Consider the following data series:

























Aug-06 93
Sep-06 257
Oct-06 155
Nov-06 172
Dec-06 101
Jan-07 204
Feb-07 49
Mar-07 139
Apr-07 184
May-07 44
Jun-07 107
Jul-07 138

Let us suppose that these figures represented the monthly change in payroll employment over the next year.  One can readily imagine the interpretation that might be placed upon the August number, due to be released tomorrow, that job growth is weakening dramatically.  The big increase in September might frighten those worried about future Fed policy, but delight those going on TV to talk about the President’s economic policy.  The October number, announced right before the election, would also be helpful.  And so forth through the year.

Here are the data in graphical form. (click to enlarge)

The data series was created using the advanced tools at The Payroll Employment Game site.  I simply specified the truth growth rate as a constant 130,000 jobs (the number Fed Chair Bernanke thinks we need to maintain employment and reasonable economic growth).  The resulting series is typical of the variation you would get from the BLS establishment survey.  The actual mean for the period is 137K and the average absolute deviation from "truth" was about 50,000 jobs.

I did it a few more times with very similar results.  Readers should go to the game site and give it a try. You will have an advantage over many of the experts you read about our see on TV!

(Next month we may make this and other experiments a bit easier to try).

Economic Pessimists In Denial

Rich Karlgaard has an intriguing commentary on upcoming economic data.  We strongly agree with this assessment, and suggest that readers follow the links in his article.

In interpreting the payroll employment report, however, be prepared for a lot of noise from the survey.  We suggest testing the prediction with our Payroll Employment Game.

Check out the story and links, and we’ll follow with more commentary as the week unfolds.

Link: Economic Pessimists In Denial.

Sure, we hit our Brian Wesbury quota yesterday — but what the heck. Today’s piece by First Trust Advisor’s chief economist is too good not to cite: This week’s economic data is going to be hard for the pessimists to…

Economic Strength?

The indicators for economic strength should be a factual matter.  One might write them down and then monitor them.  Many economists take this approach.  This week will see several important  government reports on the economy including the second quarter GDP revision, pointed to as overly weak when the first estimate came out.  There will also be the payroll employment report.  Interested traders can try out their guesses on our payroll employment game.

Barry RItholtz has done a lot of work on forecasting and explaining the employment numbers, and I am sure we’ll be seeing a prediction from him in the next two days.  You should look at his past work before playing the game.

Meanwhile, check out his latest observations about economic strength and then come back for our viewpoint.

Link: Consumer Confidence and Commodity Weakness Heralding An Economic Slowdown.

The data keeps coming, and its getting harder for the perma-bulls to rationalize the information. Earlier in the month, University of Michigan Consumer Confidence plummeted the lowest level since last October; the blame went to Terrorism fears and higher …

Continue reading…

The Payroll Employment Game

Sometimes a game makes everything clear.  The Payroll Employment Game is now available for your entertainment, education, and comment at this site:

Here is the background.

The big economic news for this week will be Friday’s report on the employment situation, highlighted by the payroll employment report.  There is always intense speculation about the pace of job growth.  The report has special significance because it is the first real data from the month just ended, and because employment growth is a fundamental aspect of economic growth.

At "A Dash" we have written about the problems in interpreting this report, provided an illustration of the measurement errors involved, and suggested some insight into how the Fed looks at the data.

Now we let readers simulate for themselves the perils of forecasting employment growth.

The Fed and the Employment Report

Today was one of almost non-stop speculation about Friday’s employment report and how the Fed will react to it.  ADP had a predicion of job growth under 100K while the consensus is for 140K.  Most of those interviewed on CNBC were talking about differences in Fed policy based upon 50K changes in the result.

Readers of "A Dash" can get an insight into how the Fed is likely to respond.  I tried to do this last month with a description of methodology and an example of the error rate.  Sometimes it takes a graphic analogy, well outside of the problem area, to make the point.  I shall try again.

As you read this, you should be aware that the Fed Governors and their excellent staff understand what I am explaining.  They get it perfectly.  Unlike most on the Street, I am not speaking to them, trying to tell them their job.  I am trying to help the rest of us understand what they already know.

Let us suppose that we are in a bar that has a dartboard, a tried and true Wall Street analogy.  Next let us suppose that we set the throw line fifteen feet from the dart board.  The bullseye represents "truth" as we call it in statistics classes.  It exists, but it can only be estimated.  In this case it is the actual payroll growth for the month.  A player (perhaps having consumed a few brewski’s) steps to the line and throws at the board.  We mark the location of the first dart.  That is the first payroll jobs estimate, based upon only partial returns from a survey.

Now we let the player step up a few feet to a twelve-foot line.  The player throws again, and we mark the location of the second dart.  This is based upon the first revision of the payroll survey results.

Finally, we let the player throw from a regulation distance of about eight feet from the board.  The result is the payroll number after both revisions, about 90% of those originally asked to respond.

Here is the key point:

We now go to the board and draw a new bullseye around the position of the first (and least accurate) dart.  We accept that as "truth."

Briefly put, it is quite possible that ADP, or, or Ed Yardeni, or some other economist made an accurate prediction of truth, because we actually score the game based upon where the first, and least accurate, dart lands.

Even the final dart has an error band of over 400K jobs at the 90% confidence level.

Most employment numbers, probably the vast majority, are off by more than 50K jobs even after the final revision.  Despite this, the Street will swing billions of dollars on this result.

The main  point is that the Fed does not do this.  They understand the nature of the estimate.  They look at three-month groupings and trends, where the error is reduced.

I’ll pursue this further with how an investor can make use of this information, but it is important to keep in mind when viewing Friday’s report — and the revisions for prior months.  Also please note that the revisions change the base for the current month — often forgotten.

The Cognitive Bias of Ed Yardeni Critics

I admire Barry Ritholtz’s enthusiasm for analyzing Street Research — not peer reviewed and often methodologically flawed.  His critcism of Dr. Ed Yardeni’s New York Times comments, however, is both incorrect and more than a little unfair.  I’ll cite specific errors, but first look at Barry’s argument:

Link: The Cognitive Bias of Ed Yardeni.

One of the more astonishing things I’ve come across recently was an utterly disengenuous article in Sunday’s Times: Navigating the Fog in Jobs Data. The article is a discussion with (former Prudential) Strategist Dr. Ed Yardeni. It seems that he has been …

Continue reading…

Redux: Household versus Establishment

While Barry Ritholtz’s blog is a big favorite for lots of us, I do not think that David Malpass, Chief Global Eocnomist for Bear Stearns, is going to drop by to comment on the employment numbers.

His work in the current cycle has been brilliant — right on target throughout, so let me summarize a few points.  The key idea is that he thinks the Household survey is a lot better.  But look first at the background from Barry, including some great charts from the BLS.

Link: Redux: Household versus Establishment.

One last item: The Labor Department’s payrolls report is also at odds with its own survey of households, which is used to calculate the unemployment rate. The household survey showed employment grew by 387,000 in June, in line with ADP’s figures.The …

Continue reading…

US Jobs Data Under Focus Today

This is a good explanation of how the market was responding.  I have tried to explain how and why this can happen — will happen — and quite often.  ADP may be proved correct.  Time will tell.

Link: US Jobs Data Under Focus Today.

David Jackson submits: Excerpt from our One Page Annotated Wall Street Journal Summary (which you can get emailed to you every morning by signing up here): AHEAD OF THE TAPE: Prediction Market Summary: The Labor Department releases its monthly jobs …

A Helpful Analogy

Let’s suppose that I get out my headphones and IPod and take the dog for a walk.  Let’s further suppose that someone has given me a gadget — a pedometer — and at the return from my walk it says that we have traveled exactly one mile — 5280 feet.  If my stride is even (sometimes a little tough on a dog walk) I might be pretty close to the estimated distance.  Let’s say the walk was actually 5290 feet.  Well getting it within ten feet would be excellent, an error of less than two-tenths of one percent.

Taking this little example another step, let’s suppose that the next day I choose a different route and I plan to stop for a break after walking exactly one mile.  Suppose that I estimated even better than the day before, missing by only one tenth of one percent, stopping after 5275 feet.

Now finally, let’s suppose that a Genie appeared and offered me a prize if I could tell him the actual difference in length between the two walks.  I would win only if my estimate of the difference was accurate to within two feet!

The problem in winning the prize is pretty clear.  While my estimate of each walk is excellent, my estimate of the relative difference in these two large distances is not so good.  An error which is small in percentage terms in the first case, is quite large when viewed as a percentage of the deviations.

The difference of two feet in the dog walks is the equivalent of a 50K change in the estimates for monthly non-farm payrolls.  No wonder it is difficult to win a bet with the market "genie" on this report.  Even if you knew the truth, you could lose because the estimate was wrong!

Employment Report Methodology

Today’s market focus was the monthly employment situation report.  The numbers get a lot of publicity, including the popular media.  Most active market participants know that the numbers are adjusted in various ways and that they also get revised.  Even the best-informed observers miss some key points.  If you read this through to the end, I’ll bet you learn something you did not know.

The featured number, the growth in non-farm jobs, uses the difference between the current and prior months.  Let’s see what that means.  (I’m going to put aside, at least for this particular discussion, the fact that there are seasonal adjustments and a "birth/death" model to account for the changing business population.  It is possible that there is significant non-sampling error.)

The reported results are based upon a survey.  About 400,000 workplaces are supposed to respond with information during the week of the month containing the 12th.  Most people figure that a survey that large must be pretty accurate, and it is.  The 90% confidence interval is about +/-0.1%.  This means that the Bureau of Labor Statistics is estimating the size of the workforce every month and doing it very accurately.

The problem is that the workforce is over 140 million, so a 0.1% error amounts to 100,000 jobs!

So today’s report estimated 135 million jobs.  This means that come September we can be 90% certain that the true number of non-farm jobs is 135 million +/- 100,000.

Why September?  The report was issued today.

Not all 400,000 workplaces report on time.  But they don’t.  Just as some percentage of people will be late filing their income tax returns, some businesses do not report employment data on time.  This is the source of the revisions in the payroll data.  (Sorry to disappoint all of the paranoid hedge fund managers who think that the President brings in a team to massage the numbers, but that is not how it happens.)

The first report, like today’s report for June employment, is based upon returns from about 65% of the workplaces in the sample.  That may seem like a good number, and it does provide meaningful data.  The problem occurs if there is some systemic bias in the businesses that are not filing on time — something business condition that relates to hiring or firing employees.  It does not take much of a bias to alter the result by a tenth of one percent — and that represents 135K jobs.

The second report, which we will not see until August, is based upon returns from about 80% of the workplaces.  The final revision, reported in September when everyone will see it as "old news" accounts for more than 90% of the reporting workplaces, which the government figures is good enough.

So what we did today was compare the first estimate for June with the second estimate for May and calculate the job growth.  We will not know the "official" job growth until September, when we have the final numbers for both periods.  Even then, the result for May and June are both +/- 100,000 jobs.

This is good to keep in mind when you see a parade of experts on CNBC trying to explain why there is a 50K difference from "expectations."

Oh — and here’s something they won’t tell you on TV.  Even those who realize that the job growth is an estimate forget it when they discuss the "internals" of the report.  (This is usually done in a deep and serious voice).

Well where does the BLS get the internals?  These are also survey results, of course!  While they do not report a sampling error for the hourly wage in the document, it is probably +/- five cents.  And to get the error down to that level we need to wait until September.

This is part of a consistent theme on "A Dash."  The immediate market reaction to data is often far too extereme, based upon the value of the new information.