In Praise of Automatic Stabilizers
Hosted from Archives: The "Cognitive Elite"

Department of "Huh?": Unemployment and GDP

I don't grok Paul Krugman completely this morning. He writes:

Growth and unemployment: What you see is that unemployment tends to fall when growth is high, fall rise when it’s low or negative. You also see that growth has to be fairly fast — more than 2 percent [per year] — just to keep the unemployment rate from rising. Why? Well, productivity is rising, so that you can produce any given level of output with fewer workers; so output has to rise to keep employment from falling. And the working-age population is growing, so you need positive employment growth just to keep unemployment from rising.

That’s how you can have a technical recovery that feels like a recession: real GDP may be rising, but if it doesn’t rise at a sufficiently high rate, unemployment keeps going up....

Has the rise in unemployment been more than we should have expected, given the slump in GDP? Looking at the data, I don’t think there’s a strong case — not so much because of the numerical exercise I did in the earlier post, but because this recession is just completely on a different scale from anything we’ve seen for a long, long time. It’s hard to know what we should have expected.

And one last point: when I say that the rise in unemployment makes sense, I mean that it makes sense given the depth of the recession. It is, of course, completely crazy and disastrous from a larger point of view.


I get the point that we are at the edge of the scatter right now, and hence it is "hard to know what we should have expected." But we did expect the relationship to be roughly linear--and it does look like we are high on the change-in-unemployment by an amount that is certainly economically if not statistically significant.

Should we have expected a nonlinearity in Okun's Law to emerge at the left end? I know that I didn't. Would I have if I were not a bear of very little brain? I'll have to think about this for a while...

Update: Gah! The horizontal axis is not the 8 quarter growth in GDP but the average annual growth rate over those 8 quarters...