The Duration Paradox (Part One of Two)
It is commonly held that, by setting only short-term interest-rates, central banks leave bond yields for the free market to determine. While mostly true under normal conditions, there are situations in which this is clearly not the case, for example, when the central bank becomes a large buyer of longer-dated bonds. As such, it is claimed by some that the US Fed is currently holding US bond yields below the level at which a free market would place them. While Fed bond buying almost certainly has some impact, there are more subtle, indirect ways in which a central bank can influence bond yields. One of these, the Duration Paradox, comes into effect in an unusually low interest rate environment, such as that we find ourselves in today. But if yields do not properly compensate investors for the risk of holding government bonds in an environment of soaring government deficits and money supply, they need to look elsewhere. Historically, precious metals and various other real assets have provided the preferred alternatives.
Our explanation for why bond yields were so low at that time was twofold: first, that much demand for Treasury bonds comes from sources that buy bonds as a matter of economic policy, rather than as a rational investment decision; and second, that the Fed was beginning to talk yields down around that time by raising expectations for another round of Treasury bond monetization, subsequently announced in early November last year and now known as QE2 (for ‘quantitative easing round two’). Here is a relevant excerpt from that report:
Let’s take a look at who buys Treasury securities, and why. Much of the Treasury market is held by government and financial institutions that, in practice, have little discretion. To the extent that there is a free-market in Treasury securities, it exists amidst much official, somewhat systematic buying from the Fed, foreign central banks, US state and local governments and US financial institutions….
Foreign central bank holdings are also a matter of government policy in that they grow with balance sheets. These have risen dramatically in recent years…
State and local government holdings of Treasury securities are also quite substantial, at $535bn as of end Q1 2010. These are not for pension funds or other investments but rather for liquidity management. As such, they are also not discretionary holdings…
US commercial banks are also large holders of Treasuries, about $1.5tn at end Q1. Now under normal circumstances in which the financial system was not in distress and bank lending was growing at a healthy pace, these holdings could rightly be considered discretionary as they would reflect to some extent the bank’s trading view on the direction of yields. However, commercial bank lending has declined sharply over the past year. If banks are unwilling to lend, presumably because they can’t locate enough suitable, creditworthy borrowers, then an obvious, low risk alternative to making loans to the private sector is to make loans to the Federal government in the form of Treasury purchases…
Adding up the Treasury holdings of these various groups we arrive at a total of around $5.5tn, or some 2/3 of all marketable Treasury debt, which we can assume is not held for primarily discretionary, investment-driven reasons.
Given the large presence of non-discretionary buyers in the market, it is questionable whether or not the observed yield curve at present is close to where a purely discretionary market would place it. There is also the issue to consider whether, given this market structure, investors are discouraged from shorting the market even if they anticipate higher yields in future…
[I]nvestors shorting Treasuries with a longer-term view that yields are headed higher someday as a debt/currency crisis eventually arrives must assume the risk in the meantime that the monetary authority chooses to monetize some portion of the debt, thereby preventing a rise in yields. In this case, those who are short are going to sustain losses in the domestic currency. And unless the domestic currency weakens, those short are going to sustain losses in terms of other currencies as well. It is only in the event that the currency devalues that investors going short will profit. Such perceptions of risk, influenced as they are by credible threats from the monetary authority, most certainly contribute to the conundrum of low yields in the face of rising risks of a debt/currency crisis in future.
Subsequent to the Fed’s formal announcement of an expanded program of Treasury bond purchases in early November, Treasury yields began to rise, with the 10y rate increasing from a low of around 2.4% in October to around 3.5% currently. On the one hand, this could be considered an entirely normal ‘buy the rumor, sell the fact’ reaction in financial markets. But consider the text in bold above. It is important to note that, last summer, US commercial bank lending stopped declining and, as such, this implied that the banks, which had been huge buyers of Treasury securities over the past two years, no longer had as large an appetite. In this context is it not particularly surprising that Treasury yields began to rise again.
US commercial bank lending stopped declining in mid-2010
Does this, however, imply that Treasury yields now more properly reflect the longer term risks associated with the current set of US monetary and fiscal policies? No, it does not. The Fed remains an active buyer and is now the largest holder of Treasuries. Foreign central banks continue to accumulate Treasuries as a matter of their own respective monetary and currency policies. US commercial banks may now be buying less but the market remains dominated not by investors but by policymakers. And there remains a credible threat that, in the event the Fed believes that Treasury yields are rising too far, too fast such that this threatens the anemic US economic recovery, they will expand and/or extend their current bond buying program. As such, the Fed is still strongly influencing investors’ perceptions of risk, making them think twice before selling Treasury securities and pushing yields higher.
There is, however, a more subtle way in which the Fed influences bond yields with changes in short term interest rates. At a recent conference in London (which can be viewed here), I presented on the topic of the ‘Duration Paradox’, a phenomenon I first identified while working as the Head of Interest Rate and Currency Strategy at Dresdner Bank in the fall of 2002. Back then, the Fed was lowering US interest rates to historically low levels. Under normal circumstances, as short rates decline, long rates follow, although by a smaller amount. This implied ‘bull’ steepening of the yield curve is normal in recessions and, as the economy eventually recovers and the central bank begins to raise rates again, the yield curve normally ‘bear’ flattens. However, back in late 2002, the yield curve was not steepening by as much as implied by the decline in short rates. In other words, Treasury yields were following short rates lower by more than they normally would, implying a strengthening relationship. Why?
The answer lay in the fact that by late 2002, market expectations were that short rates were likely to approach zero in the coming months, which is a limit below which they cannot fall. This implied that the potential path of short rates was becoming highly asymmetric: They could rise dramatically at some point, perhaps back to levels then thought normal of around, say, 4-6%. But once rates approached 1%, as they did by mid-2003, there wasn’t much potential downside left.
This asymmetry is the key to understanding why the relationship between short rates and long rates had strengthened. Consider for a moment that, in any financial market, there are always bullish and bearish investors. The bond ‘bulls’ position for declining interest rates (implying rising prices) and the ‘bears’ take the opposite view. Now in a normal interest rate environment in which the potential future path of short rates is not asymmetric (or in which the potential distribution of outcomes is not otherwise truncated), the bulls and the bears can theoretically position at any point along the yield curve, as there is both bullish and bearish potential. Moreover, the relative price sensitivities of different points along the yield curve can be related to one another by using a bond mathematics concept know as ‘Duration’, which is based on theoretical parallel shifts of the yield curve, up or down.
But as the bullish potential for short rates becomes constrained as they approach the zero bound, then the bulls have no option but to move their positions further out the yield curve. Duration no longer remains a valid way in which to compare the expected returns from a position in short rates to the expected returns from a comparably price-sensitive position in long rates, as a parallel shift of the yield curve is no longer possible. The bears, however, can continue to position wherever they like, at the short or at the long end (although the asymmetry may lure them into shorter maturities). The net result is that bullish investors are forced to migrate out the curve to where there is still a relatively unconstrained, potentially bullish path of interest rates, whereas bearish positioning can remain unchanged or, perhaps, is lured into shorter maturities.
We decided to name this breakdown in the applicability of Duration to extremely low interest rate environments the Duration Paradox. It explains why the relationship between short rates and long rates strengthens as short rates approach zero and, by implication, why the yield curve does not steepen by as much as it otherwise would. (As an empirical check on our research, we looked at what happened to the Japanese government bond (JGB) curve as the Bank of Japan slashed short rates to low levels in the mid-1990s. As we suspected, we found that the Duration Paradox was observed in that instance as well.)
The implications are clear. While it is generally assumed that central banks set short rates and that the bond market sets long rates, as the central bank takes short rates towards zero, it gains more control over long rates, if only through the indirect and subtle nature of the Duration Paradox. Yet when the central bank also buys bonds outright in a quantitative easing program and foreign central banks buy bonds to keep their currencies weak, the net result is that bond yields can be far lower than where a true, rational, investor-determined free market would place them. This is likely the case today.
[Editor’s Note: The above essay is excerpted from The Amphora Report, which is dedicated to providing the defensive investor with practical ideas for protecting wealth and maintaining liquidity in a world in which currencies are no longer reliable stores of value.]