This is a situation that may be on the verge of disappearing and more or less normalizing, but over the last couple of months US bond markets have exhibited a weird phenomenon of non-monotonicity. It has been even weirder than what we saw during the period of negative nominal interest rates, when what we saw was interest rates on US treasury securities fell from the shortest time horizon to a low usually around the two-year time horizon, with the pattern then reverting to its usual upward slope. What has been going on recently has been a pattern of rates initially rising with the time horizon in the normal pattern, then turning around and declining, then turning around yet again and rising again. I do not know what to make of any of this. I exhibit it in a table below for the three days, January 2, January 10, and January 18 of this year.
3 mo. 1 yr. 2 yr. 3 yr. 5 yr. 10 yr. 30 yr.
1/2/19 2.42 2.60 2.50 2.47 2.49 2.66 2.97
1/10/19 2.43 2.59 2.56 2.54 2.56 2.74 3.06
1/18/19 2.41 2.60 2.62 2.60 2.62 2.79 3.09
So, at the beginning of the year the rates rose from 3 months to 1 year, with rates declining to 3 years, and then rising after that. The same pattern was still holding on January 10. On January 18 things were somewhat more normalized with the mid-range decline being a decline from the 2 year to 3 year range, but then reversing to rise in the normal way after that.
“Has this change been partly in response to this weird yield curve shape and that now we are indeed seeing some normalization as the new policy gradually sinks in? I do not know, and I doubt anybody else does either.”
Despite your warning, I am nevertheless going with “Yes.”
This reflects the conundrum of forecasting. After 50 years, we have now reached a point where “everybody knows” that an inverted yield curve means “recession coming in 12 to 24 months,” so instead of ignoring it, traders (and pundits and probably policy makers) react to it immediately.
This of course changes the signal. For example, the stock market is supposed to keep rising after an inversion for 3-12 months, not drop like a rock starting the next day, And the Fed is supposed to keep blithely on, not get skittish in part due to the stock market reaction.
So a supposedly clear signal gets scrambled. Although, to disagree somewhat with another point you make, I think year end 1994 is a pretty good analog to our current situation — but of course n=1, so take with truckloads of salt.
When excess reserves exit faster than the Fed sells portfolio then remits back to Treasury increase, which is an effective reduction in rates on that portfolio balance. It was designed that way, designed to split Fed liquidity between government bonds and private sector deposits under a “zero gain loss” fixed corridor.
The ten year rate has a hard bound, above which mortgage rates will crash our housing bubble, Treasury is anxious to avoid hitting that bound. The Fed system provides a mechanism to invert the curve for a while to avoid the problem. But the mechanism causes dips to utilize gains from scale, the system breaks and a whole bunch enterprises will use the crash to restructure. It is a procyclic tipping point mechanism, designe to hold off corrections util the last moment.
The mechanism that causes the sudden stops is that the current system has all rates derived from treasury curve, and we get a two step process in adjustment. Any algorithm will be step wise doing differences during a change in the terms of trade. The algorithm reaches a prices today equal prices yesterday, and that will be zero at the bottom of a ratio in the closed system, interest rate indeterminate. It can’t happen, as Einstein noticed, so the economy avoids the tipping point with pre-planned cycles, as Einstein discovered with mercury. We do that with a meeting of the elders, non market trades behind closed doors once per recession cycle, compute the next cycle to avoid divide by zero.