There's No Floor Under Yields. There Never Was.
In John Maynard Keynes’ The General Theory of Employment, Interest and Money, the concluding chapters deal with, among other problems, the possibility of persistently low interest rates. When the book was first published in 1936, that was the issue at hand in bond markets all over the world. We are taught to associate low interest rates with stimulus when historical experience consistently shows otherwise.
Part of the confusion stems from a sort of misunderstanding, examining the level of interest rates in relation to zero or some other artificial nominal anchor. What do we really mean when we say interest rates are low?
By itself, a nominal yield doesn’t tell us all that much. Even central bankers concede this point. That is why they spend so much time and computer power attempting to calculate R* - the presumed “natural” interest rate, the magical point at which everything in employment, interest and money harmonically balances. From R* they further infer the “neutral” interest rate so that monetary policy can be squeezed in.
If the current rate is above the neutral rate, then monetary policy is presumed to be unhelpful. And if R* is so low that it has caused the neutral rate to be the same, then what are policymakers to do?
This was a good part of the subject in Keynes’ final pages. Given the persisting low interest rates in the US and the UK at the time, Great Depression after all, and the stubbornness of it, as an economist studying the very idea of equilibriums he came to wonder if the low interest rates just weren’t low enough to have reached one.
What Lord Keynes suggested was that liquidity preferences may have put a floor under yields. Bond investors are not just bondholders, they have all the same animalistic spirits that in times like those were simply being repressed. Everyone is seeking to maximize returns no matter the underlying conditions.
The whole rationale for converting money into bonds was opportunity versus risk; the risks of doing something risky with money was too high and therefore investors preferred buying and owning the safest and most liquid instruments. Since money didn’t pay any return, to generate at least some reward a tradeoff of money for interest bearing securities but mostly of the highest qualities.
Thus, low yields on safety instruments when money, and economy, is tight.
However, bondholders believed that the Great Depression wasn’t a permanent condition. Despite later theories on secular stagnation and whatnot - that the economy was broken by an unfixable mix of decaying demographics, lack of innovation, and no more empty land to settle – the flight to safety in bonds was imagined to be temporary by those doing the fleeing.
At some point, everyone knew, recovery would happen bringing with it inflation as well as opportunity. Therefore, a very bad time to be holding safe assets.
By thinking ahead, bondholders would, theoretically, only go so far in bidding up the price. They would not buy them past some unknown threshold, an equilibrium if you like, always keeping a watchful eye for when normalcy might begin to show up. And then the bond rout!
For Keynes, this might’ve been a big part of the recovery problem. If looking forward toward a prospective recovery put a floor under interest rates, and that floor just happened to be higher than the neutral and natural rates, then no matter how low people might think rates have sunk they haven’t sunk low enough!
Expectations theory has been a crucial aspect of economic study for about as long as there have been economic cycles. Going back to the nineteenth century, economists have long sought a way to model and define how individual economic and market agents process information about the future, reach conclusions maybe even a consensus about it, and then act in the present on them.
In our modern age, central bankers have developed strategies and theories. If bond yields encounter a floor, be it the one described by Keynes or just the zero lower bound, they have QE. Should bond investors balk at the obvious successes QE will obviously bring, then the bond buying itself will force rates down to where these modern mathematical equations calculate they need to go.
Which, to economists, also requires what is called forward guidance. I won’t get into the presumed differences between Odyssean forward guidance (explicit communication) and Delphic (implied), suffice to say that central bankers are aware of what’s supposed to happen to bonds in the same way as Keynes was struggling to work out.
Forward guidance is to deal with the eventual success of QE and unconventional monetary policies. Once the current rate, or set of rates, is reduced whether voluntarily or by force of central bank purchases, the bond market will become nervous about the forces of recovery – which, again, are very bad for safe instruments.
Instead, central bankers either declare (Odyssean) or imply (Delphic) that they will keep on buying bonds no matter if recovery and inflation do show up (as expected). It is a message to bond buyers that they shouldn’t fear the 100% guaranteed future success of monetary programs because the central bank will maintain the bid for these securities long after it is necessary.
If we decompose these low anticipating-recovery-yields, for lack of a better term, what we’d expect to find is rising inflation expectations within them. Recovery after nasty or/and prolonged contraction and weakness should be highly inflationary. However, if forward guidance combined with bond buying is keeping interest rates low, then what offsets the rising inflation expectation within yields?
The answer in the modern textbook is something called “term premiums.” What they are supposed to be and what they actually are can be two very different things. The conventional definition is the additional spread a bondholder demands for investing in a longer-dated bond. You need more return to lock up your money for a longer period of time, a higher rate associated with a longer maturity.
In our recovery case of low yields, rising inflation expectations are balanced by lower term premiums brought to the market via central bank bond buying and forward guidance assurances. In this view, lower rates are tortured into some conception of lower risk.
Even though bond buyers have been buying safe bonds on the premise of much higher perceived risk; the hoarding of the safest and most liquid assets is turned around by this theory into the notion of smaller risks and a safer condition. Same low rate, very different textbook meaning.
Falling term premiums and rising inflation expectations eventually replace HOLY CR$%!
But how do markets and the economy actually look at these things? When it comes to inflation and economic recovery, there is a lengthy academic history and a long list of literature seeking to define expectations. From adaptive to rational expectations, somehow, some way agents process information available today and form opinions about what’s likely to happen tomorrow and beyond.
In the fifties and sixties, economists used adaptive expectations functions to model these processes, and these models performed very well in fitting real world inflation outcomes. Pioneered by Phillip Cagan in his 1956 article The Monetary Dynamics of Hyper-Inflation and Milton Friedman’s 1957 book A Theory of the Consumption Function, both based, in part, on Irving Fisher’s earlier The Purchasing Power of Money published in 1911, adaptive expectations suggested that agents form their expectations for the future in the simplest, most elegant fashion.
They look to the past.
But, as we all know, sometimes the past is a poor predictor of future events. Things change. People adapt to changes in ways they had never acted before.
So, the adaptive expectations framework, which importantly created enormous mathematical problems for stochastic modeling, was eventually replaced by rational expectations.
It was the latter that has formed the basis for the hypothesis of efficient markets as well as our modern infatuation with the random walk. The idea that what happened yesterday has no bearing on what happens today. The slate is wiped clean at dusk so that by dawn everything is new again.
Or is it?
According to rational expectations, we would expect the bond market to behave exactly as described in the conventional theory; bond investors initially fled into safe haven assets but soon forgot all about why they had, information about yesterday that was replaced with quite rational thoughts on forward guidance and a massive load of bond buying today, tomorrow, maybe for the foreseeable future. Rising inflation expectations and low maybe negative term premiums.
What would it be, though, if inflation expectations weren’t rising all that much?
This was the puzzle that confronted the Fed’s policymakers in 2014 – the era of the final grand recovery. Your recollection might be a little fuzzy with the unemployment rate lower now than five years ago, and the force with which recovery was talked about proportional only to that particular economic account. But 2014 was in every way outside of the unemployment rate much, much better – especially its middle months.
And still it had left policymakers with a puzzle (the same puzzle, by the way, they struggle with today). Where was the inflation? Furthermore, bond yields weren’t just low they were sinking. It all seemed upside down.
As the FOMC gathered in December of 2014 to further discuss first their grand plans for normalizing monetary policies after more than a half a decade of non-standard and emergency policies, they were bothered by the inconsistencies to the point of wondering if those might actually derail the intended regime change. Were they missing something about the bond market?
As a supplement, Committee members for that particular meeting were provided with a staff paper which had been prepared in September 2014 – but only released to the public last month - just as oil prices were swinging harshly in the wrong direction. The topic was inflation expectations, specifically if long run expectations might not be behaving in the way theory, especially rational expectations, expects.
In what must’ve been a shocking proposition, the study raised the possibility that adaptive expectations might better explain observations in bond market behavior. “Such a mechanism”, the paper’s authors wrote, “may have some plausibility if the public comes to doubt either the willingness or the ability of policymakers to achieve their inflation objective.”
In other words, no random walk. Bond investors had seen all the QE’s, had heard both the explicit and implied forms of forward guidance, and may have concluded it was all bunk. Expectations about recovery and inflation were being priced on empirical history rather than what central bankers always claim they can do.
The problem with using adaptive expectations, so far as the authors were concerned, was that inflation expectations should’ve been falling since 2008.
“Kozicki and Tinsley (2001) showed that adaptive expectations provided a reasonable approximation to survey measures of long-run inflation expectations during the 1980s and 1990s. However, this approach has proved less useful for forecasting survey measures of long-term inflation expectations in recent years since it would have implied a noticeable, lasting decline since 2008 that was not observed.”
As central banks failed to stimulate with all their stimulus (talk), why hadn’t inflation expectations fallen consistently along with what were privately admitted shortcomings to the economic recovery?
And that was true; market-based measures of inflation expectations seemed at least consistent especially in long run tenors. If we use something like the 5-year/5-year forward inflation rate, a number calculated by comparing the 5-year TIPS breakevens (the difference between the nominal yield on the same maturity UST and the TIPS real yield) and the 10-year breakeven, up until August 2014 there hadn’t been much change in the distribution.
In the precrisis era, from January 2003 until the end of September 2008, the 5-year forward in the US remained in a relatively steady and narrow range. On 60% of trading days it was between 2.34% and 2.52%; very well anchored, in the parlance of central bankers. In the early “recovery” period, defined as July 2009 through June 2014, on 60% of trading days it fell between 2.34% and 2.73%.
It had seemed a tiny bit consistent with policymakers’ expectations; a distribution of key long run inflation expectations that were a little higher than in the precrisis era. But why weren’t they so much higher, as mainstream media commentary (and stock prices, post QE3) had made them?
If we run the numbers for the last almost half decade, however, the whole distribution changes. Inflation expectations, long and short run, haven’t been rising at all. Dating back to the middle of 2014, they’ve collapsed – just as adaptive expectations would’ve assumed. As the Fed’s paper said, “if the public comes to doubt either the willingness or the ability of policymakers to achieve their inflation objective.”
Since July 2014, on 60% of trading days the 5-year forward rate has been between 1.75% and 2.14%; a substantially lower distribution. In fact, 80% of the time it has been stuck between 1.67% and 2.19% - validating the staff’s worries about which doubts, which form of expectations, are being expressed by the bond market.
And why wouldn’t it? What was 2015, after all, if not one too many false dawns?
Speaking of which, the 5-year/5-year forward inflation rate is approaching the extremes of the distribution again. Just a shade above 1.52%, that put its it into the 3rd percentile of the last two decades. It has rarely been lower than it has this week; a few times in 2016 and before that the heart of the Global Financial Crisis.
Here's the thing – this 5-year rate isn’t about the coronavirus. That’s the point of using the 5-year/5-year inflation forward. It isn’t a short run measure of what the market expects for the next few months or even years after everyone has forgotten about COVID-19. It’s a market-based calculation of what bondholders are thinking about much bigger concepts like recovery, economic potential, and the ability of central banks to make or influence them (and the inflation associated).
Bonds have heard the optimism for long enough. These lower inflation expectations erase the whole chatter on term premiums, explaining lower nominal bond yields the other way – clinging to safety because risks are, and remain, paramount. Not virus risks and the uncertain fallout from the pandemic, but yet again such a perpetually weakened economic state that even the threat of one is enough to take seriously a potential break in it.
Low rates are not stimulus, they are the signals for when there hasn’t been any, falling because inflation expectations are and so are those for nominal (and real) economic growth. The market is (still) saying that central banks have utterly failed.
Does record low UST’s prove Keynes was wrong about the floor the recovery instinct supposedly puts under yields? Maybe, but maybe not. Keynes had based his proposition on the idea bondholders would fear holding bonds during a recovery they expected to happen at some point. He didn’t really work out where things could stand in the situation where bonds stop expecting any recovery.
Which is where we are right now and have been for many years. Apologies to all those bond kings and their Jay Powell-inspired inflationary imaginations driving them to negative duration bond portfolios, the real bond market has adapted to saying – especially since 2014 - there is no floor under yields. There never was.