The concept of "finding a needle in a haystack" is often used to illustrate the difficulty of finding an object among many other, somewhat similar objects. Of course, anyone really wanting to hide one particular needle successfully will hide it among a lot of other needles. That metaphor comes to mind after reading a recent report of how Sweden may be masking its true rate of unemployment: by putting the unemployed into benefit classifications other than unemployment itself. (HT: Mahalalnobis.)

The report was written by former LO trade union economist Jan Edling, who resigned from the union after claiming that LO's leadership was trying to block publication of his work. Edling's paper has since been published online by the Swedish free-market think-tank Timbro (currently available on its Swedish language website as a 1.8MB PDF document.)

The English-language Swedish newspaper The Local describes how the unemployment numbers may be being manipulated:

Five percent of Swedes of working age are currently classed as unemployed. A further three percent are occupied in state-organised job schemes. The controversy surrounding Edling's report centres on his interpretation of the 700,000 Swedes who are either on long-term sick leave or in early retirement. Edling asks how many of these people are in fact unemployed.

Edling focuses on certain regions with high numbers of people on sick leave and in early retirement, and concludes that the real reason that people are not in work is that there are no jobs.

Edling's measure of the difference between Sweden's official estimate of its rate of unemployment and his own estimate is dramatic. Mahalanobis offers an excerpt from the Wall Street Journal ( Subscription required, links and emphasis mine):

That Sweden has far more people out of work than detailed in the official 5.5% unemployment rate isn't totally new. Beyond the official rate, an additional 4.4% of the working-age population are parked in the government's elaborate array of job-creation and training programs, according to a study by Skandinaviska Enskilda Banken AB with data from Statistics Sweden. But Mr. Edling calculates that another 10% of working-age people can be identified as unemployed, using correlations between unemployment, long-term sickness and early retirement among Sweden's municipalities and regions. This makes the actual unemployment rate closer to 20% of the work force.

If correct, how Sweden reports its unemployment rate could make for an entirely new chapter in the book " How to Lie with Statistics." (If you're not already familiar with this book, I highly recommend it!) At best, this kind of tactic represents a political calculation destined to backfire. The only real question is how long can knowledge of the magnitude of the problem be suppressed before it erupts into public view?

- posted by Ironman at 5:30 AM | Permalink | |