Fuzzy Math on Construction Jobs
One aspect of the past year’s economic slow-down has been weaker job creation. The 1.4 percent employment gain over the past year is down from 1.8 percent over the preceding 12-month period. Nonetheless, a growing chorus of analysts challenge that deceleration has an upside exaggeration. Because jobs in the homebuilding industry have fallen less than related metrics imply, this view asserts, flaws in the official job count now overstate reality and a more accurate reading would depict a softer labor market.
Representative of this view is a July 10 report from Deutsche Bank Securities: “If employment in homebuilding had declined by 25% from its peak as key indicators of some construction activity have, employment would be about 750,000 lower than it is now and the nationwide unemployment rate would currently be in the neighborhood of 5%...”.
After considering the various factors potentially inflating the official tally of housing jobs, the report singles out the loss of officially uncounted jobs held by illegal immigrants as most important: “We estimate that about two-thirds of the ‘puzzle’ (or 500,000 out of the unexplained 750,000 shortfall in construction job losses) can be explained by job losses that have been incurred by undocumented workers.”
While we agree that the official measure of housing construction jobs is probably too high, the Deutsche approach has two important flaws. First, it arrives at a 25 percent estimated decline in “construction activity” by averaging a bunch of various metrics – housing starts (down 36 percent), housing completions (down 31 percent), inflation adjusted construction spending (down 24 percent), and residential investment as measured in our National Income and Product Accounts (down 17 percent) – since the start of 2006. (Actually, those four numbers average 27 percent so perhaps they were trying to be conservative).
The problem with doing that is that not all measures are equally relevant. In particular, the largest declines since early 2006 are posted by housing starts and completions, though they only represent small fractions of housing construction activity. In May, for example, 123,000 housing units were started and 128,000 units were completed, but 1.2 million units were under construction. This latter measure (altogether omitted from Deutsche’s list), is not only theoretically and empirically more relevant, it also shows a smaller decline (18 percent) than most other metrics.
Our preferred approach to checking up on the official tally of housing employment compares it with (1) housing units under construction, and, (2) inflation-adjusted spending on home improvements, each lagged by 3 months. This latter measure, which has not yet declined at all, is relevant since construction workers do more than just build new homes. And while the historical relationship among these metrics suggests the official measure of housing employment is too high by 65,000 positions, that’s hardly worth losing sleep over.
Moreover, even if the number of jobs lost by undocumented workers is much greater than we think, it is still fundamentally unbalanced to account for their elimination without also giving equal credit for their creation in the first place. So even if we’re wrong on the magnitude of undocumented job losses in housing (no, not backing down), it would still be equally valid to say that officially reported unemployment rates were too high in the past as to say that now they’re too low.
Richard DeKaser is Chief Economist at National City Bank.
NationalCity.com/Economics
Copyright 2007. All Rights Reserved.
