Nearly everyone will agree that improving employment is crucial for economic growth. Understanding the various employment indicators is important for any investor who is interested in fundamental analysis. I write a monthly employment preview where I highlight forecasts using several different methods, including one developed by my team. Here is a good recent example.
In this article I want to take a different approach -- a focus on the Bureau of Labor Statistics and the monthly estimate of the change in payroll jobs. If you will take five minutes to look at a few charts, you will have a better understanding than all of the experts you see on TV.
I will make a few simple points and let some great charts (thanks to Feray) illustrate the key points. All of the data come from the BLS Business Dynamics series (seasonally adjusted). This means actual data from state employment offices -- no guesswork or revisions.
The discussion about job creation is very deceptive. Nearly everyone confuses "job creation" with "net job creation." The monthly changes in jobs are only the tip of a wave, concealing massive shifts below the surface.
This chart from the BLS shows that job creation, until the 2007 recession start, ran at 7.5 million to 8 million jobs per quarter. Even at the worst point of the recession, job creation was nearly 100,000 new jobs for every business day. The problem is the rate of job losses (the red line).
Conclusion #1: There is always job creation -- lots of it -- even in recessions.
Counting Jobs
Suppose we wanted to measure the change in employment. One method, the one embraced by the BLS, is to count all of the jobs in the country in one month, count them all again in the second month, and subtract to determine the change. Since the BLS cannot count every job, it relies upon a survey (and a good one) of existing businesses.
The problem? Some businesses do not respond to the survey. The BLS uses great methods to improve the response rate. 65% or more respond in time for the first report. 75% or more a month later. The final result includes response rates of about 90%, which is excellent for survey work.
But what about the missing 10%? Are these non-respondents or (drum roll...) are they companies that are out of business. The difference is crucial to the result. One interesting idea is that you could just take the outcome from the respondents and infer that non-respondents were the same.
This implies something important, what I call the 'Imputation Step." The question is whether we may infer, based upon actual data, that the behavior of business deaths and births is similar to the behavior of the businesses continuing in the sample.
From this chart you can see that the contribution to new jobs from opening establishments has a relatively constant proportion to continuing businesses. Recessions are indicated.
Conclusion #2: It is reasonable to infer that business deaths and births parallel the job effects of continuing businesses.
Let Us Verify
Taking the pattern from continuing businesses and imputing the same behavior to the net change from new and dying businesses is a big step. Let us take a closer look.
This chart looks at the ratio between jobs gained and jobs lost. It compares the ratio for continuing businesses with the ratio for new and dying businesses.
- Conclusion #3A: Over almost 20 years there has been a tight relationship.
- #3B: The relationship remained strong during recessions.
- #3C: The "birth/death" relationship always showed a premium to existing jobs -- a bit stronger result, perhaps requiring an additional adjustment.
The Recent Recession
The action in the latter part of the chart deserves a closer look.
Trouble! The long-term premium of the death and birth jobs disappeared in 2009. For the record, the premium was about 5% over twenty years, and that hypothesis held up over many economic changes.
- Conculsion 4A: The "birth/death premium" was accurate for many years.
- 4B: Recessions did not change this conclusion.
- 4C: Economic "turning points" made no difference.
- 4D: Something happened in mid-to-late 2009 --- after the recession ended!
Focus on the Key Groups
This is where it gets a little tricky. I am comparing a ratio to another ratio. We are looking at a comparison of the "death/birth" business group to the ongoing businesses in the BLS sample. What we would hope and expect to see is an ongoing pattern showing a "premium" of about about 5% -- the 1.05 level.
The chart highlights the dramatic change in the ratio. In the 2001 recession and the 2007 recession, the net change in business births and deaths was actually stronger than the trend. As the recession has ended, the relationship broke down.
Overall Conclusions
There are many inferences to draw from this, but for now I'll stick to the basics.
- Most of the popular criticisms about the BLS are completely wrong. The basic methodology is not "guesswork" or "assumptions" or "creating phantom jobs." The data support the method.
- The key question is not related to the oft-criticized "birth death adjustment" but has more to do with the much larger and more significant "imputation step."
- Something unusual is happening within the last year.
Nice post. The samples given and the charts shows the gains to loss ratio. It's really perfect for my research, it really helped alot. It's easier to discuss this in class.
Posted by: legitimate paid surveys | August 30, 2010 at 11:55 AM
Chris -- Many would agree that the regulatory requirements and uncertainty have put an extra burden on small business.
This is something that we should watch closely, You are correct about the timing of the current data, so let's watch for a couple more quarters.
Thanks for your observation.
Jeff
Posted by: oldprof | July 13, 2010 at 12:10 AM
Pete -- These are good points. The underground economy may be as much as 10% of GDP, and it is difficult to measure. I have no doubt that people supplement their incomes in this way.
It would be helpful to have a good measure, but I do not know how.
Thanks for your provocative observation -- something that I will keep in mind.
Jeff
Posted by: oldprof | July 13, 2010 at 12:05 AM
I would argue that small business has more uncertainty around it in the the form of government involvement. Taxes are always at the top of the list but now Obama social issues are scaring small business. What is your risk of adding a new employee. You are now obligated to provide a 401k plan and health insurance. I myself gave up my business and joined a larger firm and commission based employee. It did cost me some money on my bottom line but it removed uncertainty which i hold at a premium right now in these times. My business had employed four people at the time including myself. I was the only one to join the new firm since they provide the service that these employee's had handled. On an outside look at your chart, the premium disappeared right when the nasty heath care debate started.No?
Posted by: Chris | July 12, 2010 at 10:30 AM
Research is good, but hard data is incomplete on an essential part of the employment-unemployment continuum and indeed to a closer estimation of the Real GDP as opposed to that Reported.
As far as I can tell economists have avoided like a plague or else intoned "ceteris paribus" over the relationship between minimum wage,the welfare system and the Gray Market, the cash and barter economy, which exists not only with respect to the employment of illegal aliens, but for a long time with the movement of people from welfare to work in the recorded economy. During my time as a small business owner, very small business, in the lower income areas, I would describe from experience the Gray Market as endemic, and having a considerable effect when multiplied across the economy.
The fact is we do not know what the true rate of unemployment is nor what the true GDP is and will never know until we can characterize the Gray Market.
This comment is not meant to denigrate the wxcellent work of the Blog author. It is rather a plea for information on what appears to be a quantum mechanism.
With that in mind I leave my email address with the hope that someone can point me in the right direction rather than just to the Egress.
Respectfully,
Pete Speer
[email protected]
Posted by: PeteSpeer | July 11, 2010 at 05:06 PM
How much of an effect has the recent trend of unemployment extentions had? Some people are collecting for over 2 years.
Posted by: John | July 10, 2010 at 07:21 PM
Very nice. I offer the following observations:
1. Some of those people who have lost jobs have started their own companies. They might not have any employees, mabye working from home. We are returning to where we were 200 yrs. ago when most people were self employed. Remember "The Waltons"? It was a depression but everybody was working, even the kids all had jobs to do.
2. Over the last 35 yrs. profit margins have fallen, but this has been masked by a speed up in transactions. Right now you are seeing a slow down in transactions, because people can't get credit. All the local banks have regulators standing right there, looking over their shoulders. Businesses can't get credit with suppliers either, it is all COD.
3. You are seeing massive uncertainty due to health care, taxes, and deficits. We will hit debt at 100% of GDP by November, 2010. So when you say, "something different is happening now," you are right.
Posted by: [email protected] | June 16, 2010 at 11:26 AM
Would it be fair to say this current employment problem is largely an effect of hours that have been cut?
Posted by: anon | June 16, 2010 at 10:41 AM
Excellent post. The description is mad easier with help of those charts.
Posted by: Personal finance | June 16, 2010 at 05:36 AM
Were the job gains 1992-2000 tracking population growth, or exceeding it?
I assume that the static-looking period 2000-2007 was "jobless" because it couldn't climb back to track population.
A falling employment to population ration is probably not "bad" in and of itself, but with rising self-reports of un- and under-employment, it probably is.
Posted by: john personna | June 15, 2010 at 03:04 PM
Very timely discussion - more than five minutes, but time well spent.
As previously stated, yes, possibly just mean reversion, but the change is larger and faster than any other over 20 years. Any guesses as to how long it will take to resolve one way or another? Six months?
Posted by: Proteus | June 15, 2010 at 01:36 PM
Andrew - You are correct in saying that it could just be variation around the mean (which is about 1.05).
I try to take my own advice (and that of Ted Williams) and stay within my "happy zone." This means a fact-based analysis and highlighting remaining questions. If I had to guess, I think it might relate to lending standards and small businesses. The problem is the timing, since the change occurs well into the recession.
Whatever the cause, I think everyone agrees that it is important to foster job growth from new businesses, so the recent shift is a negative factor.
Thanks for the good questions!
Jeff
Posted by: oldprof | June 15, 2010 at 01:06 PM
Two questions:
1. What about the other periods when the ratio was in the current neighborhood (1997-2000) and once in 2003? Perhaps the discount is a mere reversion to the mean?
2. If something unusual is happening, do you think the economic implications are positive or negative?
I appreciate your thoughtful analysis - it's refreshing to see an unbiased opinion in the blogosphere.
Posted by: Andrew | June 15, 2010 at 11:25 AM
Mike C -- I am not sure, but I am working on it. I think the data show that everyone else has been looking at the wrong things for years.
We are looking at something that happened well after the fall of Lehman.
Good question, and I wish I had a better answer.
Jeff
Posted by: oldprof | June 14, 2010 at 11:08 PM
Something unusual is happening within the last year.
What do you think is happening and why?
Posted by: Mike C | June 14, 2010 at 10:34 PM