Tuesday, November 19, 2013

A numbers puzzle

In my macro classes I like to have the students study a chart of the "output gap." That's the difference between the actual GDP (what the economy did in any given quarter) and the potential GDP (the standard estimate of what the economy "should have" done in that same quarter).

A positive output gap shows an economy in some sense out-performing (and maybe running a risk of inflation), while a negative output gap is associated with high unemployment.

If you take the output gap and divide it by potential GDP, you get the gap in percentage terms, where you can meaningfully compare the economy's performance across different decades.

I've come to rely on FRED, the very useful data tool from the St. Louis branch of the Federal Reserve. They have a vast number of data series, and a fairly flexible charting tool that lets you perform calculations on the data you've chosen and get a visual representation of the results.

This past weekend I went to make a chart of the output gap, and got this (I've added the red line at 0% to make it easier to see when the gap is positive and when it's negative):

And something struck me as not right. Because I'd made the same chart sometime in 2012 (I think in the fall), and it looked like this (again, the red line at 0% is my addition):

Either the numbers they're providing for actual GDP have gone up since last year, or the numbers they're providing for potential GDP have gone down--I didn't save the numbers behind last year's chart just the chart itself, so I don't know which of those possibilities is true, but clearly at least one of them has to be.

I've wracked my brains trying to figure out why this could be, and I can't come up with anything. It's not about a change in inflation adjustments, since both series are in nominal terms. And while the Congressional Budget Office does make some revisions to past estimates of potential GDP, they tend to be modest revisions covering only the last few years, whereas these two charts show changes from beginning to end.

I'm simply perplexed.


  1. Um. Isn't the scale on the left side different? Or is there something else going on?

    1. The scale is different--that's why I highlighted the zero line, so that the change in the data is easier to see.

      The FRED charting tool sets the scale automatically to cover the data being represented. If there is a tool to set the scale manually, I don't know where it is. And even if I did, I can't go back and redraw the older one, and if I were to set the scale on the new one to be the same as on the older one, some of the data would be literally off the chart (though it would do a nice job of making the point visually).