The Ups and Downs in Regional Employment Statistics
Our standard monthly employment report typically provides various employment statistics for the Fourth District and its major metropolitan areas. This month we take a different tack, as the District’s November employment numbers varied widely from our expectations. The Fourth District’s unemployment rate dropped sharply in November, falling to 5.2 percent from 5.7 percent in the previous month. Meanwhile, the national unemployment rate held relatively steady until December, when it saw a substantial increase of 0.3 percent. However, we are cautious about interpreting the large drop in the District’s unemployment rate as a sign of an improving labor market.
For one thing, the drop in the district’s unemployment rate is more likely the result of an atypical calendar and its effect on the way the data are reported than on something in the labor market.
First, Thanksgiving fell very early this year and may have caused retailers to move their seasonal hiring up further in the month than in recent years. Second, in response to the early Thanksgiving holiday, the Labor Department and Census Bureau moved the “reference week” for the Current Population Survey—a key input into the estimation of state and county unemployment rates—to the week of November 4–10. The reference week is the week in a month during which individuals are asked about their employment status; normally, this is the week that contains the 12th of the month. The change in the reference week this month may have influenced November’s state-level statistics. Finally, the change of the reference week, combined with the early Thanksgiving, may have introduced more noise into the seasonal-adjustment process that is applied to remove from the data any seasonally-induced swings in the labor force, employment, and unemployment series.
A look at state-level unemployment rates supports this idea. November’s declines were completely reversed in December in Fourth District states. Ohio’s unemployment rate went from 5.6 percent to 6.0 percent, Pennsylvania’s from 4.2 percent to 4.7 percent, and Kentucky’s from 5.0 percent to 5.7 percent.
We can't construct December’s unemployment rate for the Fourth District because data are not available yet for individual counties. However, we believe the trends in the state-level data will hold for the Fourth District as a whole.
A final word of caution on comparing monthly labor statistics. The 2007 data have not yet been revised and are not equivalent to those of earlier years. Revised data are updated and adjusted for seasonal factors, but they are also markedly less volatile, because the revision process smoothes out natural month-to-month fluctuations. To see this, look at the graph below, which shows the monthly percent changes in employment for Kentucky, Ohio, and Pennsylvania from 1988 to 2007. The data for 2007, which are unrevised, show considerably more variation than the data for previous years. Note though that different degrees of smoothing emerge across states after revision. Ohio’s post-revision data are relatively smooth, while Pennsylvania’s retain significant month-to-month fluctuations. However, these differences are not evidence necessarily that Ohio’s economy has lower month-to-month employment fluctuations than Pennsylvania’s. Rather, they are more likely the result of differences in the way states modify their data during the revision process.
The next chart illustrates the degree of smoothing that can occur in these data series. The chart displays the month-to-month percentage change in the employment data both before and after the last annual revision for Ohio. Note that the data for 2006 prior to the revision (the red line) show high volatility but following the revision the series is smoothed for 2006 (the blue line). Revised 2007 employment data will be released at the end of February and will undoubtedly show much less month-to-month volatility.