TCS Daily


Fun with Funhouse Mirrors

By Kevin Hassett - July 10, 2000 12:00 AM

When it comes to using government statistics to think about the new economy, what you see isn't even close to what you get. We can't go forward using the data the government gives us to think about the world without looking at why the data can be so misleading, and last week's labor market data are a great example of one of the major problems.

Markets boomed last Friday in response to another sign that the economy is slowing. Last week in this space we went over why the "bad news is good news" mentality is so prevalent. The Phillips curve believers---to add a little pseudo-biblical flavor lets call them Phillipsians---think that slowing employment growth means lower inflation. Last week's report told us that the economy only added 11,000 jobs in June, well shy of the average monthly gain of about 200,000 jobs registered last year. Apparently, the new economy engine of growth slowed to a crawl. But the bright side, the market seemed to think, is that the Phillipsians at the Fed will surely lay off the rate hikes now.

Hold on! That 11,000 job number definitely does not describe what happened in June.

Suppose, to stay with our biblical theme, God selected you, his trusted angel to go down and watch the U.S. economy very carefully. You have a simple job, count the people in the U.S. who are hired, count the people in the U.S. who leave their jobs, and provide the council of angels with a net count of job creation or destruction for each month. In June you would have sat on your perch and counted a gain in jobs in the private sector of about 1 million jobs! Imagine how surprised you would be if you handed your report to a typist, and then arrived at the council meeting to find that your number had changed to 11,000. But that is exactly what happened to the labor data reported last Friday.

How did a million job gain turn into a measly 11,000 job gain? There were two footnotes many people skipped that were a big, big deal in June.

First, the Bureau of Labor Statistics provides a special massage to the data each month before it releases the latest news. The process is known as seasonal adjustment, and what it means in layman's terms is that the agency tries to remove "normal seasonal factors" from the data so that they provide an easier to follow read on the economy. What are normal seasonal factors? Well, every year in June, for example, lots of kids get out of college and go to work selling cotton candy on the boardwalk. The flood of new summer hires shows up as a blip in the data every June, and the statistical mavens long ago decided that the data would be easier read if they took out the average June blip. How big a blip is that? The process of seasonal adjustment knocked the June number from an unadjusted 1 million new jobs down to about 200,000.

A 200,000 job increase in employment is still a pretty good month, how did we get the rest of the way down to 11,000? Well, the number Phillipsians look at is the total number of jobs in the economy, government included, and the government fired 190,000 workers in June. No, the era of big government is not over. They were all census workers who had counted their last noggin.

This kind of funhouse twist to the data goes on all the time. From now until eternity, whenever somebody tries to study the economy, June 2000's meager 11,000 job increase will be a data point that affects the results, and lots of other months in the past have been as screwy. The problem is, this twist is probably not just noise. Think about it. In the new economy, firms are getting better and better at using computers and other machines to automate their production processes in ways never dreamed of back when the old economy scholars were devising their seasonal adjustment schemes. In the old days, production and hiring would fluctuate madly about the seasons. Now, though, such fluctuations are viewed as pretty costly by firms, and are avoided whenever possible. It makes sense that the aggregate impact of seasonality would be diminishing. This doesn't mean that we shouldn't think about the impact of seasonal factors when looking at the data, but it does mean that there is a huge amount of uncertainty surrounding their trends in recent years.

Old fashioned seasonal adjustment is ok if everybody does the same thing every year. It clearly is not ok now. The application of old adjustments to new data obscures the view we have of the world significantly. A problem is clearly evident in the data. The second quarter of 2000 marks the third consecutive year that activity has dropped of precipitously in the second quarter of the year. Hold your hand up if you think we will have the second quarter blues again next year.

What is happening is equally clear. The big increase in activity expected by the seasonal adjusters is not happening because in the new economy firms are sensibly smoothing their activity more.

I suppose we can sit on the beach this summer and thank our lucky stars that the Fed is not going to raise rates because the data have slowed so much, but come September the flip side of the problem will be evident. The seasonal factors will expect job creation to be massively negative as all the seasonal hires experience layoffs. But of course, if we miss them on the way up, we will miss them on the way down, and the 800,000 jobs that were subtracted from the data in June will be added back to data in the Fall. Some month where our angel counts 50,000 net hires will show up as a huge many hundred thousand increase in employment. Expect to see news stories about how the strong numbers are once again driving the Fed to distraction, and a panicky market wondering what to do about it.

New economy believers, on the other hand, will see through the wild fluctuations in the data and recognize that times are still basically good. Growth is high and inflation is low. Heck, we even added a million jobs in June.
Categories:
|

TCS Daily Archives