Employment: Bear trap

Stephen Colbert would be pleased by the pain the employment figures have been inflicting on bears. The October figures were impressive, posting a 166,000 gain in payrolls, nearly twice the level that economists were expecting.

I should stop here, but my inner grizzly won't let me. So once again I resume my hopeless crusade and take issue with some numbers, particularly those related to construction employment.

Construction is a study in contrasts. Residential building has collapsed, with real investment falling 23% between its peak in the 4th quarter of 2005 and the past quarter, with no improvement is sight (it fell 16.4% in 3Q07 versus the year-ago period). In the same time frame, non-residential construction investment has risen steadily, a total gain of 24.5% (and 13% versus 3Q06). But given the fact that residential construction is larger than its counterpart, total construction spending has fallen over 10% over the last 8 quarters (and 6.8% in 3Q07 vs 3Q06).



And yet, total construction employment rose 1.4% between the 4th quarter of 2006 and the past quarter. During this period, residential construction employment (including contractors) fell only 3.1%, while non-residential construction employment jumped 5%. In other words, one fell very little and the other didn't rise as much as it should have. The same holds over the past year.



This graph plainly shows that the response of employment to the fall in construction investment has been remarkably muted by past standards. This is and remains very odd, in a good way, but odd nonetheless.

The construction employment puzzle, redux

Unbeknownst to me, during my summer break many bloggers commented on the construction employment puzzle (see yesterday's post). Brief recap: housing construction has, by all measures, fallen more than 40% from peak while residential construction employment has barelly budged.

Here's Nouriel Roubini's take and Jim Hamilton's is here. The WSJ's Economics Blog also has good info on this (see here and here).

Severl months onwards, the original mystery has only deepened and no one has provided a convincing answer.

I will say that the argument that states that employment hasn't fallen that much because the workers fired so far have been off-the-books illegals is nonsense. Construction employment rose sharply over 2001/2006 tracking housing construction and it should follow it on its way down.

Also take into account that the underlying force driving housing demand, household formation, is expected to be pretty steady over the next decade (a bit less than 1.5 million annually). Thus, it wouldn't make much sense to argue that residential construction employment registered a permanent jump in employment over the past few years and builders are "hoarding" employees while the storm passes.

Employment crow for breakfast

My gloomy outlook for today's payroll figures was, to put it charitable, quite misplaced. The 110k jump in September and the upwardly revised August figures certainly demonstrate a surprising degree of resilience. Which, I might add, is welcome news.

Yet....

My skepticism is not totally unfounded. Just take a look at the following graph, which charts residential construction employment and housing starts.



The relative strength of residential construction employment has been downright freakish considering that housing starts have pretty much fallen of a cliff. Yes, there are lags, but housing starts have fallen for a year and a half, while employment has fallen less than 5% from peak levels. Real estate sales and credit employment has held up just as well.

Something does not compute. In any case, Dr. Gloom himself, Nouriel Roubini, argues that the payroll figures hide some underlying weaknesses.

Unemployment Friday

Is poised to be more entertaining than usual, and I don't mean that in a good way. Apparently, the consensus expects payrolls to rise by 100,000. Now, this is not an outlandish possibility, given that unemployment claims have been pretty steady.

Nonetheless, as I argued last month, there's something very, very fishy about the employment numbers. Sooner or later they'll catch up to a rather unpleasant reality. Will it be this month? I don't know, but my hunch (yes, hunch) is that the payroll number will come in flat to slightly negative, like September. The carnage will probably start this quarter.

It'll be interesting to see how it plays out with long term rates and stocks. A flattish figure probably won't cause stocks or rates to fall much, given that most investors know that the balance of risks is tilted to the negative side.

I don't believe

That the payroll numbers released today, bad enough as they were, bear any resemblance to reality. It's simply not possible. To see why, let's just review some numbers related to construction and real estate activities.

First, a quick review of the most recent housing and residential construction data:

Housing starts: -20.9% (July 2007/2006)
Housing units under construction: -16% (July 2007/2006)
Housing units completed: -22.2% (July 07/06)
New residential sales: -10.2% (July 07/06)
Existing home sales: -9% (July 07/06)

And yet, the Bureau of Labor Statistics (BLS)wants us to believe that employment in real estate services rose in August (600 new positions) and is up 1.5% versus year-ago levels?
Am I expected to think that residential construction employment has only fallen 3.5% in this period?

It gets worse. BLS data shows that employment in the offices of real estate brokers and agents has kept on rising, growing 3% in July over the year-ago level.

My head feels like exploding.

But even if we take these figures at face value, it's clear that eventually payrolls will catch up to activity levels, meaning that over the next few months and quarters the employment numbers will look very, very grim. And if analysts were surprised now, I can't imagine how they'll react then.