Employment Situation (January)

* Mixed report. Payrolls showed surprising weakness but the household survey results were very encouraging. Moreover, readings for hours workedand wage rates point to solid gains in personal income and industrial production for the month of January. All in all, we seeencouraging signs of progress in labor market conditions and expect to see much better payroll performance (ex-census) in coming months.
* The census worker effect was only +9,000 (vs an expected +25,000). This will become a more important special factor over the course of coming months and should peak in May at an estimated +425,000 on a monthly change basis (or +700,000 on a level basis). These workers will disappear from payrolls over the second half of the year.
* We are getting much closer to calling the peak in the unemployment rate. The household survey showed a sharp 784,000 rise in employment (after adjusting for population controls) – best since early-2003. However, we are still concerned that improving employment prospects could lead to a rebound in labor force participation, implying some renewed elevation in the jobless rate. So, the unemployment rate will have to hold below the October reading of 10.1% for a few more months before we can be confident that it has, in fact, peaked.

* Today’s report included the annual benchmark revision to the payroll data. The preliminary indication from the BLS, released in October, was for a downward adjustment of 824,000 (or 0.6%) to the benchmark month of March 2009. The actual revision turned out to be somewhat larger (-902,000 or0.7%). In recent years, the benchmark revision has tended to be only 0.1% or 0.2% – the exception was in 2006 when it was +0.6%. The extrapolation of the annual benchmark to more recent months led to a 1.39 million downward revision to December 2009. So, the average monthly revision from Apr 2008 to Mar 2009 was -75,000 and the average monthly revision from Apr 2009 to Dec 2009 was -54,000. The monthly payroll data are based on a statistical sample and the benchmark adjustment incorporates complete employment records that are only available with a lag. The BLS uses a birth/death model to try to adjust for the inability to capture new business creations and closures in the monthly sample. The model worked very well during the early stages of the recession (last year's benchmark revision was very small) but went off-track as the downturn wore on.
* The most disappointing aspects of the payroll data were a reacceleration of the weakness in construction (-75,000), some surprising weakness in
financial services (-16,000), and a big drop in S&L govt jobs (-41,000). Meanwhile, the most encouraging aspects of the payroll data were another solid rise in temp jobs (+52,000). This category is often viewed as a leading indicator of labor demand. Also, manufacturing was +11,000 – first increase since Jan 2007.
* The manufacturing hours data, together with our estimates for utility output and mining, point to a 1.1% rise in industrial production in January – this would represent the best gain since August. Also, aggregate weekly payrolls (a measure which feeds directly into the calculation of personal income) rose 0.5% in January (the second sharpest rise March 2008). This points to a solid gain in January personal income.

[morgan stanley]