The net job loss in the December nonfarm payroll (NFP) is unwelcome news, but given the (upward) revision in the November figure, one shouldn’t think of this number as fixed. Figure 1 shows various employment series.
Figure 1: Nonfarm payroll employment (blue), civilian employment (red) and civilian employment “adjusted” to conform to payroll concept (green), all in ‘000s. NBER defined recession dates shaded gray, assuming last recession ends 2009Q2. Source: BLS December 2009 release, via FRED II, and NBER.
The more volatile household based civilian employment series exhibits continued downward movement.
What about the distribution of job losses? In Figure 2, I plot total employment relative to 2007M12, for total NFP, as well as for certain categories.
Figure 2: Nonfarm payroll employment relative to NBER defined peak (blue), construction (red), manufacturing (teal), retail trade (green bold) and finance, insurance, real estate-“FIRE” (brown) relative to 2007M12. NBER defined recession dates shaded gray, assuming last recession ends 2009Q2. Source: BLS December 2009 release, via FRED II, and NBER.
I’ll merely observe that finance, insurance, real estate (FIRE) hardly seems to me like a large component of the overall drop in employment. And for all the hullaballoo about the end of the consumer society, employment in retail trade has not accounted for most of the decline (although 0.9 million is not inconsequential, and does represent a big departure from trend). Rather, the biggest hit has been to manufacturing, which is highly procyclical.
In my last post on the employment situation, I cited the Valletta and Cleary (2009) piece which noted little evidence that there had been a large increase in structural unemployment. Dr. Valletta has updated his assessment in a personal communication.
I’ve updated the sectoral dispersion and related analyses into fall 2009, and I’ve also done some additional analyses (e.g., unemployment duration by industry) — nothing written up yet, though. The updated data don’t appear to provide any more support for the sectoral reallocation interpretation of elevated unemployment (and the expected recovery path) than the data through mid-2008 did. Dispersion of employment growth has continued to grow, but the extent and pattern appear largely consistent with the usual cyclical pattern of unbalanced job losses in a recession (a la Abraham and Katz in JPE from the mid-80s). Some of the evidence that I’ve uncovered suggests that workers who lost jobs in the financial sector (but not construction) are facing an unusual degree of difficulty for finding new jobs, but the number of such searchers isn’t large enough to have major implications for structural unemployment.
Bottom line: no problem with extrapolating from mid-2008, the facts and arguments don’t appear to have changed materially.
Additional research, with similar conclusions, has been conducted by Chicago Fed economist Ellen Rissman .
I thought it of interest to examine government employment, noting the impact of employment associated with the 2010 Census.
Figure 3: Government employees (blue), civilian employment (red) and civilian employment “adjusted” to conform to payroll concept (green). NBER defined recession dates shaded gray, assuming last recession ends 2009Q2. Source: BLS December 2009 release, via FRED II, and NBER.
Even taking out the Census-related workers, government employment has been declining.
One bit of good news is that aggregate hours do seem to be rising, even if private employment is still declining.
Figure 4: Nonfarm payroll employment relative to NBER defined peak (blue), private sector employment (red) and aggregate weekly hours (green), all in logs. NBER defined recession dates shaded gray, assuming last recession ends 2009Q2. Source: BLS December 2009 release, via FRED II, and NBER.
Finally, the employment numbers are an input into real time estimates of GDP.
Figure 5: Real GDP, 3rd release for ’09Q3, in billions of Ch.2005$ SAAR (blue bars), e-forecasting release of 8 January (red line), and Macroeconomic Advisers release of 14 December (green line). NBER defined recession dates shaded gray, assuming recession ends in 2009M06. Source: BEA, e-forecasting, Macroeconomic Advisers and NBER.
e-forecasting notes that these figures imply a SAAR q/q growth rate for GDP of 5%; Macroeconomic Advisers forecasts 4.2% (as of 12/14).
Update 1/9/10 10am Pacific
Chris Vavares of Macroeconomic Advisers is quoted in RTE/Hilsenrath as now forecasting 5.4% q/q growth (SAAR) in 09Q4.
RTE/Izzo also makes note of the possible impact of weather. Below, I plot the standard (seasonally adjusted) series and the unadjusted series.
Figure 6: Log nonfarm payroll employment, seasonally adjusted (blue), and NFP (red). Shaded areas denote December-January. Source: BLS via St. Louis Fed FRED.
One thing to notice is that the drop in 09M12 is larger than the average drop over the past twenty years. Running a regression over the 1989M01-09M12 period, one finds the annualized drop in December is 1.2% (log terms), while the December 2009 drop was 3.6%.
Update 1/10/10 6:20PM Pacific
Here’s a breakdown of government employment into state, and local employment. However, in contrast to this Economix post, these series exclude the education component.
Figure 7: Local government employment, ex. education (blue), and state government employment ex. education (red). Shaded areas denote NBER defined recession, assuming trough at 2009M06. Source: BLS.