Friday, October 2, 2009

ADP and BLS numbers now agree

For what it's worth, I thought I would re-post this chart with today's revisions to the BLS jobs data. Job losses per the BLS had been running lower than the ADP data for the past several months, but with today's unexpectedly large loss, and upward revisions to prior losses, the two series are reflecting nearly identical moves so far this year. So what happened today was not a worsening of the jobs situation, but a narrowing of the gap between two competing measures of jobs. I see nothing in this chart to suggest that these two measures of jobs won't continue to show improvement in the months to come, albeit in an irregular fashion.


stilettoheels said...

Wow...-824K preliminary estimate of BLS March 2010 benchmark In other words, nonfarm payrolls have declined by an additional 103K per month since January 2009.

The ADP and BLS data sets DO NOT agree. The nonfarm payroll job losses are now 10%+ deeper than reported thus far.

This a real serious deterioration in the employment picture.

Calculated Risk will need to change the vertical axis on his chart.

Scott Grannis said...

The preliminary benchmark revision is primarily going to affect the Q1 data, and that is now very old news.

Bill said...


Don't you think there is a real risk that once commercial construction projects finish in the months to come that the unemployment numbers will go way up since there are no new ones in the pipeline?

Scott Grannis said...

I have no way to speculate what might happen to certain projects and whether or not those who might be laid off will be absorbed or not by other sectors of the economy. All I feel confident about is that the trend underway is one of gradual improvement. I think the improvement in the fundamentals (e.g., financial market healing, tighter credit and swap spreads, rising commodity prices, improving global commerce) will in the end outweigh the negatives such as you mention.

Jake said...

wait... so the fact that employment dropped another 260k+ with an 825k adjustment to the downside is irrelevant? you're inability to change your point of view with conflicting data is amazing.

do you still think we aren't in a recession as you did last september DURING the worst recession since WWII (i can guess your answer):

From September 7th, 2008...

"Job losses have not yet reached the pace that is typically associated with recession, and Prof. Ed Leamer of UCLA's forecasting project concurs. He has done some good work replicating the results of the NBER (official recession-calling group) and concludes that we aren't in a recession, that things would have to get worse before we can make the recession call. So far we've only lost a little more than a million jobs. During the (very mild) 2001 recession, we lost 2-3 million jobs. That's little consolation, of course, for those who have lost jobs, but the fact remains that this is in all likelihood still a very mild economic slowdown and not yet a recession."

Scott Grannis said...

The benchmark revisions reflect changes to very old data. They do nothing to change the reality of today.

Jake said...

that's like stating there is no difference in the following two situations (i understand this is exaggerated, but it gets at the point):

situation #1:
*100 people are employed in an entire economy that typically employs 100 people
*10 people lose their jobs in the most recent period, thus 90 people are currently employed

situation #2:
*50 people are employed in an economy that typically employes 100 people (i.e. 50 people have already lost their jobs)
*10 more people lose their jobs, thus 40 people are currently employed

same 10 people, but off a smaller base and after the "weak links" in the economy already lost their jobs. the fact that 50 previous people lost their jobs is "old news", but still VERY relevant.

Scott Grannis said...

You're not thinking about this correctly. Government statistics such as payroll employment are always subject to revision after the fact. Therefore you can never accept what they say at face value. They may be correct, and then again maybe not.

Today we learned that the numbers reported earlier this year were overstating the number of jobs. Does that make the economy even one bit weaker today than it already was? Of course not. The number of people working today is the same as it was yesterday, regardless of the revisions.

The revisions are changes to the past, not the present. They don't change the reality of today, but they do of course change our interpretation of what happened yesterday.

The only change that matters today is the change on the margin today.

dr. j said...

Scott: It would be helpful for readers to know HOW the government calculates unemployment, job growth, the national savings rates and other data points we look to as divining rods. I believe "analysis" is as good as the data and the UNDERSTANDING of how the "number" was derived. Once people understand the way these data points are created, and how really twisted government logic is, the more able they will be to assess the current pickle we are in.
Job data from small businesses is 3 quarters old, when reported, I believe, and "adjusted" for several months after reporting. The large company data is more timely. It is a mish mash or old and new, accurate and inaccurate. Conclusions change as the data is adjusted.

Finally, every time the government "adjusts" a number, it is a nice way of saying"oops, we got it wrong."

Any thoughts on an easy to find location where reader's can go to find out HOW the government calculates the numbers it reports?

Scott Grannis said...

I won't go into a long discussion on government statistics, and unfortunately I'm not aware of a good book on the subject (though there must be several I'm sure). But here's what I've learned:

Employment: The government can't possibly count all the people who are working, so they need to rely on surveys, telephone polling, and guesstimates. Way after the fact they can refine their guesses about the current state of the economy by checking IRS returns. But for the most part, the government is guessing. Revisions to the jobs data can come years after the fact, and they can be huge, as much as hundreds of thousands per month. The monthly jobs numbers are thus only rough approximations, and what is 300K one month might eventually be revised to 50K or 500K. So take them with a big grain of salt.

Savings: this is so bad that I hardly ever pay any attention to it. The formula used to calculate the savings rate is deeply flawed. For example: let's say I sell an investment I made a long time ago for a large capital gain. I saved money to buy that investment, and it really paid off. But when I realize my capital gain, I have to pay taxes on the gain, and those taxes are subtracted from my income. If the taxes are large relative to my income, the government will end up assuming that I had a negative savings rate, when in fact I just made myself rich. A successful investment, according to the government's formula, will result in a decline in my savings. But if I don't realize the capital gain, the success of my investment will never be counted as savings. To summarize: realized capital gains count as negative savings, while unrealized gains don't add to savings.

If I saved $100 and used it to start a company that later became worth $1 million, my million dollar gain would never be counted as savings--only the $100 would be considered savings. But someone who saved $1 million and never made a dime on his investment would show up as having saved $1 million. Does this make any sense? No.

And even if the government could figure out how much is really being saved, the only thing that really matters is how those savings are being deployed. If the government is borrowing all the private sector's savings, then the results won't be nearly as impressive as if entrepreneurs were borrowing that money to create new businesses.

So when it comes to savings, it's best just to think about how the incentives to investing, saving, and working are changing on the margin.

Most of the statistics the government generates are full of guesses that may later prove to be wrong, and by the time they come out they are old news. So you can't really trust anything the government puts out.

What you can trust is market-based indicators like gold, commodities, the value of the dollar, bond yields, credit spreads, equity prices, etc. They give you real-time, unbiased information about what is happening.

dr. j said...

Thank you very much for the thoughtful response.