Book Excerpt (from Section 4.8)
These concerns about hedging efficiency [discussed in the previous article] will then rebound to how inflation-linked bonds fit into portfolio allocations. Since a full overview of portfolio allocation strategies is well beyond the scope of this text, I will confine my attention to classical non-levered portfolio allocations. By contrast, if we are willing to use leverage, inflation-linked bonds are more interesting and are used in constructions like risk parity strategies.
The following discussion is based on conventional portfolio principles. The idea is that we are not going to try to shift our portfolio in an erratic fashion in order to optimise returns. Instead, we assume that markets are somewhat efficiently priced, and we are trying to find a default portfolio mix that best fits our investment objectives. For example, in personal finance, a 25 year-old is typically assumed to be aiming at achieving high long-term returns (at the cost of higher portfolio volatility), whereas an 80-year-old retiree is presumed to be aiming to achieve a steady income stream. We then attempt to find a benchmark weighting for a number of asset classes. We can then vary our actual portfolio based on market views, but our risk is measured relative to our deviations from that benchmark weighting. For our discussion here, we are interested in what the benchmark weighting for inflation-linked bonds should be.
I will start with a standard baseline portfolio strategy: the 60/40 construction – 60% equities, and 40% fixed income. This was a good approximation for many institutional portfolios historically, and a common recommendation for middle-aged individuals. (Now that many pension funds are in run-off mode, I assume that there is a greater variety of investment stances.) We then want to adapt the portfolio to include inflation-linked bonds.
In a 60/40 portfolio, equity risk dominates. Not only do equities feature higher raw returns volatility, the equity weighting is 150% of the fixed income weighting. Prospective returns are heavily dependent on the equity market performance.
From an active bond manager’s perspective, inflation-linked bonds provide an amazing expansion of the set of opportunities for outperformance. It might be possible to build an entire investing platform around using inflation-linked bond relative value trades to beat a conventional bond index. However, the usual target for outperformance for an active non-levered fixed income manager is typically on the order of 20 basis points per year. From an asset mix perspective, adding 20 basis points of outperformance on 40% of your portfolio is nice to have, but it is laughably small influence on overall returns when compared to equity market risk. (The 60% equity weighting might gain or lose the same dollar amount within 20 minutes when the exchanges are choppy.)
The question from an asset mix perspective is: what weighting should we give to a dedicated inflation-linked bond holding? Unfortunately, there are difficulties in trying to carve away assets from the other asset classes.
- If we reduce the conventional fixed income weighting in an inflation-linked portfolio, we are reducing the limited ability of our bond portfolio to diversify the risk asset exposure of equities. As we saw in the Financial Crisis, inflation-linked bonds can fall in price at the same time as equities; only conventional government bonds unambiguously did well during the repricing. Furthermore, conventional bond indices typically have a weighting to corporate credits that are expected to outperform government bonds over the long term (since corporate spreads have been historically wider than default losses).* As noted in Section 4.6, inflation-linked bond indices are weighted towards central government issuers, and so the credit exposure of the portfolio would drop (without making changes to the indices used).
- If we carve out the equity weighting, we would very likely be reducing expected returns on the portfolio. So long as we assume that inflation breakevens are not totally mispriced, an inflation-linked portfolio would be expected to have a return within 100 basis points of a conventional bond portfolio with a similar duration (based on prospective inflation volatility being close to historical averages). Meanwhile, asset allocators generally expect equities to outperform bonds by hundreds of basis points. (Historically, they were too optimistic in that regard, but the bull market in bonds has largely crashed into the zero bound in rates.) Under such a return assumption, the prospective return of the portfolio would drop markedly and would be resisted by most asset allocators.
Given the trade-offs noted above, one can see that it will be a tough sell to get a benchmark inflation-linked weighting much greater than 10%. Even at just a 10% weighting for inflation-linked bonds taken out of the bond weighting, we might lose about one quarter of our risk asset diversification (depending upon our assumptions on how returns will be correlated in a crisis). However, as noted in the discussion of hedging efficiency, that only provides a hedge against inflation for 10% of the portfolio. Although that is a step in the right direction, it is certainly not going to result in being completely unconcerned about future inflation trends.
Of course, if one is somewhat more risk-seeking, one might replace the conventional bond portfolio by inflation-linked bonds based on the view that inflation-linked bonds are fundamentally cheap. However, in an institutional setting, it would be necessary to convince an asset allocation committee to make a big bet based on a relative value view, which is often quite difficult. In any event, it is hard to describe such a stance as a baseline portfolio construction, as it is unclear why one should believe that inflation-linked bonds are always going to be structurally fundamentally cheap. (Risk assets are presumed to have higher prospective returns to compensate for illiquidity or volatility.)
Commentators of a more bearish temperament may view the higher returns expectations for equities with derision. However, if one is managing a portfolio for retirement purposes (either as an institution, or for yourself), there is a cost to using low returns expectations. In order to meet retirement income targets with lower expected returns, it is necessary to raise contributions. Workers might not be happy with extremely large obligatory deductions from their pay. For an individual, a high implied contribution might lead to disenchantment with financial planning and instead lead to speculative activities. (One standard depressing result of personal financial surveys is the percentage of people that indicate that they expect to fund their retirement with lottery winnings.)**
It is much easier to find a place for inflation-linked bonds when the default bond/equity weighting is tilted more towards bonds. This might be the stance of a closed pension fund, or the recommended asset allocation for a retiree. For example, if the default bond weighting is 60%, we would have a 150% relative weighting towards bonds over equities. We no longer need to prioritise controlling equity returns volatility. Meanwhile, since the role of equities in a bond-heavy portfolio is presumably to hedge against inflation risk, inflation-linked bonds may be a better hedge vehicle.
In summary, inflation-linked bonds seem best suited as a vehicle for making relative value trades (breakevens are cheap/expensive), or as a fixed component of portfolios that are already heavily weighted towards bonds. The evaluation changes if we are willing to build our portfolio strategy around derivatives or leverage, but we would then need to understand the issues raised by such investment structures.
Breakeven Inflation Analysis is available at the stores linked here.
* Actual corporate bond relative performance has been underwhelming in practice. My reading of the data was that credit products tended to have embedded calls that ended up being a major performance drag in a secular bull market in interest rates.
** My discussion of returns expectations is following conventional logic. I am unconvinced that this is the best strategy for personal finance. My preference is to follow the “pay yourself first” strategy, and immediately save a decent chunk of each paycheque. You will find out what your sustainable retirement income is the hard way – when you retire, you see how large your portfolio is. However, such a viewpoint cannot be translated into an institutional framework. It is essentially equivalent to saying that we should fire all the actuaries, and set pension contributions by instinct. Although I believe that we cannot avoid fundamental uncertainty, we still have to pretend that we can forecast the future when setting pension policy.
(c) Brian Romanchuk 2018
What do you think about using very long duration inflation linked bonds to effectively give you higher leverage. Does this improve the prospects of including them within a portfolio?
And same question for the nominal bond portion - if you use 30% 30 year nominal bonds and 10% 30 year ILBs would that not give you both better protection in environments of both falling growth AND rising inflation without taking away the potential returns from the equity portion?
From a portfolio allocation standpoint, you want as long a duration in index-linked as you can get. The issue is that the inflation protection is equal to 100% of market value, and so if it’s 10% of the portfolio, that’s still only 10% of the portfolio with inflation protection - no matter what the maturity is. The longer maturity helps out in relative value terms, but over the long term, those will be averaged out.Delete
You can try leveraging the linker portfolio, but you need to then keep in mind the many victims of that strategy in 2008.