September 01, 2003

Initial Unemployment Insurance Claims

400,000 may be the weekly initial unemployment insurance claims number below which the economy is gaining rather than shedding jobs on average. But in order to reduce the unemployment rate employment must be growing at more than the 1% per year that the labor force grows at.

It looks as through--in the 1990s at least--357,000 is the weekly initial unemployment insurance claims number below which the unemployment rate is falling rather than rising.

Posted by DeLong at September 1, 2003 08:55 AM | TrackBack


Well, we have been far above the threshhold 357,000 new claims for unemployment every week this year. This will be a useful rule of thumb.

Posted by: anne on September 1, 2003 09:19 AM

The 400,000 # has less import than its commonly given; Especially given how sticky the unemployent situation has become, and its present decoupling from economic growth:

1) Productivity gains are allowing companies to increase production yet still not add jobs;

2) Overseas hiring of white collar (middle class) workers allows for GDP growth yet still not add jobs;

3) Post bubble, excess capacity allows for increased corporate revenues and additional layoffs;

4) Public companies have become increasingly focused on making the next quarter's numbers; Layoffs and outsourcing is increasingly becoming the way already lean firms hit their taargets;

When manufacturing jobs went overseas, workers retrained in new economy/service jobs. Thanks to globalization and the internet, those jobs are now migrating to the lowest cost provider.

This is a structural problem that will not be easily solved. It certainly won't be fixed merely with tax cuts and low interest rates . . .

Posted by: Barry Ritholtz on September 1, 2003 09:23 AM

"A scared employee is a productive employeee".,1413,36~130~1560526,00.html

Posted by: Roman Berry on September 1, 2003 10:17 AM

New York Post



September 2, 2003 -- YESTERDAY was the day America celebrated workers. On Friday we'll find out how many people aren't working.
So, as I sat at home this holiday weekend with a plate of potato salad on my lap, barbecue sauce dripping down my chin, I decided to do a different kind of column: a labor-free one.

Here's a list of some employment facts you might want to know and probably don't. These may come in handy if you want to depress a date, scare a colleague or bore a friend.

* The experts expect Washington to report on Friday that 10,000 new jobs were created in August and that the unemployment rate held steady at 6.2 percent. Those would be the first new jobs created since last January.

But the statistical margin of error for the new jobs survey is plus or minus 156,000, and plus or minus 0.2 percent of the unemployment rate. In other words, 10,000 jobs are meaningless from a statistical point of view.

* The way the unemployment rate is calculated was changed back in January, when it suddenly fell to 5.7 percent from 6.0 percent in December. That decline didn't reflect any improvement in the job market, just a change in the way the government takes its count.

* The government says it gets its new job count by surveying 300,000 businesses. But the participants are not selected randomly or scientifically. And there is no way of knowing how many of these companies reply truthfully to the government's request.

* Until this year the Labor Department arbitrarily added large numbers of jobs each month to its count for companies it assumed (but couldn't prove) forgot to respond to its survey or couldn't be reached.

Some months these "bias factor" additions - as they were called - would add up to 150,000 jobs. The bias factor never reduced the number of jobs, because the government never assumed companies that had forgotten to reply might have gone out of business and laid everyone off.

The government recently changed how it comes up with this bias factor but it still adds jobs to the count more often than it takes them away. In fact, only once this year did its bias factor take away jobs.

* According to the government survey of companies, there have been 2,690,000 jobs lost since the peak in employment in February 2001. That was the first full month of the Bush administration.

* When the government surveyed companies in July, it found a drop of 44,000 jobs. But when it asked people in their homes if they had worked at least one hour in the previous month, job losses increased to 260,000.

* The unemployment rate for people at least 25 years old with at least an undergraduate degree from college is now 3.1 percent. It was 1.6 percent when Bush took office. The 3.1 percent is the highest since 1993, when the last recession was ending.

* The average length of unemployment is now 19.8 weeks. It was 12.8 weeks in February 2001. The current level is the highest since it reached 20.4 weeks in January 1984.

* The number of people unemployed for at least 27 weeks is 1,959,000. Back in January the number hit 2,036,000 - the worst since 1993. As comparison: Back in February 2000 the number of people unemployed for 27 or more weeks was just 708,000.

Posted by: Pooh on September 2, 2003 02:08 AM


The less-looked at jobs statistics mostly confirm, or helghten the impression of, the bad news of the upward trend in the jobless rate. They are a legitimate part of the story. However, Mr Crudele's the treatment of BLS statistical adjustment efforts is probably too flippant. BLS economists and statisticians do some things that don't work very well, it's true. They also do some things that at first glance look wierd but are the best way anybody has so far come up with to deal with a very complex job of estimation. Mr Crudele could have called them up and asked for an inside assessment of how well some of these techniques work - BLS economists know their stuff and know their methods have weaknesses - but instead, he apparently intentionally tried to make the whole employment reporting effort support his "Labor Day" thesis that the labor market stinks. Not really fair to a pretty hard working bunch at BLS.

By the way - that "first new jobs created since last January" business? Big joke. Fewer than normal retail sales jobs were added in November and December for the holidays, so fewer than normal retail sales layoffs were translated into a retail jobs gain by seasonal adjustment factors in January. There was no net job gain in January.

Posted by: K Harris on September 2, 2003 05:35 AM
Post a comment