The Fed Says It Is “Data-Driven”… But The Data Isn’t Any Good

Authored by Ryan McMaken via The Mises Institute,

It’s been a big week for “the data.” At Wednesday’s FOMC press conference, Fed Chair Jerome Powell announced that the Fed was holding its policy interest rate steady at the current 4.5 percent. Powell noted that there was no need to cut the rate because the job market is “solid.” Powell engaged in the usual song and dance of declaring that the Federal Reserve’s monetary policy is data-driven or “data-dependent” and assured the attending members of the press that FOMC policy is carefully implemented in accordance with federal employment data (among other data points). 

Then, less than 48 hours later, the Bureau of Labor Statistics (BLS) released its July report which revealed that the “solid” job numbers the Fed had allegedly been using for the past two months were actually very wrong. The Bureau of Labor Statistics had greatly overestimated job growth in its earlier reports for May and June. Then, mere hours after the BLS numbers went public, President Trump announced he was firing the head of the Bureau of Labor Statistics. But, he wasn’t firing her because the agency’s data has been initially wrong. Trump was firing her because Trump thought the revised BLS data was too low and made him look bad. 

This leaves us with a couple of questions. The first is this: why are we still expected to take initial BLS job estimates seriously when they are often reduced by 75 percent or more upon later revisions? 

The second question is this: what use is the Fed’s supposed devotion to being “data-driven” when the data itself is unreliable, and the Fed is basing its policies on data that turns out to be thoroughly wrong? The answer is: we can’t. The spectacle of the FOMC making policy based on wildly inaccurate employment numbers simply illustrates the absurdity of claims by Fed officials that the central bank can centrally plan the economy by divining the “correct” monetary policy based on government data. 

The FOMC Meeting and Powell’s “Solid” Employment Numbers 

After announcing it would make no change to the policy interest rate, Powell listed a number of factors that be said justified this policy. Among these claims was Powell’s assertion that “in the labor market, conditions have remained solid. Payroll job gains averaged 150,000 per month over the past three months. The unemployment rate at 4.1% remains low and has stayed in a narrow range over the past year.”

Throughout the press conference, Powell repeatedly emphasized—as he usually does—that the Fed and the FOMC were basing their policies on federal economic data and that the Fed is, to use an often-employed phrase “data-dependent.” Notably, Powell also repeatedly emphasized that when it comes to employment data, Fed is now more interested in the unemployment rate than in job growth or total employment levels. 

All this talk of basing monetary policy on “data” is foundational to the central myth of modern central banking: namely, the idea there is that central bankers can centrally plan the economy by somehow calculating at what level to set short-term interest rates in order to fulfill the Fed’s mandates of full employment and price stability. The questions asked at the press conference by most of the reporters illustrated the success of this central-bank propaganda. It was clear that the media audience believed that the Federal Reserve can, somehow, engineer the “right” level of price inflation—currently set at the arbitrary level of 2 percent—while also ensuring a robust job market. All that it is required, the narrative goes, is that the Fed’s number crunchers convert government data—like the unemployment rate and the CPI level—into sound monetary policy. 

The Federal Job Numbers Are Very Wrong 

That was all on Wednesday. Then, on Friday morning, the BLS released new employment numbers, and suddenly all that Fed talk about “solid” job numbers didn’t look so convincing. 

Friday’s job report was weak overall with total job growth in the establishment survey coming in at only 73,000 new jobs (part time or full time) month-over-month in July. By itself, that would be fairly bad news and suggest a rapidly weakening job market. But that wasn’t the most notable part of the report. The big news comes from the fact that the total job growth for May and June were both massively revised downward. For example, the initial job growth for May was reported by the BLS to be 144,000. But, as of Friday morning, the BLS now tells us that job growth in May was really a measly 19,000 jobs. Job data underwent a similar shift for June. Until Friday morning, we were told that the US economy added 147,000 jobs in June, month over month. But, after Friday morning, we’re now told that the real total growth was a mere 14,000. That’s no mere rounding error. 

Yet when the FOMC met in June, it was using the old bloated estimate of the May jobs numbers. So, the Fed’s “data-driven” decisions at the time were based on what now are apparently grossly inaccurate stats. Then, when the FOMC met again this month, it was using what now look like inflated job numbers from both May and June. 

That suggests the entire enterprise of central planning using BLS data is pretty farcical. 

Is the Fed Lying about the Data it Uses?

Given the fact that the FOMC meeting this week occurred so close to the release of new BLS numbers, one might suggest that the Fed wasn’t really working with the old May and June stats, but actually had access to unpublished “better” data. That is, it’s reasonable to suspect that the Fed secretly knew the “real numbers” going into the July FOMC meeting. That may be true, but if true, then the Fed is deliberately participating in deception by repeating jobs numbers that the Fed knows to be false. For example, if Powell knew the “real” jobs number for May and June at Wednesday’s FOMC press conference, why did he say that “Payroll job gains averaged 150,000 per month over the past three months.” That’s not remotely true in light of the revised data. The average monthly job gain for “the past three months”—presumably April, May and June when Powell said this—was 63,000, not 150,000. (The number is even lower if we’re talking about May, June, and July.) If Powell knew the revised numbers on Wednesday then he clearly lied to the public when he claimed the “150,000” number. 

Powell’s knowledge of the troubling revised numbers may have also motivated his comments about how the FOMC and the Fed are now really concentrating on the unemployment rate. It may be that, knowing that job growth had tanked in May, June, and July, Powell wanted to drive home that he is no longer paying attention to job growth numbers but is focused on the unemployment rate. After all, if one looks only at the unemployment rate, things look pretty good. Even in the new July report from Friday morning, the unemployment rate only slightly increased, rising to 4.2 percent in July. 

This number is belied, however, by continued declines in the labor force participation rate. In July, that rate fell to 62.2 percent in July. Excepting the covid period—when the federal government was using printed money to essentially pay people to not work—the labor force participation rate is now at the lowest it’s been in more than a decade. If the labor participation rate were at more normal levels, the unemployment rate would be significantly higher.

In any case, it is strange for the Fed to suddenly start suggesting that job growth numbers aren’t that important two days before it becomes publicly obvious that job growth numbers at the BLS are so thoroughly unreliable. 

The lesson here is that the Fed’s repeated claims to be “data-driven” are mostly political theater. Even if the employment data represented amazingly accurate estimates, the Fed would still not be able to centrally plan or calculate the “correct” interest rates. As it is, the Fed doesn’t even have convincing employment numbers. 

Trump Fires the BLS Commissioner 

Within hours of the BLS’s new report going public, Trump announced that he would fire BLS commissioner Erika McEntarfer. This move was hailed by the usual MAGA-style disciples of the Trump administration, who joined Trump in claiming that McEntarger was manipulating jobs reports for “political purposes.” Trump insists that the BLS numbers are too low, and don’t reflect the fact the US economy, thanks to Trump’s economic brilliance, is booming. But one could be forgiven for being confused here. If Trump is firing McEntarfer for publishing inaccurate numbers, is she being fired for the initial estimates—which made the job market look good—or is she being fired for the revised numbers? If she is trying to make Trump look bad, why did she first release numbers that made the job market look—to use Powell’s term—”solid”? Trump certainly had no problem with the reports released by McEntarger two months ago when the initial May data was released. At the time, time, Trump declared “GREAT JOB NUMBERS…!”

In fact, McEntarger’s methods have benefited Trump. After all, it’s that initial release of data that gets the most headlines, and it’s the revisions that are usually forgotten and swept under the rug. McEntarger is doing the same thing she did during the Biden administration: release positive initial data, and then revise down the numbers after the fact. The overall effect is to benefit the incumbent administration, regardless of whether the incumbent is Biden or Trump. Trump seems to be too obtuse to understand this. 

We know the answer to these questions of course. McEntarger is only being fired for inaccuracies that make trump look bad. Trump only cares about “inaccurate” job numbers when they are “too low.” So, we have no reason to expect the data to get any more accurate any time soon.