Deinonychus antirrhopus

The Household Survey says employment rose by 629,000 jobs in July.

Which is a good sign, but offset by the decline in the payroll survey. There is also the higher energy prices, the slightly higher interest rates, and the declining growth in GDP. We may not be heading into a recession, but things have definitely softened and will likely stay soft for a while.

I hope I'm wrong.

The disparity between the two surveys is getting enormous and makes me wonder that the payroll survey has broken down in methodology.

I'm sure you're aware that the two surveys are measuring two somewhat related phenomena and what the growing discrepency is making clear to me is that the nature of work in the US is changing.

This article details the rise of self-employment, for some a positive step and for others a make-do situation until they can get back into the payroll world.

This article makes the point that an underground economy staffed by illegal immigrants without forged papers, may be forming that escapes measure by the payroll data.

Robin,

Why would you think that the payroll survey is screwed up?

Brian, not to answer for Robin, but I do a lot of statistics; when you have two measures that purport to measure more or less the same thing, but diverge this much for this long, you've got a problem somewhere.

Although I'm not an economist, I've been seeing economists arguing for at least ten or fifteen years that 5.5 percent -- this month's score -- is about "full employment". And the confidence scores, which are supposed to be very sensitive to employment conditions, are at record highs.

Anecdotally, I'm noticing that places like Safeway have big "NOW HIRING" signs out, like they did at the height of the boom.

I won't claim to be able to make an economic argument that fits all this together, but just on this examination, the one thing that doesn't fit in very well is the notion that job growth has more or less stopped.

5.5 is almost full employment
Manufacturing is booming
Auto Sales seems to be rebounding
How could the economy keep up the pace it was at?
It had to slow but oil is a problem

Seems like the people are smarter than the pundits and talking heads

Let's hope so

Charlie, I appreciate the effort, and your points are good ones but Brian is a troll. The increasing differential between payroll survey and the household survey have been discussed before here.

Steve,

I think the data do not tell us that the economy is slowing, but that it recently slowed. Whether this was the pause that refreshes or the beginning of a major downtrend is completely unclear from present data.

Personally, I would bet on a temporary slowdown. The extremely positive slope of the yield curve is a far better predictor than LEI or any other. Also, consider what we've been through in the last few years. In 2000, (when the yield curve went flat , BTW) it was hard to believe any of the incipient signs of slowdown, now it is hard to believe anything else.

Regarding the old establishment survey vs. the household survey dispute: Fine, let's split the difference. That's 330,000 jobs. That number would fit well with the initial and continuing unemployment claims data that has come out in the last month.

Robin:

This is not the first time the CES and CPS have shown this kind of discrepency. During the 1990s there were at leat 7 times that their results were different by more than 500,000.

Apr 1991
CPS + 475,000
CES - 217,000
Difference 674,000

May 1999
CPS -669,000
CES -122,000
Difference 547,000

Spet 1993
CPS -302,000
CES +240,000
Difference 542,000

Mar 1994
CPS -156,000
CES +468,000
Difference 624,000

Jun 1994
CPS -230,000
CES +310,000
Difference 540,000

May 1995
CPS -524,000
CES -9,000
Difference 515,000

Feb 1999
CPS -171,000
CES +396,000
Difference 567,000

SOrry, the second example should be May 1991, not 1999.

Pete,

Leading economic indicators are in essence, predictors of the future, granted not perfect predictors, but predictors none-the-less. My guess is we will see weak growth for the next quarter and the monthly labor numbers just wont be there since most leading indictors lead by about 8 months. Now if we were looking at coincident or lagging indicators I'd say you had a good argument.

As for the discrepency between the payroll and household surveys, I think it does point to a possible problem, but the question is where is the problem...with which survey.

Does the payroll survey understate self-employment? Probably. But is self-employment a good thing? Well I guess it is better than unemployed, but it could also increase future uncertainty for some people and cause them to reduce consumer spending...which is what we have seen in the data so far (consumer spending went from 4.1% to 1% with the only increase in services--i.e., durable and non-durable goods actually decreased). During the lase recession, consumer spending was the bright spot. Now on the plus side gross domestic investment is still in the double digits.

I'm not saying we are heading for a recession in the next few months, but the economy is slowing, and that seems to be the conventional wisdom.

My guess (and it is a guess) is that we will see another set of stats diverge: the different methods of measuring GDP. I reckon, on no evidence whatsoever, that there is a structural change underway which makes the payroll survey consistently, and increasingly, under report employment. Technological change has made self employment/contracting out/small unincorporated businesses both more viable and attractive.
We should be able to measure this by looking at how the figures for GDP diverge when counted as output, income or spending. Unfortunately, we don't get final figures for all three until about 18 months after the end of the calendar year.
As I say, I can't prove any of this. Simply noting what's going on around me. For example, I run a little business that 15 years ago, before the www, cheap telecoms and cheap Fed-Ex/air transport would be simply impossible as an independent entity. We would have been a (small) division of a large company and on the payroll.

"Brian, not to answer for Robin, but I do a lot of statistics; when you have two measures that purport to measure more or less the same thing, but diverge this much for this long, you've got a problem somewhere."

What do you think the problem is? As it is right now, pretty much everyone agrees that the payroll survey is the better indicator. Greenspan himself said this.

Until we fix the problem with the household survey, let's go by the better one.

"Charlie, I appreciate the effort, and your points are good ones but Brian is a troll. The increasing differential between payroll survey and the household survey have been discussed before here."

Robin, I don't know what your goddamn problem is. I don't come to this site as often as you do, so I missed what was said in the past (or if I did see it, which isn't likely, I don't remember it). I don't understand why it is so fucking hard for you to simply say what the problem was and then leave it at that. Jesus Christ.

Brian,

Just because everybody agrees that one survey is better than another does not mean that survey is not the one with a problem. Your argument suffers from two logical fallacies:

1. Argument by popularity
2. Argument from authority

Now it could very well be that the household survey is indeed the problem survey, but it is not a good idea to simply assume this.

"Now it could very well be that the household survey is indeed the problem survey, but it is not a good idea to simply assume this."

We're not talking about the public at large. We are talking about people like Greenspan, who aren't basing their claims on a whim. Of course it's an appeal to authority, in a way, but if that's your excuse not to be skeptical of everything, then I am not sure what to tell you.

We're not talking about the public at large. We are talking about people like Greenspan, who aren't basing their claims on a whim.

Again, faulty logic. Sure Greenspan is a smart guy. And there is a good chance he is right, but really smart economists have made some really bad blunders in the past. So, it is foolish to just assume the problem is with the household survey.

Of course it's an appeal to authority, in a way, but if that's your excuse not to be skeptical of everything, then I am not sure what to tell you.

How about you look in the mirror and say, "I'm an idiot"? I'm being skeptical of both surveys.

"And there is a good chance he is right, but really smart economists have made some really bad blunders in the past. So, it is foolish to just assume the problem is with the household survey."

We aren't talking about something completely abstract and controversial.

We can look at this in a few different ways. At a very basic level, we can say that the household survey is less reliable because the sample size is smaller. Moreoever, a lot of economists usually consider it more volatile. Additionally, studies have been done. One by Anthony Chan of Banc One Investment Advisors found that the payroll survey does indeed do better at showing an upward swing in the labor market.

And as Brad DeLong points out, not all of the claims coming from those who like the household survey more are that signigicant. Only 10% of the difference between the two surveys is the result of people being self-employed.

"How about you look in the mirror and say, "I'm an idiot"? I'm being skeptical of both surveys."

Hmm, I missed that, because I only read through comments that I made and those by people who responded to me.

Oh well, I'll never understand why I get you so angry. I'm surely not that significant.

For someone insignificant, you sure cost STeve a lot of bandwidth. Which, if I'm not mistaken, YOU're not paying for?

For someone who recognizes they're insignificant, you sure spend a lot of time ignoring comments that might, just MIGHT, have something to do w/ the comments you leave behind.

We aren't talking about something completely abstract and controversial.

And neither am I fool. Look at the Phillips curve. That was standard macro-economic theory and used in policy and was one reason that there was stagflation.

We can look at this in a few different ways. At a very basic level, we can say that the household survey is less reliable because the sample size is smaller.

While this is true, the thing is both have very large sample sizes though. Statistically speaking samples of 500 or more might be enough for the invocation of large sample properties of estimators (i.e., the asymptotic results on consistency and so forth). So this is somewhat of a red herring, IMO.

Oh well, I'll never understand why I get you so angry. I'm surely not that significant.

Because of your distortions and your belief that you know more about economics than me and how you presume to tell me what I think. The latter is one sure fire way to piss me off.

For example, the volatilitiy of the household survey implies stability in the payroll survey...but are we sure that the payroll survey is still all that stable? It may very well be, but is it? Lots of people thought the Phillips curve was stable and it turned out to be anything but. It took years for people...really smart people, to realize there was a problem. The current situation reminds me of that time. People sat around saying, "Geez the employment market should be going like a bat out of Hell...what is going on." Sound like what people are saying today?

Now I don't think it is because economists like Greg Mankiw believe in the Phillips curve, but for some other reason we don't know right now. My money is along the lines of Tim Worstall's hypothesis, a structural change (and yes, we could both be wrong).