The Employed-to-Unemployed Ratio: A Brief Clarification

Earlier today I looked at the history of what appears to be another useful metric for monitoring the business cycle: the ratio of nonfarm payrolls to the unemployment rate in terms of its 12-month percent change. I attributed the “discovery” of this indicator to Bob Dieli, an economist who crunches the numbers at But after reading the post, Dieli alerts me that I was mistaken and that he was in fact watching was the ratio of the employed to the unemployed–both in terms of the actual numbers reported each month in the household survey of the employment report. On a 12-month rolling basis, the two signals are virtually identical. Nonetheless, Dieli’s clarification is worth noting because it moves us closer to an apples-to-apples comparison in terms of the underlying data.

But first, let’s look at how the employment/unemployment ratio looks through history. (For those familiar with graphing via the St. Louis Fed’s FRED database, the relevant tickers: CE160V/UNEMPLOY). Here’s how it shakes out for a rolling 1-year percentage change:

Like the ratio I discussed earlier, the recession-warning signals in the ratio above are closely linked with those periods when the ratio falls below zero. It’s not flawless—the indicator suffers several false positives in its history. But the fact that this indicator has a strong history of dispatching early warnings of an approaching recession (or advising us that a downturn has recently started) suggests that we should keep an eye on it. As always, it’s best to look at this ratio (and every other metric) in context with a range of business cycle signals.
What’s the intuition behind this indicator? Simple, really. Imagine that the number of employed workers is rising while the number of unemployed is constant. In that case, the ratio is rising, reflecting the improving conditions in the labor market. In contrast, a decline in the number of employed vs. a constant number of unemployed translates into a declining ratio. The same principle applies if we keep the number of employed constant and compare it with rising or falling numbers of the unemployed. In the worst case scenario, the number of employed people is falling and the total of unemployed is rising. The bottom line: when the labor market is deteriorating, it’s likely to show up in this ratio. When the deterioration is severe enough so that the ratio is retreating on a year-over-year basis, recession risk is above average.
Finally, both the employed and unemployed estimates are found in the household survey of the monthly employment report, and so we’re drawing from the same sample data. In that sense, the design of the ratio is superior to the version I outlined previously. In practice, however, both are in agreement in terms of the timing of their signals through history. Both also tell us that recession risk was low last month.
You still can’t get blood out of a stone, or find flawless forecasts of the future in this ratio. But as one more metric for analyzing the business cycle, this one looks pretty good.