Friday, October 3, 2025

Alternate sources of employment data point to job losses, steady to higher unemployment rate in September

 

 - by New Deal democrat


As we all know by now, there is no US jobs report today due to the government shutdown. But there are a number of alternative measures which can give us a good estimate of what the jobs situation is.

Let me start first with the unemployment rate. Here is an update through last week of initial and continuing jobless claims, which lead the unemployment rate:



With the exception of late July and August, jobless claims have been consistently forecasting an increase of 5% (note: a percent of a percent) in the unemployment rate. Since last autumn the unemployment rate was 4.1% or 4.2%:



this forecasts a rate of 4.3% or 4.4% for September.


Which, as it happens, is exactly the mean forecast of the Chicago Fed’s new unemployment rate forecasting tool:


In contrast, the Challenger layoffs report indicated a lower number of newly jobless in September than in previous months, although the 2025 average remains elevated over 2024:



Now let’s turn to the jobs number itself.

The ADP employment report deservedly got a lot of attention several days ago when if indicated a loss of -32,000 jobs in September. Including revisions, the ADP report generally tracks close to the jobs report. Additionally, it is noteworthy that the ADP report already incorporates the preliminary benchmark revisions based on the QCEW, that won’t be included in the official jobs data until next February.

Justin Wolfers points out that the initial, unrevised ADP numbers do not correlate so well with the official jobs report:


But aside from early 2023, the three month average of the two series remains pretty close. And in the last three months, the ADP’s unrevised reports have averaged 42,000 jobs gained per month, compared with the 50,000 average for July and August in the official report. As revised for July and August - which Wolfers acknowledges is closer to the official BLS numbers, the ADP’s two month average was also 50,000, and its 3 month average including September is 23,000.

In short, the ADP does forecast very little if any jobs growth  at best “officially” in September.

Next, let’s look at the ISM manufacturing and services reports employment subindexes. To cut to the chase, here’s the graph for both covering the past three years (via TradingEconomics):



Since February, only in May has the employment weighted average of the two surveys (86% services, 14% goods producing), indicated growth. In September, manufacturing came in at 46.3 and services at 47.2, averaging 47.1, which is still better than the low average of 46.0.

Finally, let’s look at the employment components of the regional Fed manufacturing and services reports. There is very little graphic information available, but here is the average of the Empire State and Philly manufacturing numbers:


Here is Richmond’s:



And here is Texas’s:


The Kansas City Fed does not publish a graph for employment. But note that among the four that do, all trended down, 3 were negative, and 2 were close to or at their worst levels.

So the best way I have to present this to you is showing your the month over month change, and the actual number for September:

Manufacturing:
Empire: -5.6 to -1.2
Philly: -0.3 to +5.6
Richmond: -4 to -15
KC: +7 to +7
Texas: -12.2 to -3.4
AVERAGE: -3 to -1

Services:
NY: n/a
Philly: +13.2 to 9.4
Richmond: +1 to 0
KC: -14 to -12
Texas: -4.8 to -2
AVERAGE: -4.6 to -2

In sum, the alternative data sources all point to another weak, and likely contractionary, month for September. The unemployment rate likely held steady or ticked up 0.1%, and all three alternative sources for jobs information - ADP, ISM, and the regional Feds - indicate an actual decline in jobs.

Finally, let’s put this in some context for the economy. Here are the official jobs numbers, and real personal income excluding government transfer payments, both normed to 100 as of April:



Real income is flat to down, while jobs have increased by at best 0.1% in the 5 months since. This points to the economy being kept from recession by consumer and business spending, concentrated in AI-related industries and stock market wealth created by that boom (or, perhaps, bubble).