You, Neel Kashkari, can do better.- James Grant
"Their cocksureness does not become them. The dogmatists were just as confident in 2021 that a 10 basis-point funds rate was a lock through 2023."
Memo to Neel Kashkari by Jim Grant
Faithfully Excerpted from Grant’s Interest Rate Observer
A Jan. 4 post by the president of the Federal Reserve Bank of Minneapolis, “Why We Missed on Inflation, and Implications for Monetary Policy Going Forward,” is a sight as welcome as it is unconventional. Rare is the central banker (or, for that matter, journalist) who freely cops to error, but you, Neel Kashkari, can do better.
The future isn’t mankind’s best subject, all can agree. We can’t predict it, yet we keep trying to. The Fed, especially, is incorrigible. It failed to foresee the inflation of 2021 and counting, even though it was printing the money with which to finance it. Nor can it predict the funds rate (as the economist Gary Shilling has pointed out), even though it’s the rate that the Fed itself controls.
We don’t fault the central bank entirely for missing, though—you have to admit—to exacerbate the problem of lockdown-constrained supply with heavy doses of QE right into the first quarter of 2022 was no career-enhancer. Where we do fault you and your confrères is persisting in the business of economic star-gazing—of imputing more utility to what you call your “workhorse” econometric models than those broken-down nags can possibly bear.
In 1961—maybe you know this story—Edward N. Lorenz, mathematician and meteorologist par excellence at the Massachusetts Institute of Technology, was entering data into his computer, an off-the-shelf, 800-pound, $47,000 Royal McBee LGP-30. To simulate weather patterns, Lorenz modeled a dozen variables and compared one simulation with another. To save time, he rounded six-digit inputs to three digits—thus, for instance, 0.506127 became 0.506.
“To his surprise,” as Wikipedia tells the story, “the weather that the machine began to predict was completely different from the previous calculation.” The source of variance turned out to be the rounded numbers. The missing three digits seemed inconsequential enough—and were, to start— but the cumulation of the missing digits made all the difference, and errors grew like Pinocchio’s nose: “Small changes in initial conditions produced large changes in long-term results,” was Lorenz’s conclusion.