Oct. Payrolls: Noise?
Oct. Payrolls: Noise?
October payrolls growth slowed notably due to temporary factors, the unemployment rate stabilized, but the job-finding rate fell. Wage growth remains elevated.
Today’s release of the October Employment Situation report was very much affected by a number of special factors. Payrolls growth decelerated sharply, and as a consequence underlying trends also slowed. The unemployment rate stayed put but the job-finding probability for an unemployed person declined notably. Wage growth remained elevated compared to trends. Given the noise in this report the Fed will largely ignore it although it needs to keep an eye on some tentative signs of changing underlying trends.
Key takeaways:
Payrolls growth slowed notably in October, with trends running below the breakeven pace needed to keep the unemployment rate constant at current levels given higher-than-expected population growth.
The unemployment rate remained 4.1% as the labor force participation rate decreased from 62.7% to 62.6%.
The job-finding rate decreased over the month to 41% with smoothed trends at around 47%. The unemployment rate consistent with recent job-separation and job-finding rates increased, but smoothed trends suggest that in the near term the unemployment rate will likely stabilize around 4.2%.
Wage growth in composition-adjusted average hourly earnings remains elevated at paces inconsistent with both the 2% inflation target and near-term inflation expectations.
Today’s data was too noisy to change the Fed’s outlook on the labor market and recent consumption, inflation and wage data suggests at best a gradual easing of the future policy rate path.
October Jobs Growth: A Noisy Slowing
The October jobs report released today indicated that payrolls in the establishment survey surprised markedly to the downside as they were up by a mere 12,000 persons in October, compared to a 223,000 increase in the preceding month (which was revised down from 254,000 initially). Payrolls growth for August and September combined were revised down by 112,000 persons.
The BLS noted that survey data collection for the Establishment Survey was impacted by hurricane Milton as well as the length of the data collection period1. Also strikes at Boeing played a major role: a main drag was the manufacturing sector (-46,000 persons). Of this decline, the transportation equipment manufacturing excluding motor vehicles payrolls equaled 38,400 persons, similar to the number of striking workers at Boeing.
The unemployment rate remained unchanged at 4.1% in October. In three-digit terms the unemployment rate went up from 4.051% in September to 4.145%. Household employment contracted in October: from +430,000 persons in September to -368,000 persons (chart above). The labor force declined by 220,000 persons, the first labor force decline since May when it decreased by 250,000 persons. The population grew of 209,000 persons in October with the labor force participation rate decreasing from 62.7% in September to 62.6%.
Moving beyond the month-to-month movements, the chart above shows three- and six-month moving averages of payrolls changes from the establishment survey since February 2022. With downward revisions of August and September data smoothed trends in payrolls growth appear to have slowed recently and are now at or close to the breakeven pace needed to keep the unemployment rate constant in a given month (purple line in the above chart). This does not change after naively incorporating the results of the Preliminary Benchmark revision published in August, as I did recently (blue and orange dashed lines in the chart above).
Note that the estimate of the breakeven pace that keeps the unemployment rate constant in a given month is conditional on the BLS’ population growth estimates as well as the labor force participation rate in that month. There’s been a lot of public discourse on the validity of the population growth projections from the U.S. Census that are used by the BLS for its own population estimates. In particular when it comes to net immigration projections the Census has been criticized a lot. For example, the CBO’s population projections assume more aggressive net immigration numbers for 2023 and 2024, resulting in noninstitutionalized population estimates of about 268 million in 2023 (vs about 266 million from Census) and about 271 million in 2024.
The chart also includes an adjusted breakeven payrolls growth pace for a constant unemployment rate, which adjusts the BLS’ population projections for 2023 and 2024 such that they are line with the CBO’s 2023 and 2024 noninstitutionalized population estimates (gray line in the chart above). By comparing the trends in (revised) payrolls to this adjusted breakeven pace it’s clear that job creation in the U.S. currently falls short of keeping up with the higher CBO population growth projections. The CBO-implied breakeven pace is around 197,000 persons in October vs. about 113,000 using the BLS population estimates, and we need to see a sizeable pickup in the payrolls growth over the remainder of the year otherwise the unemployment rate will start to rise at a faster pace again.
Additional details about the underlying strength of the labor market can be inferred from the household employment survey. Following Shimer (AER, 2005) and Shimer (RED, 2012), we can use data on total unemployed and employed persons as well as the number of people that are unemployed for less than 5 weeks to estimate:
Job-finding rate: the probability an unemployed person in month t will find a job or leaves the labor force. This is calculated assuming that total unemployment in month t+1 equals month t unemployment plus the number of people unemployed for less than 5 weeks in month t+1.2
Job-exit rate: the likelihood an employed person in month t will exit its job either due to layoffs, quits or retirement, which depends on data on the job-finding rate, unemployment and labor force.3
The chart above shows a plot for the estimated job-exit rate. This job-exit rate has been relatively stable over the past two years, with a moderate downward shift in the first half of 2023 that stabilized between June and October, but then rose again until the summer. Note, however, that the y-axis in the chart above also makes clear that the variability in the separation rate has been really modest.
Job exits jumped up between June and August and but has been decreasing notably between August and October (chart above). The ratio of job exits is now at the lowest level since early 2024. Given the decline in the labor force in October and historically low layoff rates this is somewhat surprising, and I can only explain it as reflecting the continued decline in the quits rate.
In 2023 the job-finding rate declined a lot after Q1 2023 (chart above) but recovered to about 54% on the month and 52% on a three-month average basis in December of that year. However, in 2024 the job-finding rate declined again and continued to do so until June going into July when it stabilized around a low 43%, followed by an increase to about 52% between July and August. Between August and September, the job-finding rate slightly decreased to about 49%.
In October the total number of unemployed persons increased over the month, but the number of newly unemployed persons (less than 5 weeks in duration) declined: +150,000 vs. -34,000. This suggests that the likelihood to exit unemployment between September and October declined notably, with the job-finding rate decreasing from about 49% to about 41% for September going into October (chart above). Three- and six-month averaged job-finding rates for September into October settled at around 47%.
As in Shimer (AER, 2005) we can combine the above discussed job-exit and job-finding rates to calculate a flow-consistent unemployment rate and the chart above plots both the corresponding monthly rate and the three-month average of this rate. The (three-month average) flow-consistent unemployment rate is the unemployment rate that prevails when the job-exit and job-finding rates remain constant at their current (three-month average) levels. Deviations compared to the official unemployment rate should dissipate over time, and often leads changes in the official rate.
The chart above shows clearly that the improvements in the job-exit and job-finding rates between July and September led to a drop in the monthly flow-consistent unemployment rate to levels not seen since the start of the year. However, given the outsized drop in the job-finding rate between September and October, the flow-consistent unemployment rate jumped up to 4.3%.
Given the choppiness in this measure, it’s probably more useful to look at three-month averages. Since January the flow-consistent unemployment rate on a three-month average basis has been outpacing the official unemployment rate and leading the rise in the latter (solid blue and orange lines in the chart above). This was largely driven by a deterioration in the job finding rate over that period, which reflects labor demand weakening. More recently, and despite this month’s jump in the flow-consistent measure, three-month averages of headline and flow-consistent unemployment rates suggest that over the near term the unemployment rate will likely stabilize in the 4.1%-4.2% range.
A variety of temporary factors related to weather, strikes and survey timing, led to a down payrolls number for October. However, given revisions to earlier months and taking into account higher-than-expected population growth, smoothed payrolls growth trends have slowed and needs to pick up again this year to avoid a faster rise in the unemployment rate. According to the BLS the household survey was not impacted by the above factors but given the sudden decline in the labor force as well as an equally notable drop in the monthly job-finding rate going into October, I cannot help to suspect that in the background these temporary factors played also a role in the household survey. So, we will need to see a substantial recovery in these metrics for November and December to confirm that indeed the labor market has settled into a new steady state and is not further slowing.
Wage Growth Remains Elevated
Average hourly earnings of all private sector employees grew faster fast over the month in October at 0.4% month/month from a downwardly revised 0.3% in September but slowed in year/year terms from (a downwardly revised) 4.4% in September to 3.3%. For production and non-supervisory workers, hourly earnings went up at 0.4% month/month in October, up from (an unrevised) 0.3% in the preceding month, and on a year/year basis growth decelerated from 4.4% to 3.6% in October.
The wage data from the jobs report are notoriously noisy, given that they are revised often and do not correct for the sectoral and skills composition of jobs growth over the month. There are better quality wage data available, such as the Atlanta Fed Wage Growth Tracker and the Employment Cost Index, but the Atlanta Fed does construct a rudimentary composition correction for average hourly earnings from the jobs report, which can be found here.
We can observe from the chart above is that the year/year wage growth rate indeed slowed, despite the composition correction, but on a three- and six-month basis composition-adjusted wage growth have both been accelerating since August.
I can combine labor productivity and labor share trend estimates with the 2% inflation target, along the lines I have done in my “Wages and Inflation Expectations” notes and incorporating the Q2 update of productivity data to get a medium-run annual wage growth rate consistent with 2% inflation. Additionally, instead of the 2% target one can use my “Main Street” year-ahead inflation expectations proxy, which after incorporating September inflation survey data suggests that the trend in near-term inflation expectations eased since Q2 to around 2.5% in year/year PCE inflation terms in September. Incorporating preliminary University of Michigan data as well as Atlanta Fed Business Inflation Expectations data, both for October, led to this common near-term inflation expectations trend to rise to 2.6% in year/year PCE inflation terms in October.
Compared to both the composition-adjusted AHE data for production and non-supervisory workers and the unsmoothed Atlanta Fed wage tracker into August, annual wage growth rates still outpace the 2.5% pace consistent with 2% PCE inflation in the medium term (green line in the above chart). In fact, wage growth also runs above the wage growth pace consistent with “Main Street” near-term inflation expectations (blue line in the chart above).
November Rate Cut
After the conclusion of the September FOMC meeting the Fed clearly signaled that the 50bps rate cut following that meeting was nothing more than a recalibration move, and that it would ease policy rates at more moderate pace beyond that meeting. This was based on the notion that the U.S. economy was on a moderate growth path with a labor market cooling towards a steady state with unemployment rates in the low 4% range and inflation that continues to ease back towards 2% over time. Today’s data was too noisy to corroborate whether the Fed’s view on the labor market, with unemployment rates seemingly stabilizing in the 4%-4.2% range., still holds or not.
I expect a 25bps rate cut at the November FOMC meeting and likely a similar rate cut at the December FOMC meeting. However, employment growth needs to pick up over the remainder of this year in order for it to be in line with the Fed’s own view. Furthermore, with wage growth momentum remaining stubbornly elevated as well as a pick-up in underlying inflation momentum and household spending growth there’s a lot of uncertainty with regards to the policy rate outlook going into 2025.
As per the BLS October jobs report: “A larger influence on the October collection rate for establishment data was the timing and length of the collection period. This period, which can range from 10 to 16 days, lasted 10 days in October and was completed several days before the end of the month.”
Given this calculation, the job-finding rate will run up to September utilizing data on (short-term) unemployment for October.
As the calculation of the job-separation rate depends also on (short-term) unemployment for October, we cannot go beyond September.