The birth/death adjustment is indeed a statistical contrivance, but it pales in comparison to the seasonal adjustment, which usually ranges from around -1 million in June to +2 million in January. I wonder how many of the people who are are reporting this 32K increase as the worst economic news since the Great Panic of 1819 are aware that unadjusted payrolls dropped by 1.2 million in July. Or that it was the smallest July drop since 1999. If the seasonal adjustment factor for July (0.12%) were replaced with the seasonal adjustment for July of last year (0.26%), we'd all be talking about a 216K increase in July. Then presumably the stock market would be soaring, Bush would be a shoo-in, and Democrats would be jumping out of windows instead of Republicans.

All of these adjustments are just attempts to smooth out the data so we can try to compare things that aren't really comparable. Monthly employment changes are much too volatile and too imprecisely measured to be talking about a 16 basis point shortfall from expectations as if it meant something. If you must analyze the trend, look at year-over-year increases. They're much more consistent, and they don't require any seasonal adjustment.

July had the biggest year-over-year increase in unadjusted payrolls since February, 2001 (1.27%, or 1.64 million jobs). But it was still only about half the average in the late 90's. Reasonable conclusion: it appears that payroll job growth continues to strengthen, but we're not exactly booming yet.

Nice post Tom. What is puzzling then is why doesn't Wall Street or the economists understand this? If the lower than expected number was mostly due to seasonal adjustments/guesstimates then why didn't they figure this out and act accordingly? And you'd think something was up when the household survey showed a significant increase for July. Maybe this will lower expectations for the August number which then will come in much higher.

Tom:

What AWW said--very informative post, and characteristic of what makes this blog so interesting.

AWW:

I worked on Wall Street for 10 years, and I can tell you with some confidence that 1) nobody can predict anything and 2) nobody would ever admit it in a million years, because if they did, they wouldn't have a job. Economists crank out predictions, analysts crank out recommendations and earnings estimates, and investors act like they actually mean something. When they turn out to be wrong, as they usually do, everyone is shocked, but completely undaunted. It's a great game. It's also why my money's in index funds.

The discrepancy between the household and payroll surveys is a whole nother can of worms. I wouldn't be surprised if there's something systematic going on that's causing the discrepancy to grow. About 15 years ago, the BLS discovered a glitch that was causing some firms to overreport their payrolls. They had to revise down their estimates by a huge number.

Tom, everything you've said about the imprecision of the statistics is known by the professional investment community, correct? So these assumptions are built into their expectations beforehand. The investors are not so irrational in selling off after the surprise as they were in their earlier buying on an expectation of a higher number.

You can't celebrate the payroll numbers when they are good as proof that the economy is taking off and then disregard them as meaningless when the numbers are bad.

If more employment is being accounted for by non-payroll sources, what does that tell you about the direction for employment? Are more people deciding to become entrepreneurs and voluntarily leaving salaried or wage paying positions, or are they turning to self employment out of necessity because of the inability to find those salaried or wage paying positions? I think the latter. It is not a sign of an improving employment picture.

Robert:

Why would you believe that they look beyond the conventional wisdom?

Robert:

I think most in the professional investment community would disagree with me about the imprecision of the numbers (and about a lot of other things too). Though the fact that the market was only down 1.5% yesterday indicates to me that, despite the headlines, most investors didn't think it was the end of the world. The market has moved that much on about 1 day in 5 in the last few years.

As far as disregarding bad news, I consider an increasing year-over-year growth rate to be good news. There's no question that payrolls didn't grow as fast in the last couple of months as they did in March and April, but I don't think anyone should have expected them to. We added more private sector jobs in March and April than we ever had before in those months, and that pace wasn't sustainable. By my criteria, payroll growth appears be solid and getting better, but not spectacular.

I won't speculate about whether people chose to become self-employed or were forced to. You may be right. It's certainly true that the percent of the increase in employment due to self-employment is much higher for this recovery than it was in the past. But it may also have something to due with the fact that self-employment was at an all-time low when employment bottomed out in Jan. 2002. The self-employed are currently 6.9% of all employed, which is lower than 93% of the last 50 years.

"As far as disregarding bad news, I consider an increasing year-over-year growth rate to be good news."

Yes, the numbers this year are better than last year. The question is, what should we expect next year's numbers to be, and what are the monthly numbers telling us about the direction that employment is taking. If growth is slowing, then it may be getting ready to peak and then reverse.

Excuse my pessimism, but there is a lot riding on the answer to this question.

Robert:

They tell us the econmomy continues to grow strongly.

Robert:

It's not just that this year's numbers are better than last year's; they're better by more each month than the last. The year-over-year growth rate has increased for 11 straight months. It increased less in the last three months than it did in March and April, but I don't necessarily see that as a problem. If growth stayed positive but stopped increasing, that would bother me a little. If growth stayed positive but started decreasing, that would bother me more. Maybe it will happen next month, but it hasn't happened yet.

If you want to look at shorter-term growth, like month-to-month, you have to expect more volatility, and you have to use seasonally adjusted data, which I don't believe is precise enough. It would be nice if monthly payroll data was so precise and economist's predictions were so accurate that I could worry when a number was 16 basis points less than what it was supposed to be, but I don't believe they are.

Your pessimism is understandable, but I prefer to wait and see what next month brings. Growth trends usually last a lot longer than this one has, but that doesn't mean this one will. And when the trend reverses, there won't be any warning; it will just happen.