the significance of monthly gains (how much is due to changing Internet activity, for example).
The employment components of both the manufacturing and non-manufacturing purchasing managers surveys gained in May. The manufacturing employment index rose to 59.5 in May from 58.5 in April, and the non-manufacturing employment index rose to 50.4 in May from 49.5 in April. These are diffusion indices, where readings above 50.0 indicate expansion. There appears to be a warping of the purchasing managers data, which has developed, due to the loss of participants as a result of the severe economic downturn (see comments in the durable goods benchmark revision section).
Foreclosure activity remains a major distortion in these data, with NAR estimating 33% of new home sales for April in the "distressed" category (the March estimate was 35%, but the number of foreclosures still has been picking up slightly). Census acknowledges that a portion of new home sales is from foreclosure activity but offers no estimates. Purportedly, foreclosure activity is on the rise, and some in the construction trade have difficulty competing with the pricing of foreclosed properties. Until the foreclosure problem works itself out, monthly changes in these numbers cannot be taken as meaningful indicators of trends in underlying activity in homeowner real estate, as it relates to general economic activity.
The following updated graphs reflect different measures of home sales activity since February 2009. The numbers, through April 2010, reflect the seasonally-adjusted level of monthly sales, rather than the annual rate usually published.
As shown in the accompanying three graphs, the 2007 and 2008 revisions were unusually large, with stronger than previously estimated orders coming into and in the early official stages of the recession, and with a sharper relative decline in 2009 than reported previously. Beyond that, I would not read much of significance into the reporting. It appears likely that recent durable goods reporting and revisions to same — subject to future benchmark revisions — have been overstated due to distortions from the severity of the economic downturn.
Specifically, the reporting includes estimates of new orders for companies that have failed to report their information. Prior levels and general trends in the non-reporting company’s industry are used to estimate the firm’s orders. As with the payroll survey, however, the government has no timely basis for determining whether a modeled company just failed to report or if it went out of business. In the current circumstance, failed companies have become a meaningful issue.
Separately, but in a related area, the purchasing managers survey also likely has seen some upside reporting distortions resulting from the severity of the economic downturn. Companies reporting the weaker numbers would tend to be those that failed and that would drop out of the survey. The stronger companies remaining in the survey would tend to pick up business from the lost competition.
In terms of the new orders revisions, the first graph shows both the new and old data smoothed with a six-month moving average. The second and third graphs reflect the seasonally-adjusted new orders level and year-to-year percent change, respectively. None of the graphs are adjusted for inflation.