July 26, 2012
Another Business Cycle Indicator Worth Watching
Bob Dieli of NoSpinForecast.com emailed me an intriguing chart earlier this week: the ratio of nonfarm payrolls to the unemployment rate. The ratio, which is based on monthly data, has an especially intriguing relationship with the business cycle, namely: it peaks just ahead of recessions.
Here’s how the ratio plays out over the past 60-plus years:
The trouble with using the ratio as an objective signal for estimating recession risk is that the peak's numerical value fluctuates through time and so it’s unclear how to interpret the latest data point. But that’s a minor problem. Transforming the ratio into year-over-year percentage changes solves the interpretation issue, as the second chart below shows:
Note that the annual percentage change falls below zero either a) just before the onset of a new recession, or b) in the early stages of a downturn. Alas, it’s not perfect—nothing is. But the false positives are relatively rare. The last one arrived in 1996, when the ratio inched ever so slightly into negative territory for a couple of months without a subsequent recession. On the other hand, the ratio hasn’t missed an NBER recession yet.
It's worth emphasizing--again--that perfection doesn’t exist when it comes to looking for a lone indicator for clues about estimating recession risk. Even an intelligently designed combination of predictors, which is essential, won’t be infallible, in part because no one really knows what the optimal mix of indicators will be, or if it's stable or an evolving pool. And then there's the gray area of deciding how to weight the indicators. Oh, well. That’s the nature of macro, and so all the usual caveats apply with this ratio.
That said, it looks like a worthy addition to the quantitative arsenal for evaluating business cycle risk—assuming, of course, it’s used in context with a range of other indicators--along these lines, for instance.
I can’t say that no one’s ever used this ratio before, although I don’t recall reading about it. No doubt I’m not as well-read as I thought on matters of the business cycle literature. In any case, I’m labeling the ratio the Dieli indicator from here on out (thanks, Bob).
Oh, yes, one more thing: the Dieli indicator tells us that the risk of an imminent recession, as defined by NBER, appears relatively low as of June. If it’s wrong, and suffering from a massive, unprecedented breakdown, we’ll soon see the evidence. Never say never in macro. Meantime, cautious optimism has a new statistical friend.
Update: See this post for a clarification on the ratio.
Posted by jp at July 26, 2012 7:46 AM
PAYEMS is a mirage. This is just 1 / UNRATE. The result simply says, recessions are preceded by increases in unemployment.
Posted by: wcw at July 26, 2012 2:17 PM