Mixed results from the Employment Situation July 2

Discussion in 'Economy' started by pinqy, Jul 2, 2009.

  1. pinqy

    pinqy Gold Member

    Jun 8, 2009
    Thanks Received:
    Trophy Points:
    Northern Virginia
    The Monthly Employment Situation and it's not really clear what is happening. Payroll Employment (from the Establishment Survey) clearly went down from May to June and by a lot more than from April to May (-467,000 preliminary, compared to the -345,000 preliminary Apr-May which has been revised to -322,000). But Unemployment only went up to 9.5%. If you can do the math (from the technical notes of Employment and Earnings (page 190) you'll see that none of the changes from the Household Survey for Labor Force, Employment, Unemployment, or Not in the Labor Force are statistically significant in that at the 90% confidence interval they all include 0 in their range. That makes analysis tricky. But my guess, and what's probably the majority opinion, is that directions indicated by the estimates are correct and Employment is going down, the Labor Force is going down, Unemployment is going up, and more people are dropping out of the Labor Force. At 68% confidence the range for the estimates are significantly different from zero.

    Since the payroll employment number changes are significantly different from zero (the range is from -566,435 to -367,565) it's quite clear that Employment is going down, regardless of the margin of error for the Household data.

    Overall, it is wore in June, but there might be a tiny glimmer of hope that things will start to improve, though they haven't yet. 3.2 million more people joined the ranks of the Unemployed between May and June, which is down from the 3.27 that became unemployed between April and May, but it looks like fewer people gained jobs, since Employment dropped by a lot more.

    And I apologize to everyone who has never studied statistics for undoubtedly completely confusing you. In short, the numbers produced are only estimates. And because the Household survey is only 60,000 households a month to represent over 235 million people, the margin of error can be rather large. BLS doesn't publish the error, but does give the parameters to calculate it to get the range. For example, for Unemployment, the estimate is 14,729,000. The standard error is +-205,208. So 68% of the time, if you took different samples, the result would be somewhere between 14,523,972 and 14,934,028. 90% of the time it would be between 14,391,729 and 15,066,271. The 90% level is what's usually used.

Share This Page