Is Short-Term Unemployment a Better Predictor of Inflation?

Apr 8, 2014Mike Konczal

Alan B. Krueger, Judd Cramer, and David Cho of Princeton recently released a Brookings paper on the state of the labor market titled "Are the Long-Term Unemployed on the Margins of the Labor Market?" Their big headline result is that the long-term unemployed are going to have trouble finding steady work, both as a historical matter and from what we've seen in the Great Recession. It's fascinating work we'll revisit here.

But what does that mean for the job market right now, with its mix of short-term and long-term unemployed? The second takeaway is that if we only look at short-term unemployment, the economy makes more sense than if we look at total unemployment. As Tim Hartford wrote, this research shows that if "we replotted the Phillips curve['s mix of inflation and unemployment]... using statistics on short-term unemployment... it turns out that the old statistical relationships would work just fine." Some are arguing that we should just focus on short-term unemployment for the moment as an indicator of how the economy is doing.

Is that the case? Not really. We should be careful with this argument now, because this is really a matter of 2009-2012. Back then, the question was why inflation was as steady as it was given very high unemployment. In 2014 the question is very different: why is inflation so low given high unemployment and the relationship of the past several years? We need to explain a different problem.

Let's look at a key chart from the Krueger paper (green boxes my addition):

This is the change in core inflation versus unemployment. (There's a similar dynamic with wage inflation in a different chart.) The left graphic is the change in core inflation versus overall unemployment, and the right graphic is the change versus short-term unemployment. As the paper's authors argue, it's a much tighter relationship if you just look at short-term unemployment. But there are three things to note here.

First, as flagged in the green box in the left graphic, the outliers are the years 2009-2012. Looking at their wage inflation version of this in particular, the authors note that they get a higher R-squared and better predictive value using short-term unemployment. But replicating this chart (data), if you simply take out 2009-2011, you also end up with the higher R-squared and better predictive value.

More importantly, as a second matter look at where we are now via the 2013 data point. The total unemployment number for 2013 is right on the line in the left graph. However, as we can see from the green circle on the right, using short-term unemployment shows inflation much lower than anticipated. This is not surprising; one of the more important economic stories of 2013 was the collapse of inflation. Note that if the labor market were actually getting much tighter, inflation should have been increasing during this time period. More broadly, if the problem were the preponderance of long-term unemployed in the general labor market, we wouldn't expect 2013 to go into freefall and hop over the trendline as it did.

I'm very interested in why we didn't collapse into deflation from 2009 to 2011. I imagine the Fed has something to do with it. But as a third point I'd be a little cautious about using just short-term unemployment during that time as an important indicator about the labor market, as job separations collapsed during the crisis. A low short-term unemployment rate reflects people simply not leaving their jobs more than it reflects the idea that the economy was doing better than we'd expect.

But this question is also a historical one. Krueger and his co-authors acknowledge this, using phrasing like "since 2009" as the basis of their paper. But other people might not catch this, and assume that the short-term unemployment rate is crucial for right now. But that doesn't reflect our current situation of low inflation, a falling rate of long-term unemployment, and an unemployment rate that is going to be stuck in the mid-6% range for some time. We shouldn't use a way of adjusting data to examine what was going on in 2010 to argue there's less slack than there actually is out here in 2014.

Follow or contact the Rortybomb blog:

  

Share This