Let the recession begin

Payroll employment fell by 17 thousand in January, the first month-to-month decline since 2003. The downtick was led by a decline of 27 thousand in construction employment, from 7,475,000 to 7,448,000. Residential construction employment, for builders and special trades (subcontractors), fell by 28 thousand, from 3,107,700 to 3,079,600.

A big drop in construction employment was slow in coming, but declines accelerated late in 2007. Even though housing starts, completions, sales, etc., had fallen sharply, employment in residential construction has declined much more modestly. Between March 2006 and December 2007, residential construction employment fell by less than 10 percent. Single family starts fell by 50 percent during that period, and the number of single family units under construction (a more relevant measure than starts) fell by 37 percent. Total units under construction fell by 25 percent.

The number of multifamily units under construction didn’t decline, but the average multifamily unit uses only about half as much labor and material as an average single family home. As a share of new residential construction value put in place, multifamily accounts for less than one-sixth of the total, even after the single-family collapse.

Revisions in the payroll data that were included in today's release showed weaker growth in total employment as well as construction employment during the past 2 years, but the overall picture didn't change. Various hypotheses have been suggested for the disconnect between construction activity and employment, such as a failure to count illegal immigrant workers, deficiencies in the payroll survey, and confusion in distinguishing between residential and nonresidential workers, but none of those explanations was really adequate. Just as there was initially a lack of apparent impact of housing on consumption, it was mainly a question of lags, rather than the absence of a relationship. Because employment in residential remodeling is included along with new construction, however, there is some cushion.

For some of the discussion and analysis that has addressed this, see the July 2007 comments by Richard Berner of Morgan Stanley, the October 2007 BLS analysis of the construction payroll data, and the discussion of the data in the January 2008 analysis of the effect of the housing slump on productivity from the Federal Reserve Bank of Chicago.

It's not official, but......

Although recessions are commonly thought of as consisting of 2 quarters of negative GDP change, the NBER committee that is effectively the arbiter of business cycles actually uses monthly statistics and judgment to determine when recessions begin and end. Often, however, recessions are well-underway, or even over, before they declare that one has started.

As explained in a memo from October 2003, the Business Cycle Dating Committee looks primarily at 4 measures: real personal income excluding transfers, payroll employment, industrial production, and real manufacturing and trade sales. I’m not completely sure I’ve used the same series as they do, but it appears that real personal income in December (the latest number available), was down slightly from a possible peak in September 2007. Industrial production may have peaked in July. Sales are only available through November, but were down from a possible peak in October. With the January employment decline, it looks like a duck, walks like a duck, swims like a duck, and now quacks like a duck. I think it’s a duck.

February 1, 2008 -

Leave a comment

| | hidden

Trackback  |  Subscribe to the comments via RSS Feed