Graphic2.jpg (11074 bytes)
 

 

 

 

 

 

A FULL EMPLOYMENT RECESSION, EH?

The Providence Journal Editorial Page, December, 1996

After examining Rhode Island’s October labor market data, I can now say that I’ve seen it all! True, traditional patterns of economic activity have changed throughout this entire recovery. And, our historic generators of cyclical economic momentum - new home construction and manufacturing - have largely remained on the sidelines throughout this recovery,

But never in my wildest dreams did I imagine that it was possible for a state to reach full employment and in the same month see employment in retail trade, health services, manufacturing, construction, and, if that’s not bad enough, total payroll employment, all decline!

Only in Rhode Island could this phenomenon occur. If the employment data are accurate, what we saw in October can be viewed as the ultimate "negative" for Rhode Island: When we finally reached full employment, the number of jobs created here actually fell! Oh, the wonders o the service-oriented economy! How is it possible to have declining employment and at the same time reach full employment?

To answer this, it is necessary to realize that there are two monthly employment numbers. The one most people are familiar with is payroll employment, the number of jobs created by Rhode Island firms. Think of growth in this measure as indicating internal job growth. According to 1996 data, Rhode Island’s internal job growth had been falling until October, when it became negative.

The bright spot for 1996 has been the behavior of resident employment, which measures the number of Rhode Islanders who are working. Obviously, Rhode Islanders can either work here or in other states. And, with the improving economies of Connecticut and Massachusetts, not to mention the new casino in Connecticut, Rhode Island’s resident employment has been rising, at times by annual rates approaching 5 percent. It is changes in this type of employment that are directly linked to movements in Rhode Island’s unemployment rate.

So, if the labor market numbers are to be believed, Rhode Island recently attained full employment largely as the result of its residents having to go out of state for jobs.

But, are the data correct? Rhode Island’s payroll employment data, like that of so many other states, are often inaccurate when first released. A year ago, the most recent data indicated that payroll employment had been falling (year-over-year) for every month since May. The conventional wisdom back then was that Rhode Island was in a recession that it would have been unable to escape in 1996. When February finally came and the rebenchmarked labor market data were released, we realized that employment had actually risen every month, and that payroll employment was 10,000 higher than we had been led to believe.

So much for the conventional wisdom in this state!

This same "petering out" phenomenon appears to be occurring again. Payroll employment has either fallen year-over-year (October), or risen by ever-smaller amounts since the second quarter of this year. I expect to see both payroll employment and the unemployment rate to be revised upward, and resident employment to be lowered.

If I am correct, October’s bizarre combination of full employment with declining internal job growth was little more than a statistical fluke.

For October, 1996, Rhode Island will neither have attained full employment nor will its payroll employment have declined relative to the past October.

Unfortunately, these types of data inaccuracies only reinforce the negative opinions Rhode Islanders have about their state. Make no mistake, first impressions do matter. The average person has no technical understanding of these data. The common presumption is that highly accurate labor market data are the rule. In fact, I heard someone who, after analyzing the Rhode Island economy, proudly proclaimed: "Numbers don’t lie."

Obviously, that person wasn’t terribly familiar with Rhode Island employment data.

by Leonard Lardaro

 

bott2.jpg (7892 bytes)