State Employment: Even Worse Than It Looks?

The location of these 300,000 "phantom" jobs is anyone's guess; indeed, the BLS is aware of the problem but has not been able to bring its samples into closer reconciliation. One place to look for clues may be yet another BLS dataset: the household survey, which measures the number of people working, as opposed to the establishment survey, which counts the number of jobs. Because of their similarity, the two surveys tend to move together, but the discrepancy is quite large in several states. The household results are much worse on a percentage basis in Alabama, Montana, and Colorado, and on a level basis for California, Illinois, and Texas, and could suggest that all of these areas are due for a downward job revision next March.

The sectoral differences between the state and national samples may also provide insight into the geographic location of the missing job losses. BLS data combined with Global Insight estimates show that administrative/support services (NAICS 56) has the largest gap, with 175,000 fewer jobs at the national than state level. This is not entirely surprising given the volatility and difficulties in measuring a sector that is dominated by temporary workers. It could suggest, however, that the already troubled Florida and Arizona economies, which have the highest concentration of jobs in this industry by far, are doing even worse than indicated. Retail trade (87,000) and durables manufacturing (64,000)—mostly transportation equipment, furniture, and wood products—also have sizable discrepancies. The latter could be an indication of the state survey's inability to keep up with the massive losses in those manufacturing sectors hit hard by the housing and credit crises.

A somewhat crude way to analyze these sector differences is to distribute them across the country based on each state's industry concentration. While a few small states (Wyoming, District of Columbia, and Alaska) show stronger performance using these adjusted estimates, the vast majority of states are revised down fairly significantly, suffering a 0.3–0.4 percentage point fall in job growth this year. The largest drops show that some of the state economies hit hardest by the housing crisis may be in even worse shape.

There are no obvious conclusions as to the source of the state vs. national discrepancy, and an analysis of industry and household estimates does not point to a clear answer. But a 300,000-job gap clearly shows that something is awry. Assuming the national data are closer to actuality (almost always the case), all state job-growth estimates for 2008 should be interpreted with caution, as there is a good chance they are too high.