Bluematter.

Via Greg Mankiw:


Be very careful how you interpret this graph. In fact, it's better to try to forget it immediately and resist the urge to draw any inferences whatsoever.

The level of unemployment benefit affects three important variables: the government's fiscal position, incentives to work and hardship endured by the unemployed.

On the first of these, it is clear that the data reveal very little, at least in the absence of information on unemployment and a note on how they were derived. (Is 'averaged across a range of wages, family types and lengths of unemployment' self-explanatory to you? It's not to me)

With regards to incentives to work and hardship endured by the unemployed, the figures again reveal hardly anything. Greece is near the top of the table above, yet it enjoys strong family bonds - social norms dictate that the unemployed are handsomely supported. Also, how much of a loss in consumption does unemployment in each different country entails? If high quality health care, education and entertainment is provided by the state free of charge to both unemployed and employed, is the loss of employment income so important? For that matter, does the figure refer to pre- or after-tax pay?

No piece of information allows us to draw any inference without first assuming some factors or relationships remain constant. The challenge when faced with data of any type is to decide whether any such assumptions can reasonably be made: in the case of the figure above, the answer is no.

Of course, numerous other seemingly relevant facts and statistics are cited every day both in print and in conversation. My advice? Release a sigh of despair, mumble some French, and move on.