Benchmark Revisions Boost U.S.State Job Growth

The new data are from the BLS' March benchmark revision of state employment, which aligns the initial sample-based figures with more extensive information from unemployment insurance records that virtually every firm has to file. The revisions, which also bring the state numbers into line with the national ones, affect (in most cases) the non-seasonally adjusted data for the year ending last March, and seasonally adjusted data back to 2002.

California, whose 2006 growth is now measured at 1.7% (versus 1.1% in the original survey), has proven even more resilient as the real estate bubble rapidly deflated. Texas, already surging to the top rank of states following a prolonged slump earlier in the decade, is growing even stronger than earlier reported (3.2% versus 2.5% for the year).

Data collection in Louisiana has been hampered over the past 18 months as a result of Hurricane Katrina; the new, more accurate numbers indicate that the state lost 180,000 jobs after the storm, not the 240,000 reported earlier. Furthermore, the bounce back in 2006 now appears stronger: as of December 2006, the Bayou state's job totals were up 4.4% y-o-y, revised up from 3.4%, putting the state with 60,000 of its pre-hurricane level.

Michigan, lagging the rest of the country all decade and still in the throes of restructuring in the domestic auto industry, gets further bad news, though of course it is no surprise to the state's workers. Losses in 2006 totaled 49,000, more than double the early estimate. Florida growth was revised downward from 3.3% to 2.6%.

Overall, the benchmark changes were mild, with just an average change of 0.5% (and 0.4% if Louisiana is excluded), suggesting that the BLS new sampling system—begun in 2003—continues to pay dividends. There is, however, one area of significant concern: the December 2005/January 2006 jump. While the national employment series is smooth over this period, the state data has an abnormal surge, which once again leaves us uncertain of exactly how to interpret the data. In particular, December 2006/ December 2005 job comparisons may be misleading.

By Dave Iaia