Stay informed with free updates
Simply sign up to the US economy myFT Digest — delivered directly to your inbox.
My 18-month-old feels his needs intensely, particularly those relating to strawberries. Similarly, investors today really, really want to know whether the US is in a recession, and although last week’s markets tantrum has mercifully ended, there is still some lip wobbling. But whereas I can judge whether my son has eaten enough (“that last one contains a worm”) the economic data is offering no such clarity. And a growing crop of “now-cession” indicators is only adding to the confusion.
Chief among these is the Sahm rule, originally intended by the economist Claudia Sahm to trigger a fiscal stimulus. It draws on the historical regularity that since 1970, every increase exceeding 0.5 percentage points in the three-month average of the unemployment rate relative to its low over the previous 12 months has coincided with a recession. Worryingly, in July this indicator flashed red.
At this point the grown-up thing to do would be to build a more nuanced judgment based on other data points. But investor outbursts are loud and unpleasant, so analysts have been busy working on supplements. One presented by UBS based on the employment-to-population ratio, for example, offers the soothing message that a recession is not yet here.
On August 11, Pascal Michaillat of the University of California, Santa Cruz, and Emmanuel Saez of the University of California, Berkeley, waded in to create the Michez rule (my name, not theirs). This combines a slightly modified Sahm rule with a similar indicator of changing job vacancies, and has tended to detect downturns with impressive speed. Indeed, it suggests the US was in recession as long ago as March.
It is not great that both the Sahm and Michez rules suggest recession is already here. But while history is littered with wishful claims that this time is different (I have learnt to be sceptical of wails suggesting that my son’s tolerance for acidic fruit has risen), these indicators come with real health warnings.
Start with the Sahm rule. As Ernie Tedeschi of Yale University pointed out to me, between January and June of 2024 the rise in unemployment was almost entirely driven by new entrants and re-entrants to the labour force. That smacks of rising unemployment driven by a higher labour supply, not dangerously depressed demand. In July there was also an increase in temporary unemployment, which Tedeschi says “bears its own grain of salt”.
This health warning applies to the Michez rule too, which also relies on changes in the unemployment rate. And although the vacancies data has historically been a useful gauge of the labour market, more recently it might be throwing out false positives, if the recent drop in job openings simply reflects normalisation from an extraordinary peak. Over the past few years it looks as if the relationship between unemployment and vacancies has been a bit off, suggesting that history may be a poor guide to the present.
Overall, this is a pretty uncertain situation. We might be in a recession, though a range of other indicators suggests that is unlikely. What about an indicator to reflect this murkiness?
Michaillat and Saez have had a go, defining two thresholds between which certainty climbs from zero to 100 per cent. The bottom one — passed in March — is the lowest level of their indicator that all recessions since 1960 have cleared. Their upper threshold is the highest level of their indicator that all recessions have surpassed.
They define the chances that we are in a recession as the share of the way the current indicator is between the lower and upper threshold. Based on the latest data, that is 40 per cent.
If you are scratching your head, think of it this way. Suppose my son’s stomach ache has always started after six strawberries, but on every stomach ache occasion he has eaten at least 10. Eight strawberries in, this indicator would suggest that there is a 50 per cent chance that Caspar will get a stomach ache.
If you are still scratching your head, to be honest so am I. Aside from whether the academics are really describing a probability rather than a sort of “confidence index”, this method seems rather sensitive to the chosen timeframe. Had the academics gone back to the 1930s, their upper threshold would have been lower, raising the probability of a current recession to 67 per cent.
The fundamental challenge here is that there have been nine recessions since 1960, not enough to identify patterns that would definitely hold in a freakish post-pandemic period. So with apologies to my son, and investors, some frustration is inevitable.
Follow Soumaya Keynes with myFT and on X
The Economics Show with Soumaya Keynes is a new podcast from the FT bringing listeners a deeper understanding of the most complex global economic issues in easy-to-digest weekly episodes. Listen to new episodes every Monday on Apple, Spotify, Pocket Casts or wherever you get your podcasts
Read the full article here
