Market Pulse: April 1, 2024

We are, I think, at a juncture where the nuances of the labor market hold more interest for the Fed as they strive to make their next move, the correct move; the nuances are worthy of close inspection.

For instance, an important component of the Bureau of Labor Statistics (BLS) Establishment Survey invites skepticism.

In the last year Government Employment is up 629k, about 23% of the total increase in Nonfarm Payrolls (NFP). Government jobs make up only 14% of NFP, so in the last year it is hitting way above its weight. Job creation for federal, state and local governments are all in a bull market for the last year, at least. But in any case, in the last 12 months, the growth rate for Government jobs is about fifty percent faster than the growth rate for all other Payrolls, not counting government.

Take Government payrolls out of the equation and in the last 12-months NFP is up about 2.1 million. During the last year the Birth/Death (B/D) is +1.260 million, or more than half the total for NFP – Government jobs/ (Government is not a part of the B/D).

B/D, is purely a mathematical formula which estimates the number of jobs from opening business and the number of job losses from closing businesses. As you can see on the chart below, the 12-month average of the B/D has never been higher than it has in the last couple years, excluding the pandemic spike.

Business Employment Dynamics (BED) is a quarterly BLS report, it comes out at least a half year after the end of the quarter on which it is reporting, and that gives them time to get it right, I would assume. Among the things that BED tracks is the number of jobs from opening businesses and the number of jobs from closing businesses. Sound familiar?

The BED is current through Q2 2023.

In the year from Q3 2022 through Q2 2023 the B/D accounted for 1.296 million jobs. During that same time, solving for the same group of workers, the BED accounted for 521k jobs. So, the B/D figured there were 775k more jobs created from opening and closing businesses, than a different BLS report that takes six months to get it right.

So, if we give BED all the credit and reduce the B/D calculation accordingly, then the payrolls for that year, from Q3 ’22 through Q2 ’23, gets reduced from 2.748 million down to 1.973 million. Using that math, the 12-month average for total NFP for that period of time goes from 229k down to 164k.

Not bad, but less good.

The BLS Household Survey tracks the number of Employed people. If you have two jobs, you are still just one Employed person.

The BLS Establishment Survey tracks the number of Nonfarm Payrolls. Key to note the title excludes jobs that are on farms.

Nonfarm Payrolls tracks all jobs minus farm jobs. Employed counts all the employees. Because of this construct, the Employed figure has always been higher than the payrolls figure. But the number of farm workers has fallen significantly in the last 75 years, by ten million or so. Therefore, the spread between the number of Employed people and the number of jobs, excluding farm jobs, has narrowed over time. See the chart below.

While the narrowing trend is clear there has been some volatility to it over the years. Recently, after the pandemic spike and bounce, the spread sharply narrowed. The latest reading puts Employed at 3.16 million greater than Nonfarm Payrolls, not far off the narrowest reading, hit during the pandemic abyss.

In the last twelve months the number of Employed, individual workers, increased 667k. However, Payrolls, the number of jobs, were up 2.748 million. The 12-month moving averages are +56k and +229k, respectively. The spread between Employed and Payrolls narrowed by about 2.08 million since early last year. Why is that?

It has been suggested that the recent discrepancy in the separate measures of workers and jobs is the result of undocumented immigrants finding work.

So, as it is told, the new employees add to the Nonfarm Payrolls totals, which are reported by those who manage the establishments. Some of the new employees may be undocumented. The estimates of the number of undocumented people entering the US the country in recent years vary greatly; two to five million increase in the last couple of years is a loose range. I haven’t counted.

It is reasonable to assume that undocumented people are less likely to answer a survey conducted by a government representative from the Household Survey, fearing that responding to such a survey puts their situation/home/families, in jeopardy.

Neither survey is designed to identify the legal status of a worker, but the BLS readily admits that “it is likely that both surveys include at least some undocumented immigrants.” But of course, this dynamic suggests the Establishment Survey would be more reliable in its task of counting the number of jobs than the Household Survey would be in its task of counting the number of Employed, because of undocumented new arrivals finding a job.

Therefore, according to this line of thinking, the stronger Establishment Survey is more indicative of the state of the labor market than the lagging Household Survey.

In the chart below, Immigrant is defined as: The term "immigrants" (also known as the foreign born) refers to people residing in the United States who were not U.S. citizens at birth. This population includes naturalized citizens, lawful permanent residents (LPRs), certain legal nonimmigrants (e.g., persons on student or work visas), those admitted under refugee or asylee status, and persons illegally residing in the United States.

(Note: the mid-century dip in immigration numbers. It is generally bounded by the exclusionary/racist Johnson-Reed Act, the immigration legislation of 1924 that narrowly defined who could and who could not emigrate to the US, and the Hart-Celler Act of 1965 that abolished the national origins quotas of the 1920s.)

The surge in immigration from 1970 forward slowed in recent years, blame the pandemic. As you can see by referring back to the Household/Establishment chart above there is no predictable rhyming pattern to the growth of immigration and the direction of the spread.

Establishing an exact number of unauthorized immigrants living in the US is an inexact science. A recent study by Pew Research covers the story from 1990 to the mid-pandemic period. This accounting of the unauthorized immigrants is more to the point than the previous chart. But once again there is no reliable/consistent pattern between a surge in the undocumented and an outperformance by the Establishment Survey Nonfarm Payrolls over the Household’s Employed.

In regards to hiring of undocumented immigrants, what jobs are they getting?

Not likely Government, which had the second biggest job gains of any sector in the last twelve months, +629k. That takes the total Nonfarm Payroll outperformance from 2.1 million down to 1.4 million-ish right off the top. Then I have to ask, what if the BED is right and the B/D is not? If so that takes the 12-month outperformance spread below 700k.

Some sectors in which undocumented people have found work in the past have been relatively strong in the last year. Private Education and Health was the number one sector in the last twelve months, with a gain of more than one million; of that +271k was in Social Assistance. The third best sector was Leisure/Hospitality, of which the Food Services/Accommodation category had a gain 278k in the latest year. Construction hiring was decent at +215k.

Some of these jobs were likely attained by people in the US without authorization. But how many?

I don’t know the answer, but I am skeptical that the immigration surge is enough to answer all the questions I have about the slack Employed component of the Household Survey and the questionable Birth/Death Model being such a large percentage of the increase in Nonfarm Payrolls.

I do believe the nuances of the labor market are worthy of close inspection at this stage of the game.