Everyone realizes that employment is a paramount consideration in evaluating the economic recovery. With jobs, we try to squeeze information out of each data point. Sometimes the information is just not there.
Most media sources do a terrible job of interpreting the information. Here is an easy way to test yourself.
How many jobs do you think were created by new businesses in the second quarter of 2010?
The correct answer is over 1.2 million. If you do not know this, you are a novice when it comes to employment data.
The "official" monthly verdict comes from the Bureau of Labor Statistics in the form of the Employment Situation Report. This includes a survey of businesses and a survey of people. Most focus on the business survey, the payroll employment report and the monthly change.
For several years I have written about the variation and error bands in this report. Others have been much more aggressive in criticizing the official results. This is especially true for those who lack the time or expertise to evaluate the data.
Even those of us who accept the best efforts of the BLS understand that there is plenty of noise in the monthly report. How much signal is left?
This month is especially troublesome. Let's start by looking at our approach.
Our Own Estimate
The non-farm payroll report is based upon a monthly survey, attempting to estimate the total of all payroll jobs for the week including the 12th of the month. Each month my team asks the question, "What change in payroll employment would be consistent with other economic data from the same time period (the week including the 12th of the prior month)?
My answer to this question is not a forecast, per se, since we do not posit any causal relationship among the variables in our model. They are all concurrent indicators of economic activity.
- We use the four-week moving average of initial unemployment claims, culminating in the week of the employment survey. This is the best direct indicator of new job losses. The current level is 413K, with plenty of wild variations in the last few weeks. This value has declined significantly from the five-handle that was so worrisome a few months ago, but is still at levels that do not indicate a real recovery. Please note that tomorrow's report (2/3/11) is not part of this survey period, nor was last week's. Many observers get confused on this point.
- We look at the University of Michigan sentiment survey, which we find to be more useful than the Conference Board's sentiment index. Michigan uses a panel, where some families are carried over from month to month. This is a good technique. Sentiment is strongly influenced by employment. When people have lost jobs, or know others who have, they get worried. It is a very good concurrent indicator. The Michigan index is at 74.2, better than recent levels, but far from what we see in good times.
- We use the ISM manufacturing index. The January report came in at a surprising 60.8. The ISM index is an important read on employment, and it remains the most bullish of the various indicators. Most people do not know much about interpreting this index. The January reading, if annualized, is consistent with GDP growth of 6%.
Our long-term preview record has been very good, especially when compared to the final revised data. This makes sense because our model was derived from the final data. Our approach also makes logical sense, because it involves some factors related to jobs lost, and some related to job creation.
For the current month, our estimate is for a gain of 77,000 jobs. This is a very pessimistic estimate, weaker than the other sources.
I have noted that the preliminary reports have been running "hot" by about 100K jobs, a problem I discussed in this article. The last two years have included annual "benchmark revisions" that have lowered the final job change results. In the final tally, our estimates look much better, but they were too pessimistic in real time.
With tomorrow's benchmark revisions, it might be a good time to check out the score. Who really has done the best in forecasting the actual employment changes? The official score would be actual data from the state employment agencies, the basis for the final benchmark revisions. The test would be which method most accurately forecast those changes in real time.
It is always interesting to compare the job forecasts from different sources. We follow several because of the widely varying methods they use. A wise interpretation would be to consider all of these disparate sources of information. Each method provides an independent measure of important information.
ADP has proprietary data because of its payroll management business. Looking only at private sector jobs -- no government, no census effect, ADP sees a gain of 187,000 private sector jobs.
TrimTabs has another valuable approach -- tax deposits. They are playing it close to the vest this month, but published reports quote them as forecasting an "upside surprise."
Briefing.com cites the consensus as a gain of 148K overall and their own estimate as a bit lower. With the end of the census jobs effect, the reason for distinguishing between public and private jobs has been reduced.
For a deeper look, especially into the ADP report, check out Steve Hansen's preview at Global Economic Intersection.
To summarize briefly, the market incorrectly focuses on predicting the BLS preliminary estimate -- mostly ignoring the 100K confidence interval, the seasonal adjustments, and the benchmark revisions. I am probably the strongest supporter of BLS methodology and integrity, but I still see their approach as only one method out of many.
The jobs report is so important -- and we are all so interested -- that we seize upon whatever information we have, even when we should accept the limitations.
- This month's report will be even more difficult to interpret than most. Once a year the BLS "squares up" their findings with the actual data from state employment offices. This process is described as benchmarking, and it is done annually. We already know from the preliminary estimates that the job changes for the last year will be revised downward. This will affect the overall number of jobs, but not the monthly changes. Hardly anyone will understand what this actually means. If you really want to understand this, one of my best articles from last year will help you get started.
- The BLS will also change to a more frequent revision of the Birth/Death adjustment. This means that the modeling for job creation, an essential part of the estimation process, will now be done more frequently and more accurately. It is good news, but it is a change for this particular month.
- The weather, especially during the survey week, will invite commentary from all sides.
The changes in methodology will improve estimates in the long run, but merely add to the confusion this month. The process is so difficult to understand that it becomes an easy target for the spin masters. This usually means that a good number is immediately viewed as suspect. In general, it has been wise for traders to be neutral or short going into this report. Even if it seems good, there will be enough negative commentary to let you cover.
For my trading accounts I am not doing anything special this month, but I am waiting until after the number to initiate new positions.