Bloated Purchasing Managers Survey. The December 2009 ISM purchasing managers manufacturing survey surged in yesterday’s (January 5th) reporting, with the seasonally-adjusted diffusion index (50.0 and above is positive) rising to 55.9 from 53.6 in November. Set annually by the Department of Commerce, however, the seasonal adjustments have changed meaningfully during and due to the recession. If the pre-recession seasonal factors used in 2007 (based on 2006) were applied, the November index would have been at 53.3 instead of 53.6, but December would have been 54.0 instead of 55.9. 

Although the index still gained in December using the 2007 seasonals, the gain was 0.8 versus the 2.3 points reported. The employment component officially increased from 50.8 in November to 52.0 in December, but using the 2007 seasonals, the increase was from a contractionary 49.8 to 51.1.

In other component indices, official new orders rose from 60.3 to 65.5, but the 2007-based sub-index increased from 60.0 to 62.4. Diverging in direction was the production index, where the official index rose from 59.9 in November to 61.8 in December, but with 2007 seasonals, the index fell from 59.6 to 57.8.

The non-manufacturing survey, which has little meaning outside of its employment and price indices, has minimal seasonal adjustment to the employment number, which was reported at a contractionary 44.0 in December, up from 41.4 in November.

Employment Indicators. Beyond the ISM’s December employment indices, which tend to be leading to the government’s January, not December, reporting and generally were less negative or slightly positive versus November, the Conference Board’s help-wanted advertising indices rose both for newspapers and online. While these data are seasonally adjusted, the unadjusted data are not available. Accordingly the impact of seasonal distortion cannot be quantified, but it likely was a major contributing factor. 

The seasonally-adjusted November newspaper index rose to 10 after two months at its historic low of 9. The series was down by 23.1% year-to-year. Seasonally-adjusted new online advertising in December was up 13.2% from November, but down 6.3% from a year ago. In November, ads were down 0.8% for the month and down 23.9% for the year. If seasonal factors were bad here, such impacted both the monthly and annual rates of change. Again, it is the November numbers that tend to be leading to the broad December employment/unemployment reporting.

As to the seasonally-adjusted weekly new claims for unemployment, using 2006 pre-recession seasonal factors would have generated a four-week average as of the latest reporting (December 26th) of 541,000 instead of the official 460,000. Also, as discussed a month ago, new claims reflect only the jobs lost, not the pace of hiring, which usually would be signaled by activity in help-wanted advertising. Unfortunately, anecdotal evidence does not support the reported surging seasonally-adjusted online advertising.

Payroll Employment and Unemployment Rate (December 2009). Due for release on Friday, January 8th, expectations for the December employment and unemployment data are reasonably optimistic, with expectations for an unchanged level of payroll employment, versus an 11,000 jobs loss in November, and a 10.1% unemployment rate, versus 10.0% in November (Briefing.com). Irrespective of economic reality and high risks of big surprises in either direction, my betting is that payroll reporting could be somewhat more positive and unemployment could be more negative than consensus estimates.

Anecdotal evidence suggests ongoing economic woes, and economic reality likely remains much more negative than what will be reported. Given the upside seasonal biases in other reporting, risks are fair that payroll employment reporting could be stronger than expected, showing an actual gain on a monthly basis. That all could disappear in the next month’s benchmark revisions. John Crudele of the New York Post reported yesterday (January 5th), however, that the Bureau of Labor Statistics is backing away from revamping its terribly flawed birth-death model. More on this will follow in the coming weeks.

The December unemployment survey has the best chance for reflecting something close to reality, with recent unemployment rates being revised for new seasonal adjustments with this week’s report. To the extent that human intelligence is applied to the revisions, some of this year’s bumps should disappear, such as the 9.8% September unemployment rate rising to 10.2% in October and then falling back to 10.0% in November. The revised pattern of rising unemployment should be more regular, with some upside risk of the December rate coming in at 10.2%, above the 10.1% consensus.