(This is the second part of a discussion of how long-term investors should view bonds. The first part is available on Seeking Alpha. I want to repeat one point here – I am looking at what should be the default weighting for “bonds” within the portfolio of an investor with a long investment horizon. For example, a young person who has built up a cash buffer already, and is now building up a retirement portfolio.)
In the previous article, I argued that the problem with bonds is that long-term bond yields are quite a bit lower than expected long-term equity returns. I argued the main reason to hold long-term bonds is the uncertainty about those equity market returns. I will now list a few other reasons to hold bonds. (One other issue that is the question of taxation. In many jurisdictions, equity returns are privileged over bond returns by the tax system. However that is a very country-specific discussion, and tax shelters can nullify this disadvantage for bonds.)
Negative Return Correlations Should Persist
From the point of asset allocation, whether government bonds and equities will continue to have negative correlations is a key question. This property implies that if equity prices fall during a period, it is likely that government bond prices will rise, and vice-versa. (Note: I am speaking here about developed market governments with control over the currency, like the U.S. and Canada. Euro peripheral bonds have switched over to have a positive correlation with their stock markets.)
The chart below shows the quarterly changes in the price of the S&P 500 equity index versus the change in 10-year Treasury yields (from the Fed H.15 release) since 1995. The positively sloped trend line implies a negative returns correlation – since Treasury bond prices and yields move in opposite directions (“yield up, price down”). The relationship is weak for small changes - the scatter is a cloud of points in the middle - but it holds strongly for the quarters with the most negative equity returns (lower left quadrant). (Note that the period chosen was not random, as the correlation has changed over time. In particular, it had the opposite relationship in the 1970s.)
This is an important but complicated issue, and I cannot cover it within this article. As a result, I am going to assert that the negative return correlation seen in the last couple of decades should continue to hold, at least when it matters (when equity prices drop quickly).
To give a quick explanation of why I believe that returns correlation will remain negative, I appeal to the insights of the economist Hyman Minsky. The private sector has an innate tendency to make increasingly risky financial arrangements as the economy expands. Eventually, the private sector ends up over-extended, and the financial sector comes under strain. This scares equity investors, while the central bank is forced to cut rates (lifting bond prices). And note that this is not just based on the last crisis – all of the Fed tightening cycles in the last couple of decades have been ended by some sort of a financial crisis.
Simultaneous (significant) falls in stock and government bond prices would seem to require an inflationary environment, such as we saw in the 1970’s. I do not see such a change of regime as being a high probability event over a reasonable forecast horizon.
The implications of a negative return correlation are:
- the volatility of the portfolio value is reduced if we own assets with a negative return correlation;
- portfolio rebalancing (slowly) enhances returns.
The first point is a fairly standard observation, but I am unconvinced that we really should care about the volatility (standard deviation) of returns. That said, it's probably the best proxy available for what we are really worried about - a more vague concept of "risk". The second point is more interesting. As long as stocks and bonds move inversely, rebalancing towards target levels implies a mechanical value strategy - sell the asset that outperformed, buy the one that underperformed. (I may discuss this concept in more detail in a later article.)
Government Bonds Hedge (Most) Economic Risks
The key assumption behind being able to focus on long-term returns is that you do not need to touch your portfolio in the short term. However, you may be forced to do so as the result of unexpected developments. And the most likely reason for this happening is losing your job. Statistically speaking, you are most likely to lose your job in a recession, which is when government bonds do well. (Things such as illnesses are presumably not correlated with financial markets.)
Government bonds are going to be hurt by inflation, as is well known. It is possible that the inflation could rise due to supply side constraints, notably energy costs. That sort of inflation is hard to hedge (beyond buying commodity-related equities). However, generalised inflation can normally only be sustained if wages are rising. Being employed is a much better hedge against inflation than a portfolio of financial assets. (Owning a home can be viewed as an inflation hedge, but I give reasons in this article why I am cautious about housing as an asset.)
Overall Government Bond Indices Do Not Have Long Maturities
Many investors hold bonds in the form of mutual funds or ETF’s. In many cases, these are based on the overall bond index (typically all bonds with maturities over one year). Although it depends on the index, the average maturity is typically less than 10 years.
The implication is that long-term bond returns are more uncertain than was implicitly implied by the first article. For example, we would have to roll over a 10-year bond twice if we want to look at a 30-year return horizon. The yields at rollover could be much higher than the current yield level. Thus, we are no longer sure that very long-term government bond yields will be as low as is implied by the current yield curve.
Investment Grade Corporates Should Enhance Returns
Many bond funds do not just hold government bonds, they also hold investment grade corporate bonds (and in Canada, Provincials). (Investment grade is typically defined as bonds with a rating of AAA,AA,A, or BBB.) These bonds have a slightly higher yield than governments (the “spread”), which should enhance the returns over time. (You cannot be certain by how much, as there will be credit losses over time, usually in the form of price drops as bonds are downgraded to "junk" and fall out of the index.)
Moving To Higher Risk Categories Within Bonds
(This point was also raised by commenters on the first article on the Seeking Alpha site.)
I believe that investors should base their policy portfolio (baseline allocation) using a simple breakdown like Stocks-Bonds-Cash. E.g., a common baseline is the 60:40 portfolio – 60% equities, 40% bonds.
The reason to do it that way is that it is easier to understand the qualitative aspects of your baseline allocation. And if you fix the Cash weighting, you only have one free variable in your choice of weightings - once you set the Equity weighting, the Bond portion of the portfolio represents what is left over. You can more easily understand how portfolios behaved historically as you shift that single parameter. (Even though past results are not predictors of future performance, you should at least have a qualitative idea how a portfolio weighting behaves under different market regimes.) I have my doubts that it is possible to really understand the trade-offs that occur in a policy portfolio which includes a dozen asset classes (many of which do not have long runs of historical data).
But once you move from the policy portfolio analysis to reality, you need to decide which specific assets to hold. I would then suggest carving out portions of the “basic” assets to more specific asset classes, but with the idea that the aggregate risk characteristics will be similar. For example, 10% of your “equity” weighting may end up in Emerging Market equities. (This means that if you have a formal policy portfolio guideline, you may end up with an "official policy portfolio" with a dozen asset classes in it, and an "analysis policy portfolio" which was used to determine the risk characteristics of the "official policy portfolio". For the purposes of this article, I am talking about the "analysis policy portfolio".)
And so one way to mitigate low government bond yields is to shift some of the “bond” weighting over to riskier related assets like High Yield bonds. Although you should expect to suffer some credit losses over time if you hold High Yield bonds, you can hope that these losses will be covered by the extra yield they provide. (Disclaimer: I do not follow the High Yield market, and I have no idea whether this will be true or not going forward.)
These riskier assets will act as a mix of stocks and bonds, so I would adjust weightings accordingly. For example, a 10% High Yield weighting could be considered to be really a 5% “bond” and 5% ”equity” weighting when comparing to the policy portfolio target. High Yield bond returns will be correlated with equities, so you will lose some of the diversification benefits government bonds bring. (Investment grade bond spreads will probably not move by enough to lose the diversification benefit.)
Policy Portfolio Inertia Is A Good Thing
Although laziness is not normally considered a good quality, most people probably should aim to keep their portfolios near a policy weighting over time, and not allow those target weightings to swing wildly in response to market fads. A lot of trouble would have been avoided if people had rebalanced away from equities towards a more sensible portfolio weighting in the late 1990’s.
However, a tendency toward inertia means that people should avoid making drastic changes in response to what may be a temporary condition, as the decision may not be undone if the condition goes away. E.g., even if you convince yourself that a 100% equity weighting makes sense based on the low level of bond yields now, it may be a bad idea to hold that stance for 10 years if bond yields rise to "attractive" levels.
Yield Enhancement Via Currency
In some cases, there may be a reason to invest in foreign currency bonds. For example, a great number of retired Canadians travel to the U.S. during the winter, and so they will expect to need U.S. dollars in the future. And you may be able to convince yourself that there is a chance that the foreign currency will appreciate versus your home currency in the long run. In which case, the returns from the foreign bonds when valued in your home currency may look adequate as a result of this enhancement. (Disclaimer: I have bought U.S. Treasury bonds based on this logic in the past; the rapid drop in the Canadian dollar makes the logic less compelling at present.)
Foreign currency bonds also act as a form of insurance against your home currency losing a lot of value on foreign exchange markets (possibly as the result of inflation). This probably makes more sense for citizens of smaller countries; the U.S. is so large that it will drag everything down with it.
There are a variety of reasons to hold bonds as the result of their diversification value, but the potential drag on performance cannot be ignored. As always, the answer depends upon the individual investor's preferences with regards to return targets and risk tolerance.