Tuesday, September 25

Lies, damned lies, and statistics: Spanish and Greek youth unemployment much lower than reported

One of the most commonly cited Eurozone crisis statistics over the past several years has been youth unemployment, which in hard hit countries such as Spain and Greece has been reported to be as high as 50%.

In a recent post over at Project Syndicate Steven Hill dissects Eurostat's unemployment rate methodology and comes up with markedly different figures:
Unemployment estimates also are surprisingly misleading – a serious problem, considering that, together with GDP indicators, unemployment drives so much economic-policy debate. Outrageously high youth unemployment – supposedly near 50% in Spain and Greece, and more than 20% in the eurozone as a whole – makes headlines daily. But these numbers result from flawed methodology, making the situation appear far worse than it is. 
The problem stems from how unemployment is measured: The adult unemployment rate is calculated by dividing the number of unemployed individuals by all individuals in the labor force. So if the labor force comprises 200 workers, and 20 are unemployed, the unemployment rate is 10%. 
But the millions of young people who attend university or vocational training programs are not considered part of the labor force, because they are neither working nor looking for a job. In calculating youth unemployment, therefore, the same number of unemployed individuals is divided by a much smaller number, to reflect the smaller labor force, which makes the unemployment rate look a lot higher.
So what we have here is a simple division problem: the unemployment numerator is accurate, but the labor force denominator has been fudged.

What are the real youth unemployment figures in countries like Spain and Greece?
The youth unemployment ratio – the number of unemployed youth relative to the total population aged 16-24 – is a far more meaningful indicator than the youth unemployment rate. Eurostat, the European Union’s statistical agency, calculates youth unemployment using both methodologies, but only the flawed indicator is widely reported, despite major discrepancies. For example, Spain’s 48.9% youth unemployment rate implies significantly worse conditions for young people than its 19% youth unemployment ratio. Likewise, Greece’s rate is 49.3%, but its ratio is only 13%. And the eurozone-wide rate of 20.8% far exceeds the 8.7% ratio.
Certainly these much lower youth unemployment figures are still a matter for serious concern. And as Hill notes later in his post it is likely that at least a significant portion of young people who are in school are there because they cannot find work.

There is, however, a substantive difference between the 50% shock headline figures and the real picture of youth unemployment, and this difference may explain why we have not seen a full-on revolution in countries like Greece or Spain (at least not yet).

The final question is why has the media only reported the much larger youth unemployment figures and not the arguably more meaningful, lower youth unemployment ratio? Certainly the larger figure is much more sensational and attention grabbing.

At the risk of sounding conspiratorial, another way of asking this question is who benefits by reporting the larger figure? Undoubtedly larger figures aid the narrative of the pro-bailout and pro-stimulus, anti-austerity contingent. 50% youth unemployment sounds pretty drastic, and drastic times call for drastic measures.

As they say, "never waste a good crisis".


  1. Your "adjusted" ratio effectively lumps discouraged workers, students, people not in the official labor force as productive, which is even more misleading by over inflating the labor force and the unemployment rate within the labor force. If there were 20 stay at home parents and 20 unemployed in a population of 100, why would you include the 20 stay at home parents in the active labor force when calculating the unemployment rate?

  2. Dear Anonymous,

    First off, the adjusted ratio is Eurostat's, not mine. Second, I agree that the adjusted measure is no panacea. See the comments section of the original piece I linked to on Project Syndicate for a discussion of other issues with calculating unemployment (e.g., how do you measure people who have left the country to look for work elsewhere?)