Someone sent me an email this evening with some details on the Paul Krugman response to James Montier which I discussed here. I had previously stated that the Krugman response was lacking meat. But it’s actually worse than that. It’s actually highly misleading and appears intentionally so.
In the post Dr. Krugman tries to show how much interest rates matter by comparing the Fed Funds Rate with Housing Starts. He shows a chart and declares that there appears to be a strong correlation. Except, as this emailer notes, he appears to have shifted the chart to make it appear as though there’s a correlation where there isn’t one.
Here’s the Krugman chart:
And here’s the version that would have originally shown up when the data is pulled from FRED:
See what was done there? The period in the early 1960’s was removed and so was the period from 2000 on. In other words, out of a 55 year time period Dr. Krugman decided to remove 20 years worth of data because it fit his argument better. For those keeping track that’s removing almost 40% of an entire data set just because the data didn’t fit the narrative. And when you add those years back in you get a result that shows a very weak correlation:
I can understand why he might remove the period from 2008 on. But why remove the 1960’s data and the early 2000’s? After all, the 2000’s were the period of Alan Greenspan’s famous “conundrum” where interest rates appeared to have no correlation with the housing market. That’s not just an important part of this discussion, it’s a critical part given that it includes the housing bubble and is outside of the mythical Liquidity Trap era….
This is why people often complain about economics. When economists take a data set and just blatantly alter it to fit their argument it doesn’t do much to help build credibility for their work. Especially when you do it within a post that basically declares economists are smarter than everyone else who says they might not have the whole world figured out.
NB – James Montier wrote an important paper discrediting an important modeling technique used by many economists. And it deserves a better response than “I am super smart and here’s one [altered] chart to prove it”.