Sunday, September 4, 2016

WSJ on natural rate of interest, economists fail at stats edition


WSJ - think you know the natural rate of interest? This part here stuns me:

Economists — and investors — tend to ignore the level of confidence in a calculation. (If you doubt that, look at the attention paid to small beats or misses of expectations for the nonfarm payrolls report, which has a margin of error of plus or minus 100,000 jobs, with 95% confidence.) Holston, Laubach and Williams helpfully published the margin of error around their estimates, and it is big enough to drive a truckload of economists through.

The error margins they produced allow a 95% confidence interval to be calculated, and for some regions it is just silly: They are 95% sure that the natural rate of interest in the eurozone is currently somewhere between plus 12% and minus 12%. Frankly, I’m 100% sure the natural rate sits in a much narrower band than that, without even picking up a calculator.

I think we can take that confidence interval as an indication that their "natural rate of interest" is the deranged fantasy of someone who fell off a bike at high speed while not wearing a helmet.

Seriously, this right here is why I'm taking extra stats classes beyond what is required for an economics BA. I could probably do one paper a year just on bad statistics.


No comments:

Post a Comment