A reader commented some weeks ago that they liked this newsletter because it featured more questions than answers. Today’s edition especially so. Email me if you have some of the answers: robert.armstrong@ft.com
The noise in real yields
At the beginning of this week, before the Chinese market started throwing its toys out of the pram, the lead story on the FT’s markets page was about record-low real yields. Here is the bollocks par (or, translated into US profanity, the nut graph):
The real yield on 10-year US Treasuries fell further below zero on Monday as growing anxiety over the outlook for economic growth added fuel to a recent rally in bond markets.
The idea here — and it is a very widespread idea — is that significant moves in real yields tell us something useful about economic growth. But how do we track real yields?
Usually they are thought of as nominal Treasury yields, minus expected inflation. But that’s not how they are actually calculated; actually, they are not calculated at all. We just treat the yield on Treasury inflation protected securities (Tips) as the real rate, and subtract that from the yield on standard Treasuries, to find out what inflation expectations are. Here’s how those three have been acting lately (using 10-year maturities):
Broadly, since April, Tips yields and Treasury yields have fallen together, implying stable inflation expectations and falling real rates and hence “anxiety over the outlook for economic growth”. But two things strike me as funny here, making me suspect real yields may not be telling us what we think they are telling us. One is minor, one major.
The minor point, pointed out to me by César Pérez Ruiz of Pictet Wealth Management, is that the gold price is not rising. Gold is strongly inversely correlated with real yields (gold yields nothing, so the opportunity cost of owning it falls with real returns available elsewhere). But in the past few months, as real yields have fallen to all-time lows, gold has fallen. Here is a chart (data from the Federal Reserve) of gold and Tips yields, with the gold price inverted, to show the correlation and its recent breakdown:
As Ruiz put it, why is gold at $1,800, instead of heading for $2,100? He thinks gold prices are anticipating that economic growth is going to disappoint, so inflation will be lower and real rates are going to rise.
He may well be right about growth (my guess is he is), but notice what this implies about how the market works. It suggests that an inflation premium is not added on top of a real interest rate that corresponds in some way to the general level of real growth in the economy. Instead, changes in inflation expectations can pull real yields up or down, depending on what nominal yields do.
If you think that nominal yields are not determined by economic expectations alone, but also by “technical” supply/demand/portfolio composition issues, this makes sense. But it draws into question the assumption that the decline in real yields is best understood as a reflection of growth expectations.
Now the second, more important funny thing. Real yields have fallen below the level of the long, dreary, pre-vaccine stage of the pandemic, between August of last year and February of this year. Is the growth picture now really as dim now as it was then, before we knew how effective vaccines were? We are expecting earnings for the second quarter to be up 70 per cent or so; GDP is expected to come in at something like 7 per cent; inflation is running at 5 per cent or so. And things are as bad as a year ago?
Something is wrong with the picture that Treasury and Tips yields are drawing — which suggests, again, that these two securities might not be economic bellwethers so much as creatures of the more or less eclectic forces of supply and demand.
If you think the government bond market might be sending bad signals about the economy, it is easy to guess where the static is coming from: the Fed, and its $120bn a month of bond buying. What would yields on Treasuries and Tips be if the Fed was not in the market?
The Fed does buy Tips — in fact, it now owns about a fifth of the outstanding supply of them, according to data from the Fed and the Securities Industry and Financial Markets Association. That compares to almost about 29 per cent of outstanding Treasury notes and bonds, but the Fed’s holdings of Tips and the share of the total outstanding they own grew faster in the past year than for Treasuries.
I asked James Athey of Aberdeen Standard Investments about this, and he gave the following picture — which sounds a lot like the way Ruiz sees things:
The problem here is that Tips are an order of magnitude less liquid and less utilised by investors when compared to nominal Treasuries — therefore the Tips market isn’t as efficient, and nominal Treasuries exhibit the first-mover status.
Which leads me to my working model for this stuff: Nominals move first, the market has some idea where [inflation expectations] should be and so real [rates] move as a remainder . . . ie. Tips yields merely serve the purpose of generating an “acceptable” (equilibrium? Market derived?) break-even inflation rate given the nominal rate is what it is.
So the Fed is lowering real yields by buying Tips, and its even heavier buying of Treasuries suppresses real yields, too, because the Fed can’t control inflation expectations, so as nominal rates go down, real rates must also go down given stable inflation expectations.
The more you think about this stuff, the less it looks like real yields are a good growth indicator. We don’t know what the true level of real rates are. More importantly, we can’t be sure Fed interventions affect Treasuries and Tips symmetrically, so it is not clear that changes in real yields tells us all that much about growth, either.
Bob Michele of JPMorgan Asset Management told me that “right now, yields are anything central banks want them to be” but suggested we might look to historical experience to determine what real and nominal yields would be without Fed intervention:
Pre-pandemic, the real yield on the 10-year Treasury averaged about +0.5 per cent from 2014-19. If we look at the current level of real yields (-1.13 per cent) and replace it with the 0.5 per cent average, you lift the 10-year by 1.63 to 2.88 per cent.
But using post-great financial crisis and pre-pandemic yields as a guide may be conservative. Before the financial crisis, QE was not such a standard part of the Fed arsenal:
I like to look at the 1992-94 experience. That was a time when the Fed put in place very aggressive monetary policy following the savings and loan crisis — but, without a QE programme distorting markets. Put differently, they left the yield curve to investors like me to price! The 10-year UST traded at a spread of almost 4 per cent above the fed funds rate in November 1992, over a year ahead of the Fed’s first rate hike in January 1994. Anyway, in that period where the Fed was behind the curve but not implementing a QE programme, the 10-year UST traded ~2.5-4 per cent above the fed funds rate — implying a yield of 2.5-4 per cent today based on that comparison.
If Michele’s arguments are even directionally correct, not only do we not have a good grasp on what today’s real yields are telling us about the economy (because we don’t know if the impact of QE on Treasuries and Tips is symmetrical) but we should be ready for the possibility that the rate environment will change significantly when — or more to the point, if — the Fed tapers its asset buying programme.
One good read
An academic paper suggesting that Treasuries are actually too cheap, of all things (hat tip to Marginal Revolution for pointing it out).
Source: Economy - ft.com
