Here are a few facts that you probably do not know about the Payroll Employment Report (unless you are a regular reader of "A Dash"). You should know them, because the Fed does.
- The Bureau of Labor Statistics (BLS) does not actually measure the change in jobs from month to month! We know this may seem confusing. The change in jobs is what everyone talks about, but it is not what the BLS measures. They try to estimate the total number of jobs, using survey techniques. They then compare the estimate from one month to the estimate from the next to calculate the change.
The result: They can be great at estimating the total, and still have a huge error band for the change. If you want more explanation on this point, we covered it here.
- The original report is revised for two reasons, but not because the government is cooking the books. The first reason is that many of the businesses in the survey do not send their reports in on time. What a surprise! Some businesses NEVER respond. The BLS does two revisions, based on more complete returns, and then declares the result to be final — for a time. The second reason for revision is that the BLS sample for the survey includes only businesses that existed at the start of the year. The dynamic economy is gaining and losing businesses all of the time. The BLS eventually takes actual data from state employment offices and compares it to their own count. They adjust the methodology based upon the actual count, using something called a birth/death model.
Result #1: Anyone who claims to see "a trend in the revisions" is being fooled by randomness. You can tune that person out, and move on to useful work. The revisions reflect businesses that were late. Who knows the reason for that?
Result #2: Critics of the birth/death model generally provide a very lame analysis. Any such critic should step up to the plate and suggest a better method. The BLS takes the actual state count at the end of the year. They go back to their methods and try to improve them by building this in. Many critics do not like the need for this revision, but they have nothing better to offer.
- The so-called "internals" are also a survey result. All of the data that people talk about — the job growth by sector, the hours worked, or whatever — are also survey results. The error range for subsets is even greater than for the overall number, since the surveyed population is smaller.
Big-time commentators do not understand this. It is a pretty cheap shot to criticize the government, claim that revisions are political, and offer one’s own view of something like a "birth/death model". The BLS economists do not go on CNBC to defend anything, so it is an easy way to look smart and sell research to institutions. Do not be fooled by this!
- We will never know the real truth–the actual change in payroll employment. The final result, entered more than a year later into official statistics will be the outcome of two surveys.
The result: Even if you knew the actual change (my old stat prof said "God whispers in your ear and tells you TRUTH") you could be seriously wrong. Why? The market is trading on the BLS survey estimate — something that has a wide error band. This is true even though the survey is professionally designed and has a large sample.
You can test this for yourself! My long-time friend and colleague Allen Russell has assisted us in devising the Payroll Employment Game. Those of you reading this can get inside information, the actual figure entered as "truth" for this month — 129,000 jobs. This is the result of our excellent regression model using the various key indicators charted at the PEG site. If you repeatedly enter this number as your guess, you will have the best possible result. You might even win our prize.
The information described here is useful because most of the Street does not get it, but the Fed does. You may not be able to guess the number in advance, but you will have a better sense for the way it is viewed in the FOMC. As we have described in detail, the Fed knows better.