U.S. employment data in line with estimates

The Employment Situation Summary for April was released today by the U.S. Department of Labor and the headline data was right in line with estimates.  The unemployment rate came in at 8.9%, up from 8.5% in March. Job losses for April were 539,000 (plus an additional 66,000 due to revisions of prior months).  This is bang on the 8.9% rate, 600,000 job losses median economist estimate.  The initial market reaction seems to be positive.

Here are some of the finer points from my analysis of the April data.  (For more analysis see Calculated Risk and EconomPic Data.)  My overall feeling is that seasonal adjustment factors are back, and they are making this report look somewhat better than it actually is for the Establishment survey, but having the opposite effect on the household survey.

I have not looked into the Birth-death model, which has added phantom jobs during the recession.  When the data are revised, these jobs will disappear from the data and the losses will be much worse.

Establishment survey

• Non-Farm Payrolls (NFP) were down 539,000 in April. However, revisions to February and March months puts them down a net 605,000. The NFP change for February was revised from -651,000 to -681,000, and the change for March was revised from -663,000 to -699,000.
• The last twelve months have the NFP decline by 5.2 million
• Seasonal adjustments were favourable.  For instance, January, February, March, April, unadjusted is down 749K, 688K, 693K, and 611K.  That’s 81K worse.
• Average hours worked was steady at an all-time low of 33.2 hours with seasonal adjustments, but it declined to 32.8 hours on an unadjusted basis. This suggests that even those employed are not being fully utilized.
• Average weekly earnings were $614.53, which is up 2% in the past year.  Watch this statistic because a recovery will be difficult if people are not earning more money.  By the way, the yearly increase was 3.6% in April 2008 and 4.2% in April 2007.

Household Survey

• The unemployment rate was 8.9%, which is 3.9% higher than last year.  This year-on-year change keeps increasing. For example, it was 3.5% in March and 3.2% in January. Generally, an acceleration in the second derivative means the statistics are getting worse.  Using non-seasonally adjusted data, prior recessions back to the Great Depression have generally ended right about the time this measure peaks, usually just before (see my post “Chart of the day: unemployment as a recession indicator”).
• The broad unemployment measure, U-6, showed unemployment increasing from 15.6 to 15.8 percent.
• The unadjusted (NSA) data here is generally better.  For example, the unemployment rate (U-3) fell from 9.0% to 8.6%. Broad unemployment (U-6) fell from 16.2% to 15.4% – that is very meaningful, by the way.  The number of unemployed actually declined 650,000 in the NSA household survey. While it went up 550,000 on a seasonally-adjusted basis – that is a huge divergence.
• The employment /population ratio stayed at a cycle low of 59.9%. Two years ago, it was 63.0%.  This suggests people are leaving the workforce because no jobs are available.

The report was weak.  However, parsing the data found some peculiar things.

There was an enormous divergence in how seasonal adjustments were affecting the household and establishment surveys.  In general, because the seasonal factors were so large in the household survey, I tend to discount them.  I look at the massive decline in the unadjusted U-6 statistic, our broad measure of unemployment, as a potential leading indicator.  On the other hand, hourly earnings and the average work week point to some serious slack in the economy.

So, my general take is this: this was a weak report showing that the labor market has considerable slack.  I expect the unemployment rate to increase into double digits.  More insight is not available because seasonal factors have taken on an over-riding significance.

Source

Employment Situation Summary – U.S. Department of Labor