Rates (mortgage and otherwise) dropping; here’s what’s going on

Long-term rates have continued in a narrow band, but in very odd behavior have trended ever-so-slightly down. The 10-year T-note today trades at 1.70 percent, the lowest since the April 7 top at 1.93 percent, and lowest-fee mortgages are sliding under 3.75 percent.

Why odd? This slide began two weeks ago, all through the week prior to release of April payrolls. Nobody trades that way, buying bonds and MBS (mortgage-backed securities) ahead — the payroll data will kill you, often arriving way off forecasts.

And the gentle slide continued this week, in which the Treasury sold at auction $62 billion in new bonds, more than half 10-year maturity or longer — never buy bonds in advance of a Treasury dump.

Stick with simple. No need to put beer cans on the rabbit ears to improve reception. German and Japanese 10-year bonds compete for buyers with ours, and in the same two-week span, their yields have declined also, pulling down our yields. German 10s have slid from 0.30 percent to 0.128 percent, and Japan’s from minus-0.067 percent to minus-0.109 percent, approaching April’s all-time low, minus-0.13 percent.

Central banks have engineered low rates, but this trading is a repudiation by markets of potential central-bank success in global recovery.

Here in the U.S. we are moving toward — or are already in — a period of lunacy. And I haven’t even gotten to Donald Trump. Our bond and mortgage yields are sky-high versus the world because the Fed insists that the U.S. economy is growing too fast. Right now, too fast! And the Fed intends to do something to slow it.

Eric Rosengren, Boston Fed prez, one of the very best people at the Fed: “The market remains too pessimistic about the fundamental strength of the U.S. economy, and the likelihood of removing monetary accommodation is higher than is currently priced into financial markets.” He continued that the U.S. non-inflationary growth limit is 1.75 percent GDP (gross domestic product), and that U.S. job growth must slow to 80,000-100,000 jobs monthly.

Rosengren has shown good feel for the economy in the last 10 years, although his speech this week nearly ignored the outside world, and alleged a commercial real estate bubble (not). Roberta Mester, prez of the Cleveland Fed spoke this week also, in Germany but ignoring overseas impacts on the U.S.

It’s worth scanning these speeches. Just visit the regional Fed websites, and you’ll get an immediate ego check on each prez: you have to dig a few clicks to find Rosengren’s texts; the Cleveland home page is all about Roberta. Roberta is new, and writes in the bossy tone of the insecure, and is an exceptional mathematician and hence takes refuge there. Math is the basis for Fed action. They don’t gather to act on hunches. But their math and models are busted.

Even Roberta said, referring to the recent fantastic failure of the Philips curve models to predict inflation: “We need to acknowledge that the measures of slack in the Phillips curve relationship are imprecise because they depend on unobservables like the natural rate of unemployment and the level of potential GDP.”

What in Hades is an “unobservable”? That is the euphemism economists use to describe assumptions beyond our ability to measure, but we plug them into models anyway. GIGO (garbage in, garbage out).

[Tweet “‘Unobservable’: A euphemism economists use to describe assumptions beyond our ability to measure”]

The Fed stands surrounded by model wreckage. Roberta on one repair effort, “…building model archives to aid in the systematic comparison of empirical results across a large set of economic models. One such archive… includes 61 models.”

I feel for mathematicians whose math does not work, but piling failed models on top of each other to get a better result? Uh-uh.

So we wait to see if U.S. data emboldens the Fed again, and then to see how the world responds to Fed tightening into a world in which every other central bank is in full-scale stimulus panic.

This election is going to matter in markets. Not by playing a favorite, but by undermining confidence until one or both make a positive contribution. Neither have expressed plans which would raise the Fed’s estimate of speed limit, such as sound ideas for productivity and success in competition.

U.S. 10-year T-note, one year back, in the process of a third try to break 1.65 percent. Traditionally — and I don’t know if tradition applies today — a bond rally like this in the face of Fed saber-rattling should tell the Fed that tightening policy now is nuts.

U.S. 2-year T-note, the most Fed-sensitive, one year back. A different pattern deepening: descending tops and ascending bottoms, an unsustainable “wedge.” How the pattern breaks, up or down, will be predictive.

Stare long enough at this chart of monthly payroll growth, and you’ll see a tail in the last six months.

Same tail in small business (in each of the NFIB — National Federation of Independent Businesses — measures, only one shown here).

The next three charts are from Mester’s speech. She dodged acknowledgement (as the whole Fed does, Lael Brainard excepted) that the forecast above looked exactly the same as now in the spring of each of the last six years, and completely mistaken in actual result.

Above are the forward forecasts of two regional Feds. Both have been completely mistaken as to actual result, inflation at no time rising near 2 percent, and the two don’t even resemble each other. Quit. Do something else.

I believe in the Fed. Got to have one. We can’t be trusted with money, nor can the President — nor, heaven knows, Congress.

Here is Mester’s slide for an evolving Philips curve equation. Philips had a simple idea: low unemployment results in rising inflation, and vice-versa. But, given Fed assumptions for today’s unobservables, the curve doesn’t work. Thus a classic mathematician’s solution: make the equation longer.

A far better idea: ask and ask, why is the world today so different?

Lou Barnes is a mortgage broker based in Boulder, Colorado. He can be reached at lbarnes@pmglending.com.