The recent dismissal of the Head of the Bureau of Labor Statistics, a key US statistical agency, rightly made headlines. Erika McEntarfer’s termination was triggered by the publication of US job creation figures for July 2025. While July's numbers were not particularly bad, they came with a severe downward revision of the two preceding months. The magnitude of this revision was so significant that the US labour market shifted from job creator to job destroyer in May to June 2025. The Trump administration chose to shoot the messenger, but this is likely a hasty way to address the underlying problem.
The mystery lies in job creation figures differing from the unemployment rate. Job creation figures point to a deteriorating labour market whereas a competing and independently calculated figure for the unemployment rate is not giving the same message, even though both measures attempt to gauge the health of the labour market.
The gap between these two statistical indicators reflects differences in calculation methodologies. At this point in the economic cycle, these methodological differences are incredibly informative about the true state of US employment and have significant implications for future growth, inflation and monetary policy.
This week, Simply put asks whether the US labour market is in good shape or not.
Read more: Why Trump's policies haven't made an economic impact (yet)
Household versus establishment surveys
In the US, labour market health is measured through two surveys: the Establishment Survey (or Nonfarm payroll report) polls approximately 131,000 businesses about the number of people they employ and aggregates their responses. The Household Survey polls a group of 60,000 households directly, asking them if they were employed at the time of the survey.
These surveys may have the same objective, but the results can be very different. There are four main differences:
- Workers with multiple jobs are counted multiple times by the Establishment Survey but only once by the Household Survey
- Undeclared workers are absent from the Establishment Survey but present in the Household Survey
- The Establishment Survey can undergo significant revisions, as smaller businesses may delay their response to the survey, especially during cycle troughs
- A strong influx of legally employed foreign workers will inflate the Establishment Survey more than the Household Survey, due to the lower rotation of the survey sample in the latter.
Thus, when the Establishment Survey shows strong employment growth that is not reflected in the Household Survey, several explanations can emerge. Let’s start with immigration: new immigrants finding work are quickly captured by the business survey, but take longer to be well represented in the household sample. Another explanation is multiple job holdings. Faced with the rising cost of living (inflation), more people take second jobs. The result is that more jobs are created (which is counted in the Establishment Survey) without the number of people working increasing (which is counted in the Household Survey).
Obviously, it is extremely difficult to determine why the two figures differ; these explanatory factors are themselves fluid. Perhaps it is simpler to consider whether this latest revision of the Establishment Survey means that both measures now align?
Two figures, two scenarios
Figure 1 compares the unemployment measurements calculated using each of these surveys. The Household Survey directly presents a ratio between the unemployed and the labour force (the definition of an unemployment rate). For the Establishment Survey, one must be creative and examine the difference between total employment and the labour force – this difference provides another measure of the unemployment rate, once presented as a ratio of the labour force itself.
The graph shows a broadly coherent evolution: both lines generally rise and fall at the same time. However, some differences remain:
- In the 1970s, the Household Survey rose faster than the Establishment Survey during periods of economic deterioration. This was notably the case during the two oil shocks of 1974 and 1979
- We see a similar gap in both 2001 and 2008, with the former outpacing the latter.
The gap between these two statistics could well be informative and linked to major economic shocks. More recently, we notice what most macro analysts already know: until recently, the Establishment Survey presented a more positive view of US employment than the unemployment rate itself, hence the alarm triggered by the rise in the unemployment rate last year, which the Sahm Rule links to recession risk. So, what does this gap between metrics tell us about the current situation in the US?
FIG 1. Unemployment measurements based on the Establishment Survey and the Household Survey1
Real signals and false alarms
Figure 2 compares periods of US recession with the difference between the two unemployment surveys. When this difference is negative, the Household Survey indicates a less positive scenario than the Establishment Survey. As can be seen on the graph, a highly negative difference tends to occur alongside major US recessions. This was notably the case during the double recession of the 1980s, the investment crisis of the early 1990s, as well as during the recessions of 2001 and 2008. Of course, this type of statistic also presents a number of false signals, particularly in 1977 and 1984-1985. More recently, the picture is deteriorating again.
Historically, very few false signals have occurred when the difference between the two data points approaches 1.5%, but future data revisions could still modify the point we are currently observing. Nevertheless, it remains essential to keep in mind that these two surveys can provide very different results, and that this difference can inform us about the risks facing the US economy. For now, it represents a clear red flag that only hard data will confirm or refute.
FIG 2. Gap between the Establishment Survey and the Household Survey relative to US recessions2
Read more: US economy: peak uncertainty as growth indicators misfire
The investment implications for multi-asset investors
Similar to recent weeks, while the overall scenario for the equity market remains very positive, our signals still call for moderation. Trends are indeed positive for equities, but weaker regarding duration. Our risk model continues to perceive volatility in equities close to long-term averages, while the slow normalisation of bond market volatility is ongoing. Finally, our risk appetite indicator has recently shown signs of a jolt, suggesting that market sentiment could reverse quickly.
Simply put, measures of US unemployment are delivering different messages. This divergence could point to a deterioration in the labour market.
To learn more about our All Roads multi-asset strategy, click here. Macro/nowcasting corner
The most recent evolution of our proprietary nowcasting indicators for global growth, global inflation surprises, and global monetary policy surprises is designed to track the recent progression of macroeconomic factors driving the markets.
Our nowcasting indicators currently show:
- Our world growth nowcaster declined this week, mainly due to a decrease in the Chinese indicator. The decline came from different sectors
- Our inflation signals remained flat in the high but declining regime
- Signals from our monetary policy nowcaster decreased and the China indicator passed below the 50% threshold.
World growth nowcaster: long-term (left) and recent evolution (right)
World inflation nowcaster: long-term (left) and recent evolution (right)
World monetary policy nowcaster: long-term (left) and recent evolution (right)
Reading note: LOIM’s nowcasting indicator gather economic indicators in a point-in-time manner in order to measure the likelihood of a given macro risk – growth, inflation surprises and monetary policy surprises. The nowcaster varies between 0% (low growth, low inflation surprises and dovish monetary policy) and 100% (the high growth, high inflation surprises and hawkish monetary policy).
