July 03, 2003

The Help-Wanted Index

Slate's Daniel Gross has a nice piece on the Conference Board's help-wanted index:

No Help Wanted for Help-Wanted - A tribute to the Help-Wanted Index, the economic gauge that shouldn't work but does. By Daniel Gross: Feeling more optimistic about the job market? Check out the Conference Board's Help-Wanted Index. The 52-year-old index, which measures the volume of help-wanted classified advertisements in newspapers, stood at 36 in May (with 1987 as a 100 baseline), its lowest level since 1961, and off 50 percent from February 2001.

Should we care about this seemingly Jurassic metric? Surprisingly, yes. The Help-Wanted Index is oddly comforting, an economic gauge that seems archaic but remains incredibly useful. The index is derived from a medium that is slowly losing its audience. It ignores technological advances that have utterly changed what it is measuring. And it lost its bearings for a period in the late 1990s. (Truth be told, who didn't?) Even so, this dead-tree index is still vigorously alive.

Since the Truman administration (1951), the Conference Board has been diligently collecting information on how many help-wanted ads were sold by one newspaper in each of 52 representative labor markets. (One can envision rows of clerks with green eyeshades sitting by their Bakelite phones waiting to punch the latest numbers into their NCR adding machines.) At its inception, the surveyed cities—which included megaplexes like New York and Los Angeles and smaller burgs like Charlotte, N.C., and Omaha, Neb.—contained about half the non-agricultural employment in the United States. Today, the reports are filed by 51 newspapers representing the same areas (when the Newark Evening News went to newspaper heaven in 1972, the city was excised). And they are still representative of the national labor picture, despite the intervening demographic and population shifts. "My predecessors made sure individual regions weren't over- or underrepresented," said Ken Goldstein, an economist at the Conference Board.

It turns out the board's failure to adapt is admirable. The usefulness of a statistic lies in the fact that it doesn't change all that much over time. If you continually altered the mix of leading economic indicators, then you wouldn't have much basis for historical comparison. And with no easy way to measure job vacancies, help-wanted ads—which are easily tabulated by those who sell them—have emerged as a useful proxy.

There's an inverse relationship between unemployment and job vacancies, known as the Beveridge Curve, after the British economist who described it. The curve thus predicts that periods of comparatively low-volume advertising would coincide with high unemployment, and periods of comparatively high-volume advertising would coincide with low unemployment. (This paper has a good description and representation of the Beveridge Curve.)

The Help-Wanted Index shows that's pretty much been the case. The index fell from 100 in 1987 to the low 60s in 1991 and 1992—the depths of the last recession and jobless recovery—and then rose throughout the mid-1990s as payrolls swelled. But in the late 1990s, the neat Beveridge Curve began to get sloppy. Even as unemployment plummeted to record lows, the Help-Wanted Index never rose above its 1987 baseline of 100. It peaked in February 1999 and then faded amid the height of the frenzy. "We wound up with the anomaly of advertising volumes that were below the 1986 average, and yet a labor market tighter than we've seen in decades," says Ken Goldstein. His theory: The labor markets were so tight—and for so long—that personnel managers stopped believing that putting out another help-wanted ad for a Web designer was going to help.

Once the bubble popped, the historical relationship came neatly back into line. Since early 2001, more than 2 million jobs have been lost, unemployment has risen, and, as one might expect, the Help-Wanted Index has plummeted.

But those of us in the Internet business wonder if there was also a structural change that would explain why the index didn't top 100 in the late '90s. After all, people don't read newspapers like they used to. Job-seekers and potential hirers now have a range of outlets to choose from, many of which are cheaper than newspapers. A one-week help-wanted ad in the New York Times costs $549 and lets you reach the 1 million or so print subscribers for seven days. A job-seeker has to pay $3.50 to get the Sunday edition, which has the most ads. A 60-day ad at Monster.com costs $305 and allows an employer to reach 15 million unique visitors per month. And a job-seeker doesn't have to pay anything to browse through the site's 800,000 job listings. Mediabistro has 297 media job listings—compared with none in today's Times. And virtually every company posts job openings on its own Web site.

But for all the audience and economic advantages it offers, the Internet has not taken much of a bite out of newspapers' help-wanted revenues. In 2002, according to the Newspaper Association of America, newspapers hauled in $4.38 billion from help-wanted ads. That's way down from the $8.71 billion in 2000. But industry types say it has more to do with the anemic economy than with the vigor of online competitors. Monster.com, the biggest online job site, reported $416 million in help-wanted revenues in 2002—less than 10 percent of the newspapers' total. And help-wanted revenues for both Monster.com and the newspaper industry fell by about 20 percent for the year. Besides, many of today's online ads are sold by ... newspapers.

The Help-Wanted Index may have lost its utility as an absolute gauge. The number of help-wanted ads in newspapers has declined, so it will be hard to reach the '87 figure even in good times. But as a relative barometer, it remains as valuable as ever. When help-wanted ads rise, it takes a few weeks for people to respond, have interviews, and then get hired. That makes it one of the best leading indicators of employment we've got. Until the Help-Wanted Index perks up, don't expect payrolls to grow.

Posted by DeLong at July 3, 2003 10:37 AM | TrackBack


This article is interesting for the light it sheds on those "Bushies are lying about when the recession started" posts farther down the page. The Help-Wanted Index has been falling since February, 1999.

Posted by: Paul Zrimsek on July 3, 2003 10:52 AM

The recession started in March 2001. Remember, by the way, the Fed went through an interest rate raising cycle in 1999 and 2000. The last rate increase was 50 basis point in May 2000.

President Clinton and Robert Rubin and Lawrence Summers did a superb fiscal job with fiscal policy! Superb. Remember the surplus? Remember how Alan Greenspan worried about the surplus that might devour Atlanta and so opted for the tax cut of 2001? Remember how many jobs the tax cut of 2001 was supposed to create? Where are the jobs?

Posted by: anne on July 3, 2003 11:10 AM


Bush's Record on Jobs: Risking Unhappy Comparisons

For George W. Bush, the race has begun to escape comparisons to Herbert Hoover.

With more than two million jobs having disappeared since Mr. Bush took office in January 2001, he finds himself in danger of becoming the first president since Hoover to oversee a decline in the country's employment. Economists disagree on how much blame, if any, Mr. Bush deserves for the long slump, but even White House aides view the economy as one of the only big threats to his re-election campaign....

Posted by: anne on July 3, 2003 11:11 AM

Local newspapers are best for local job searches. A lot of local businesses do not webvertize but must have print ads to meet EEO guides. Local newspapers filter jobs that are not local- good for students seeking work that don't want to relocate. Even if one is not a subscriber to a newspaper, a daily is a cheap (less than$1) place to start, even cheaper if it comes from the rack at MickeyD. The NYT is not a good example because there are less expensive papers that ad the less expensive jobs. Plus large segments of the US are not online including most of the minimum wage workers that would apply for jobs that newspapers advertise. While professional jobs may move their ads to other media, labor jobs will likely continue to appear in black and white.

Posted by: bakho on July 3, 2003 11:20 AM

The help wanted index came up on this site several weeks ago. I tried to point out then, and still believe, that its a particularly bad proxy for the employment situation, due to structural changes in the ways that companies do recruiting these days. In other words, while its still probably useful for tracking month to month changes in the labor market, it fails miserably when comparing anything going on today with anything going on just a few years ago, let alone a few decades ago.

For what its worth, Slate's "Moneybox" feature was supurb when James Surowiecki, one of the best business journalists working today, was writing it. Since Rob Walker and Daniel Gross started writing the column its been a pale reflection of its former self. Such a shame too, given the high quality of most of the rest of the content on Slate.

Posted by: sd on July 3, 2003 11:22 AM

I am not fully convinced by Daniel Gross' argument.

For one thing, even though local newspapers may be the main places in which certain classes of local jobs are advertised, I wouldn't expect to find many types of white-collar jobs in them. For instance, how many advertisements for Java or Visual Basic .NET programmers would one expect to find in a local, as opposed to Monster or a specialist IT recruitment site?

It is perfectly possible for employment to start rising even as this particular index continues to deteriorate for structural reasons. As much as Gross would like to use revenues as a proxy for number of jobs listed, it simply isn't obvious, or even likely, that there is still a relationship between the two quantities. If anything, the Internet should be seen as an essentially deflationary technology, through which rising use and falling revenues often go hand-in-hand.

I would expect that the higher the skill level and professional training a position demands, the less likely it is that said position will be advertised in a local paper, as opposed to a forum of international reach like Monster; it will probably always pay to put an ad for a babysitter or a carpenter in the local daily, but not much that is higher up on the occupational ladder. Since this recession has hit white collar workers unusually hard, I imagine that the effect will be even more pronounced if and when a recovery in the job market begins.

In ignoring the technological changes that are taking place, the Help-Wanted Index is like a Consumer Price Index in which the goods included never change, even as consumers are spending ever more of their incomes on products that have only been recently invented. Rather than waste time on an index that has seen its' day, why not ask or pay the big jobsites to provide the sort of statistical information required as a web service? Form a working group, define an XML schema that lays out precisely what sort of information you want, and pull the feeds from various sites into a single program that then assembles a coherent picture of the state of the job market as reflected in online advertising. Let technological progress work FOR you, rather than trying to work AGAINST it.

Posted by: Abiola Lapite on July 3, 2003 12:31 PM

Might one suggest that the Help Wanted Index reflects jobs for people who are NOT avid Internet users, especially blue collar workers and low-paid service industry workers? How many waitresses, carpenters, nurse's aides and heavy equipment operators rely on the Internet?

It was also mentioned by bakho that businesses who want local employees may advertise there to avoid relocation costs.

Like all measures of unemployment, the Index has limitations. But it can provide useful trend information.

Posted by: Charles on July 3, 2003 12:47 PM

"For instance, how many advertisements for Java or Visual Basic .NET programmers would one expect to find in a local, as opposed to Monster or a specialist IT recruitment site?"

Back at the height of the boom, there were lots of such ads in the Chicago Tribune, my local paper at the time. When the boom faded, there were many fewer ads.

The number of technical job ads in the Hartford Courant has dropped off dramatically lately, but the number of Hartford-area jobs on Monster has also dropped.

Don't forget that many, many papers have a relationship with CareerBuilder.com, an online job listing site. The Hartford Courant's help wanted section is basically a CareerBuilder-branded section of the paper. I'm pretty sure the jobs are posted on the web and in print.

I wouldn't be surprised if positions are posted in multiple venues anyway, especially if the job is being handled by a recruiting firm. The recruiting firms also seem to like posting generic ads in the paper, which just list a variety of unrelated tech skills they're looking for, with no specific position named. Those skills might be related to specific positions being advertised in detail online.

There may have been some permanent movement of job ads from print to online, but I think the online listings and paper listings are both rising and falling in response to the number of jobs out there. The number of print ads probably isn't way out of line as a measure of the employment picture.

It may not be useful for comparing job statistics between now and the 80's, but it's probably useful for month to month or year to year comparisons.

Posted by: Jon H on July 3, 2003 01:00 PM

"Low-paid service industry workers...."

Folks, we have just used and had a fine response and hired several well-paid nursing specialists through our local paper. These were desirable jobs!

Posted by: bill on July 3, 2003 01:03 PM

Interestingly, in my small-town local newspaper (The Cheshire Herald) I've been seeing a *lot* more ads for services, of the sort offered by one person. These are both classified-style ads, and larger display ads.

They strike me as being the kinds of services people will provide to pick up some extra money when they're out of work in their field, or to generate more income for the family when a spouse is out of work or underemployed. Tutoring, odd jobs, lawn work, etc.

That's probably another indicator that could be used, in addition to the job ads.

Posted by: Jon H on July 3, 2003 01:33 PM

Even if this index is still valuable, now might be the time to really start considering a similar index of new media job listings. This data might not be useful now, but if squirreled away and correlated to the conference board help-wanted index, it could be immensely valuable when there really are structural changes that make the old index no longer valid.

Indeed, starting to record this data now could put us in a situation where we continue to have a valid help-wanted index that goes back to 1951 even as the media for such help-wanted ads change with time.

Posted by: MrO on July 3, 2003 02:59 PM
Post a comment