Posted 8:17 AM by Robert

Not The Best News For An Incumbent, But ...

... not reason for him to panic, either.

The Bureau of Labor Statistics (BLS) announced today that during July employment increased by 32,000 according to the payroll survey. According to the household survey, which is used to calculate the unemployment rate, total employment increased by 629,000. The unemployment rate fell to 5.5 percent. Payroll job growth was slower than expected ? Wall Street had expected roughly 240,000 new payroll jobs. Manufacturing added 10,000 new payroll jobs and the average duration of unemployment fell to 18.6 weeks from 19.9 weeks in June, its lowest level since March 2003. But job growth in May and June was less than initially estimated.

Charts:

How Many New Jobs?

Unemployment Rate Declining

11 Consecutive Months of Payroll Job Growth

It would have been better re-election news for the incumbent, of course, if the July numbers had revealed a nice big increase in both the payroll and household employment numbers.

Much recent political dialogue has been cast in terms of the number of "new jobs" created. That has led to some fussing over the now-fairly-well-known divergence in the payroll/household surveys. But the best historical re-election correlation is with the actual unemployment rate - which declined to 5.5% and, of course, is calculated from the household survey. In some prior months the unemployment rate has declined without much job growth - or even with some job loss - because "discouraged workers" withdrew from the market - and much has been made of that in some quarters. But July's decline in the unemployment rate was not caused by discouraged workers who withdrew from the market. This decline in the unemployment rate was caused by a jump in household survey employment by 629,000 new workers.

It seems likely that higher energy prices are behind much of what is showing up in these numbers, including relatively soft recent retail sales (although those sales were more influenced by weather than has been emphasized in the media - and this New York Post column suggests a much more benign explanation for the July numbers, a little too benign in my opinion). It is also worth noting that July often has an eccentric employment profile. July is often not a big month for manufacturing hiring and a large part of July hiring is often of a seasonal and/or temporary nature - often of younger people and related to leisure activities. Those jobs are especially vulnerable to spikes in gas prices. While it would likely have been better if people seeking such positions could have found them in larger numbers, the political impact of such people's employment status should not easily be considered to be equivalent to that of people seeking other types of jobs. And then there's the point the Post article makes: [T]he Labor Department's so-called Net Birth/Death Adjustment tabulates July as one of only two months in which there are more companies dying and taking jobs away than creating new jobs. And that "tabulation" has sometimes been seriously out of whack.

Much of the media coverage is treating the unemployment rate almost as a misleading detail. The ever-spinning New York Times drolly puts it today:

The unemployment rate fell slightly, to 5.5 percent, last month. It is based on a smaller survey than the job growth numbers, which are widely considered the more reliable gauge of employment.

That smaller survey than the job growth numbers is the household survey. Contrary to the Times' spin, the matter is vastly more complex than the jobs survey being widely considered the more reliable gauge of employment than the household survey. If what the Times suggests were correct, the unemployment rate would be calculated from the jobs survey. The Federal Reserve Bank predominantly uses the jobs survey (or "payroll survey") for many purposes in connection with choosing its short-term interest rates, and some mostly short-term stock investors use the payroll survey as a basis for placing some of their money. But for many other purposes the household survey is the better measure. For example, a rise in household survey employment caused by a big rise in sole proprietorships will likely not be well-reflected in the payroll survey (no payrolls). Nor do sole proprietorships and payroll employment have the same characteristics for the purposes of many Wall Street investors - so they respond negatively to a jump in one survey but not the other. But a prosperous sole proprietor is still a happy voter. The relationship between the two surveys is considered by essentially all competent observers to be just not understood.

Because the Fed predominantly uses the jobs survey in choosing its short-term interest rates, the employment figures released today may deter the Fed from raising interest rates further for a while. Fed moves today won't have any substantive effect on the economy until long after the election - but the effect on the stock market could be immediate.

Voters will not make their choices in November by referring to abstract government economic numbers. If that were the way voters acted, Mr. Bush's poll numbers would be a good deal higher than they are. As noted here often, voters will choose in November mostly on the basis of the domestic economy - which means how prosperous and economically secure they feel at the time. In the past, that feeling has had more to do with the current unemployment rate than almost anything else - just ask Grey Davis, who was drummed out of office during a period of relatively high state unemployment while shouting that a lot of "new jobs" had been created on his watch.

RELATED: From the Congressional Budget Office (CBO):

CBO now projects a budget deficit of $422 billion for fiscal year 2004, about $56 billion less than CBO projected in March. Nearly all of that change results from higher-than-anticipated revenues.