Jul 12 2007 5:47AM | Permalink | Email this | Comments (1) |
Blog This! using: Blogger.com | LiveJournal |

The reports over the last year on construction employment from the Bureau of Labor Statistics are not consistent with the reports on construction spending for the same period from the Census Bureau. Over the last fourteen months construction spending, adjusted for inflation, declined about 8% but construction employment is essentially unchanged. A change in the workweek does not explain the discrepancy. The index for average weekly hours in construction was the same in May, 2007 as March, 2006. Both time series show a similar shift from housing to nonresidential activity.

At least one of the time series will eventually be substantially revised, probably in the annual revisions next spring. But it is not clear which one. Either jobs are being over-counted or construction activity is being under counted. Some of the discrepancy could be due to the layoff of “off the books” homebuilding workers or contract workers who appeared on the payrolls of temporary labor companies. If so, we will never know.

Several economic forecast services predicted a sharp drop in construction employment beginning early in 2007 and are baffled that it has not occurred. Consensus opinion is that the jobs data is more wrong than the construction spending data. That said, it is likely that construction spending will be revised up somewhat for the past year, especially early 2007. If so, this is a partial explanation of why materials cost inflation has accelerated in recent months.