I tried to conclude the post I wrote on overseas QE with a discussion of how the term premium at the 10-year maturity of the US curve is pretty well correlated with expectations for the Fed.
The argument was (and is) that the term premium is more about perceived rates vol and asset correlations than about the net supply manipulation achieved by QE, here or overseas.
I figured a good proxy for the outlook for the funds rate, and the economic environment bearing on that, would be just the 5-year note yield. And I was hoping for a pretty good fit. I was not hoping for this fit!
From the dawn of ZIRP pretty much exactly until its end, the 5-year Treasury note yield almost entirely “explains” the 10-year term premium. They even have the same bp vol over the ZIRP period, implying a “beta” of 1, if you like. Note in the chart that the left and right scales have the same range, just offset by a constant 150 bps. So my intuition is confirmed!
Mmmm, not really, I realized after failing to write it up coherently a few times. Two numerate friends have since explained to me that the estimated term premium is largely a function of the gap between short-term note yields and consensus estimates (among forecasters) of the path of the funds rate.
The forecasters’ stubborn and wrong-footed adherence to the logic of “renormalization” has not been shared by the bond market, which has created an impression, I would say false, of a huge negative term premium at the 5-year maturity. At other times, the bias has been less extreme, but the fixity of the consensus estimates has made all the little moves in the five year note yield look like term premium changes.
Within Kim-Wright, this apparently gets extrapolated out to future maturities, although with some dampening. That would explain why the 10-year term premium is so correlated to the 5-year and why the difference between the premia is a function of how hawkish or dovish the forecasters are relative to the actual market. Recently, the 5-year premium has been more negative than the 10-year, for example, because the forecasters have been particularly stubborn, relative to the market.
Kim and Wright’s estimate of the term premium would seem currently, then, to have little to do with what we would conventionally think of as the term premium, that is the expected excess return to funding long duration investment with rolling short-term debt — or by bearing a commensurate opportunity cost.
* This is not my thing, obviously, but I wonder if one issue here might be that forecasters’ estimates of the path of the funds rate might have become particularly unrepresentative of the market with the dawn of ZIRP in 2008. Upon the arrival of ZIRP, the Fed began to deliver interest rate guidance and, separately, the policy rules favored by Fed watchers predictably became defunct. To the extent that Fed watchers are particularly beholden to the Fed’s views and particularly likely to be taken in by policy rules, it is possible that they became more-than-typically unrepresentative after late 2008. If so, then the Kim-Wright estimate of the term premium might have assumed a durable downward bias to accompany the excessive volatility already seemingly built into it by design.