I found this critique of the Bureau of Labor Statistics’s monthly Labor Situation at Kitco News by Ernest Hoffman interesting. The bottom line is that it’s neither accurate nor timely enough to be of use to investors, the Federal Reserve, or policymakers. Particular issues he takes note of include loss of confidence due to large revisions, decline in participation by employers from 60% to 40%, failure of the probabilistic model, and the complexity of dealing with immigrant labor. I would add the issue of outsourcing.
The piece has a number of salient quotes.
Grady said the Trump administration’s criticisms on this front are well-founded. “They’re using 1970s metrics to get this data,” he said. “You can’t run an economy based on that. It just doesn’t work.”
I don’t think that’s quite right. They’re 1990s metrics devised by people who went to school in the 1970s.
“They have to be perfect, all the data that goes into the system, because the markets are trading on it,” Grady said. “But when it comes to the government, they don’t have the same criteria. It’s just wrong.”
Grady said he’s certain that if the Federal Reserve had had the correct data, if they had known about the 911,000 decrease in net jobs, they would have cut rates earlier.
I don’t think he quite captures the difference in incentives between investors and analysts working for the BLS.
“It’s based on a voluntary survey, voluntary participation,” he said. “It used to be 60% [participation rate] and now it’s 40%. The trading is with algorithms; they’re trading numbers in the milliseconds, but you’re trading based off year-old data. Who does that like this? It’s just wrong. There has to be a better way to calculate this data, because it’s too important. You can’t run an economy on it.”
That’s a point that I have made on more than one occasion. Perhaps the question that should be considered is whether the Labor Situation report is performing a useful function at this point. Or the BLS for that matter. There’s an old expression they might consider: “a miss is as good as a mile.”
Adam Button, head of currency strategy at Forexlive.com, also believes traders and Fed officials are being led astray by the monthly jobs report, but he thinks it’s the Trump administration’s immigration policies that have rendered it all but useless.
“We may have entered the post-payrolls era,” he said.
In a Sept. 26 interview with Kitco News, Button said the fundamental assumptions on which the health of the jobs market is built no longer apply.
“I think the market has realized, or will soon realize, that low employment growth in the headline non-farm payrolls number in the United States is no longer really indicative of anything, because of immigration changes,” he said. “Ken Griffin was talking about it this week, and [Richmond Fed president Thomas] Barkin today mentioned it, that flat U.S. jobs growth may be enough to keep the unemployment rate low.”
I don’t think it’s just immigration. I think that offshore outsourcing is considerably greater than they recognize. When even mom and pop shops are offshoring some of their functions (they are), it adds up.
Button said all the models currently in use – whether by the government, the Federal Reserve, or the private sector banks and hedge funds – are built around a set of parameters about demographics and population growth that can’t account for mass immigration OUT of the United States.
“So much is model-driven in markets right now,” he said, “and the models get all out of whack. Whatever rule they have around jobs, or correlation, or anything else like that, it looks like a recession. And [the impact of net migration] might not come all at once; this might unfold over a number of years.”
Button said all this adds up to a very challenging situation for the Fed and other policymakers – and a very risky situation for investors and traders. “I just think you need to be really cautious with the headline jobs number right now.”
I saw several article the other day claiming that LLM AI wasn’t claiming many jobs in the U. S. Clearly, they hadn’t looked at the numbers coming out of India, where the unemployment rate among information technology workers has become quite severe, becoming worse over each of the last three years and much attributed to AI.
