> Sahm Recession Indicator signals the start of a recession when the three-month moving average of the national unemployment rate (U3) rises by 0.50 percentage points or more relative to the minimum of the three-month averages from the previous 12 months.

Why 0.5 and not 0.4? Why three-month averages and not four? There are some free parameters that have clearly been fit to make the graph look good historically, and having been proposed in 2019, it has successfully identified one recession, which literally any other indicator also would have identified. Frankly there's a very high likelihood that it's over-fit. I'm guessing after it fails to adequately identify the start of the next recession, economists will want to replace it with one that was made 1 year before that recession, being over-fit even more. Note that these gray regions have clearly been post-hoc altered: the chart pretends that the Sahm Rule would have identified the 2008 recession as starting in Dec 2007, but that's not +0.5 over the min of the last 12 months; that doesn't happen until April 2008. Why did they move the gray box back 4 months? They're basically lying about what their metric would have shown at that time.

I was going to say it successfully "predicted" one recession, but: did it? In 2020, the U3 was at 0.0 in Feb, 0.3 in March, and 4.0(!) in April. You just said that based on this indicator, we are "nowhere close" to a recession. You could easily have said that in Feb or even March of 2020, so "nowhere close" could be as little as one month, or the smallest unit of time this metric considers.

There is no way to say whether we are close to a recession or not using this metric. The absolute value of the y-axis has nothing to do with it: the 1981 recession is identified as starting from a trough of -0.13. The current trend has nothing to do with it: the 1981 recession started immediately after a huge downward trend. A long period of stability ~0% doesn't mean anything (2001, 2020). A period of volatility doesn't mean anything (~1963, ~1986 -- volatility with no recession). All that matters is what the rate of increase is just about to be. So it doesn't make sense to say we are "nowhere close" -- the current value and recent history of this number seem to have almost no bearing whatsoever on when unemployment suddenly spikes. The spikes come out of nowhere. This is not a leading indicator at all: the best it can do is look at a terrible jobs report and say "looks like the recession has already started boys", which is what everyone else would say too if unemployment spikes by 2% in one month. It gives absolutely no information about whether one is about to start next month or not.