by Calculated Risk on 3/09/2007 06:59:00 PM

So far BLS reported residential construction employment has only fallen 4% from the peak in 2006. Meanwhile housing starts have fallen about 35%. What gives?

Click on graph for larger image.

The most important reason employment hasn't decreased significantly - yet - is that employment tracks completions, and completions are still near record levels. This graph shows starts and completions since 1968. Clearly starts have "fallen off a cliff", yet completions are still near record levels. But completions will follow starts off the cliff soon ... and so will residential construction employment.

As a techinical note: historically, on average, completions have followed starts by about 6 months for single family homes according to the Census Bureau, and by about 9 months for buildings with 2 units or more. However because of the increase in building size in recent years, the multi-family buildings are now taking an average of over 11 months from start to completion. Because of the longer period from start to completion, it has taken a little longer this time for completions to follow starts off the cliff.

There are other possible reasons too: In the construction industry there are many cash workers and illegal immigrants. These workers were probably the first to be let go, and they don't show up in the BLS statistics. See this WaPo article: Immigrants' Jobs Vanish With Housing Slowdown

"There's no work here anymore."
Amilcar Guzman, immigrant construction worker from El Salvador, Dec 27, 2006

Another possible reason BLS reported jobs haven't fallen significantly is because some employers might be hoping for a spring rebound in the housing market, and they don't want to lay off valued employees only to have to search for skilled employees in a few months. At my company, we would avoid layoffs during slow periods if we thought a turnaround was only a few months away. So this is another possible explanation.


And finally a possible technical reason based on how the BLS reports employment.

Residential construction employment is very seasonal. This graph shows both the Not Seasonally Adjusted (NSA) and Seasonally Adjusted (SA) residential construction employment from the BLS. Note: the residential specialty series starts in January 2001, so earlier totals were estimated from the residential building series.

This graph shows a key point: the next four months (March through June) are the main months for hiring construction workers. If NSA construction stays flat through the summer, the BLS will report approximately 300K lost residential construction jobs over the next four months. With the excess inventory in the housing market, it is very possible that NSA residential construction employment will actually fall during the peak hiring months!

So I'm sticking with my forecast of 400K to 600K residential construction jobs lost over the first 6 months of 2007.