Signs of weakness in the US labor market have sparked concerns about the economy being in or headed for a downturn.

However, the Sahm Rule is not really a ‘rule’ in the sense that the economy must be in recession when unemployment rises in this manner. Instead, it is an empirical regularity, which holds in many, but not necessarily all, cases. For example, it applies way less often outside of the US.

There are good reasons to believe the Sahm Rule might not indicate a recession in this cycle. First, rising unemployment due to strong labor supply growth might not provide the same negative signal about the health of the economy as an increase driven by layoffs.

To illustrate, across the last four downturns excluding COVID (2008, 2001, 1990, and 1980), employment growth in the household survey fell over the first six months of recession. By contrast, employment is up over the past six months, albeit modestly and not sufficient to keep up with a growing labor force (Chart 1).

1. How foolproof is the Sahm Rule?

The Sahm Rule finds that a 0.5 ppt increase in the unemployment rate from its low over the past year has almost always been associated with the early stages of recession.

The recent rise in US unemployment triggered this rule in July, with the unemployment rate increasing to 4.3%. Judging by history alone, this is an ominous sign. Indeed, we found that the Sahm Rule has a near-perfect record in identifying post-war recessions, sounding a false alarm only once over this period (Chart 2).

Using the employment-to-population ratio rather than the unemployment rate to construct an alternative Sahm Rule works nearly as well as the traditional measure in identifying recessions and stripping out the impact of changing labor force participation.

The employment-to-population ratio is currently down 0.3 ppts from its recent high, indicating some deterioration in the labor market. Crucially, it is not yet consistent with recession (Chart 3).

Second, there is a greater risk of mismeasurement of the unemployment rate following the huge increase in immigration into the US in recent years.

Recent research estimates that the household survey is undercounting the US population by about 3 million people, or around 1%. This blind spot might sound modest but it’s large enough to create some error in the unemployment rate should recent arrivals work more or less than the broader population. This somewhat lowers our confidence in signals from the household survey.

Bringing the pieces together, we are not arguing that the Sahm Rule should be discounted entirely. At the very least, the increase in unemployment is concerning and solidifies our view that the labor market is slowing. However, this rule could be broken in this already atypical business cycle.

2. What does other data say?

The Sahm Rule may be flashing red, but there are relatively few other data consistent with recession right now.

The National Bureau of Economic Research (NBER) formally defines when a recession starts, using various indicators to pick up a significant decline in activity across the economy lasting more than a few months.

The six indicators it uses suggest that the depth, diffusion, and duration criteria are close to being met (Chart 4). Few signs of distress are present in them.

Moreover, a broader range of hard and soft data show signs of slowing as opposed to distress.

In the labor market, initial unemployment insurance claims have risen slightly, but not yet in a manner consistent with a downturn.

There are pockets of weakness in the manufacturing sector – with the manufacturing ISM soft at 46.8 – and the housing sector – where new starts are down 16% over the past year – which suggests some bite from high interest rates on rate-sensitive sectors.

However, the broader services sector looks healthier, with headline PMI/ISM surveys running around 55, helped by a still-solid trend in consumer spending.

It is possible that the data will be revised in the future. Indeed, large revisions are common around cyclical turning points, which explains why recessions can be hard to spot in real time.

Additionally, while the NBER indicators tell us if an economy is already in recession, they are less useful as leading indicators as they can deteriorate suddenly when a downturn begins (Chart 5).

Still, the broad suite of evidence should make us more confident a downturn has not started, even if some caution is warranted.

3. Does the recent tightening in financial conditions risk becoming self-fulfilling?

Market volatility spiked in early August, following the equity selloff in response to weak US data, amplified by a rapid unwind in the Japanese yen carry trade. The VIX jumped to its highest reading since the peak of the pandemic. Other measures of financial stress rose, too, but only reached the highs seen during selloffs in 2022.

Against this backdrop, the market was briefly pricing a series of near-term emergency-style 50 bps rate cuts from the Fed. However, this pricing was insufficient to offset market fears that the economy could be headed for recession.

Since then, more reassuring US data and a stable yen have helped to unwind much of this stress (Chart 6), with major equity indices now trading close to where they sat before the July payrolls report.

This should lower the risk that the financial stress shock itself undermines the cycle by causing companies and households to rein in spending in the face of falling equity prices, negative wealth effects, widening credit spreads, and tighter credit standards.

At the same time, pricing for Fed easing has moderated, although investors still expect close to 100 bps of cuts this year, up from the 70 bps expected before the July payrolls report. Moreover, the pricing for the terminal rate continues to fall. At 3.2%, it is around 40 bps lower than in July and 100 bps down from the 2024 peak when markets were concerned about a “no landing” scenario.

With clearer evidence that policy is tight, markets anticipate that the Fed will need to dial back its restrictiveness rapidly. To the extent that it is interpreted as pro-active easing, delivered ahead of a major deterioration in activity, it can support both bonds and equities.

By contrast, if the Fed is perceived as being behind the curve and delivering reactive cuts to try to catch up with the economy's slowdown, this is likely to result in a more painful environment for financial conditions.

We expect a sequence of 25 bps rate cuts at every meeting between now and the middle of next year. However, should data continue to deteriorate, the Fed will need to accelerate to 50 bps steps.

4. What do our recession risk models say?

On balance, our reading of the broad suite of US data and financial market dynamics suggests that the economy is slowing but not heading for imminent recession.

When we cross-check this judgment with our recession models to get a more quantitative assessment, we find that they point to a low risk that the economy is heading into recession over the next three months. Models that look a little further into the future show a slight increase in recession risk. The latest read from our 12-month model sits at 27% (Chart 7).

However, in absolute terms, these probabilities are not high or consistent with recession becoming a base case. And it is worth noting that the models are actually showing a lower risk of downturn than had been the case over recent years.

Our longer-term models, which look 24 months out, are ostensibly more alarming, putting the chance of recession at 64%.

However, these lean heavily on macroeconomic imbalances, such as the output gap, which we think can continue to unwind benignly. Indeed, these models also show a lower probability of a downturn than was the case in 2022 and 2023, consistent with the progress that has been made in addressing imbalances and bringing down inflation.

Overall, our quantitative indicators align quite well with our judgment at the moment. The US economy is not in a recession right now, and while risks are more elevated than usual, these have not reached levels where we should shift our base case forecast to incorporate a downturn.

5. What probability do we attach to a recession?

A US soft landing remains our base case. But clearer signs of weakening activity have raised the risk of a downturn. As such, we have made several changes to our global scenario distribution.

First, we have increased the probability of a “rolling end cycle” in which the US and other economies have slid into recession to 20% from 15%.

Second, we have raised the probability of our base case of a global “soft landing” to 35% from 30% previously. This is because signs of cooling growth (and inflation) open the door for the rate cuts in the US and elsewhere, ultimately necessary to deliver a soft landing.

Finally, we have lowered the probability of “no landing” to 10% from 20% previously. The risk of a strong growth and inflation backdrop forcing the Fed to hike again has receded as inflation and activity data have cooled.

The remaining probability mass is split between our other global scenarios – including a trade war, further escalation in the Middle East, and positive developments on the supply side, including from AI. But, on balance, these changes push the distribution of risks towards lower growth, inflation, and interest rates.

Important information

Projections are offered as opinion and are not reflective of potential performance. Projections are not guaranteed and actual events or results may differ materially.

AA-230824-182189-1