Home > Uncategorized > Shift-F9!


For the second month in a row, the Employment report was stronger-than-expected. Let me attach two caveats, however. First, this month’s exceeding of expectations was much more marginal than last month’s (unless you buy the Deutsche Bank argument; see below). Second, the report was stronger-than-expected mostly because expectations had been ratcheted back by other weak data. When success comes largely because a previously lowered hurdle was then cleared, we call that either (a) the soft bigotry of low expectations or (b) “normal,” if it involves technology company earnings reports. But I digress.

The Unemployment Rate, rather than rising as had been generally expected after the plunge last month, was unchanged. Exactly unchanged, to three decimal places! That will get some conspiracy nuts exercised, but of course any half-witted conspirator would make some change in the Rate to avoid suspicion. It actually sounded more like someone forgot to hit Shift-F9 on the spreadsheet (for those who don’t use spreadsheets: Shift-F9 is the Excel keystroke to force calculation of the page). On further review, however, it turns out that the Unemployment did in fact rise… to 9.6872% from 9.6866%. I think that is called, literally in this case, “close enough for government work.”

Payrolls fell 36,000, but with net positive revisions to the prior two months were effectively unchanged. Also, there were only 15,000 Census workers added, compared with roughly twice that expected, so on both the Payrolls and Unemployment Rate counts the data scored slightly better-than-expectations.

But not that much better. I was impressed that other sub-indicators improved, such as the diffusion index (more industries – roughly half – are now showing payroll growth) and the median and mean duration of unemployment. This is, as far as it goes, good news. If it is true that the employment situation is improving, then it is probably also true that (despite the recent declines in survey measures of confidence) the economy is growing. I find it hard to square those two things with the other data, but all will become clear in time. Perhaps in a deep recession the likes of which we are (perhaps) emerging from, consumer confidence lags all other indicators. Perhaps consumer confidence improves with payrolls – we usually assume this – but not in a linear fashion…when unemployment is very high, perhaps the level of the ‘Rate, and not the change, is what matters. As with many of these indicators, it pays to remember that we don’t have a lot of good data to mine from similar sorts of historical periods (because there aren’t many similar periods), so we need to be careful of making strong statements about the significance – either bearish or bullish – of the data we are seeing, when it seems to be confusing. Maybe it is just that we have the wrong context.

However, some observers did indeed make very strong statements about the Employment Report. People really got exercised by the fact that around a million people said they had a job but were not able to make it to work because of the weather. The normal average for this measure, in February, is 300,000. So, you might say, wow, 700,000 would have been counted as employed but weren’t?!?

Well, not exactly. The BLS doesn’t tell us how many of these people were counted as employed anyway. This measure comes from the Household survey, which establishes the Unemployment Rate; to be counted as employed in that survey, you need to have a job whether you got paid for it or not that week – if you couldn’t work because of weather, you still had a job, so you’re not unemployed; and this is why the Unemployment Rate is probably a better indicator this month than the Payrolls figure.

But to be counted as employed on the Establishment survey (where the “count” of Payrolls comes from), you had to be paid for some work done during the reference week. You are counted as employed, in this tally, if you got paid even though you didn’t come to work (for example, you are salaried), or if you worked as little as a single hour during that week.

There is absolutely no way to tell how many of those million people who were unable to get to work would have been counted as employed had they been able to get to work. More to the point, we don’t know how many of them in fact were counted as employed because they did make it to work for even a little bit, or were salaried. There is absolutely no way to tell. In such a case, there is no way to discern if the economists’ a priori estimates were too strong or too weak on this score, because none of them forecast this (logically useless) number. The BLS also made the astute observation that “some industries, such as those dealing with cleanup and repair activities, may have added workers.” In other words, heck if we know.

Deutsche’s economist (who is on record as saying that a strong recovery has begun), though, speculated that without weather effects, Payrolls would have been up about 300,000. No one else saw anything like the size of that effect, which would also be at odds with an unchanged Unemployment Rate (remember, that’s the cleanest piece of data we have in this report). Deutsche immediately forecast a 350,000 Payrolls rise for next month as a catch up. This is despite the fact that the ‘Rate was unchanged, none of the surveys of opinions about the job market are improving, initial claims are rising, and ADP – which has not historically shown any impact from weather – was about as-expected.

Personally, I suspect that in this economy, the weather would have less impact than otherwise because there isn’t much freedom to shut down business for a whole week, nor to stay away from work if it is humanly possible to make it to work. I suspect the Deutsche Bank economist will be off by at least 250k; I would say the miss would be by more except that there are census hires coming to the fore. (However, I also expect he will revise his guess to not be a huge outlier next month).

As I said on Thursday, what all these crosscurrents mean is that we have bigger error bars on the data than we otherwise would. That means that it is harder to reject the null hypothesis about the rate of growth in Payrolls than it otherwise would be. With the number near expectations, it is extremely hard to claim that we have any information about a sudden shift to a much higher rate of employment growth, especially with every other survey of employment conditions showing at best tepid expansion.

On Monday, there is no economic data on tap. I suspect we will hear some talk about the implications for the U.S. if China severs its peg to the dollar (many people assume the dollar will weaken and China’s unit will strengthen if this happens), which speculation was brought to the fore when the head of China’s central bank was quoted in the Financial Times this weekend. China continues to sound like she wants to find ways to take advantage of the position the U.S. is in right now with respect to needing the financial flows from China. I am not sure whether I would want, if I were China, for the U.S. to call my bluff. If you owe the bank $1,000, you have a problem, but if you owe the bank $1 billion, the bank has a problem. China is the bank.


As an aside, I really enjoy reading Dr. Thomas Sowell’s work. Sowell is a very good economist (and also a very sharp conservative thinker who gets the arguments right, so if that doesn’t describe your politics and you are not open-minded then just ignore the rest of this comment). His book The Vision of the Anointed: Self-Congratulation as a Basis for Social Policy is the only book I have ever sent to several friends, and I am looking forward to reading his book Intellectuals and Society, which was just published in January. But he also writes an online column (that is syndicated at Jewish World Review.com) and recently wrote a few brief, easy-to-understand articles on health care. One of them is here. It is about a 3-minute read, but it encapsulates a bunch of the key arguments against the notion of government-run health care in simple, clear prose. If you are attentive to the health care debate, I recommend you look at this article (and he has a part 2, 3, and 4), even if you disagree (since these are the best points your opposition will make, and you will need to marshal reasonable counterarguments).

Categories: Uncategorized
  1. Robert
    March 8, 2010 at 5:24 pm

    Let us not forget the DES birth/death model employed by the BLS that added 97,000 phantom jobs. So if you take that out, then we actually lost 133,000 jobs.

    The CES birth/death model figures can be found at:

  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: