The majority of liability-driven investing (LDI) portfolios employed by private defined benefit plan sponsors include some combination of long-dated credit and long government (or Treasury) bonds. The optimal weighting between the two components depends on a number of plan-specific factors, including the nature of discount rate methodology and the plan’s asset allocation, as well as the duration and curve exposure of the liabilities. Whatever those weightings may be, the decision to combine credit and government bonds in blended LDI mandates (the bundled approach) or to separate them in sector-specific mandates (the unbundled approach) is one of the most consequential decisions that plan sponsors will have to make.
Dedicated passive (or low-discretion) long government mandates typically carry a management fee that is lower than that of actively managed LDI portfolios. Thus, the proponents of unbundling have suggested they could reduce the overall investment management fee on a plan sponsor’s LDI allocation by carving the long government bond allocation out of the actively managed LDI portfolio. It’s critical, however, to balance potential fee savings against the implications of unbundling. In the final analysis, we believe that the opportunity costs associated with unbundling far outweigh the potential moderate fee savings.
One enduring myth often used to justify an unbundled approach suggests that separating the management of long government bonds from credit allocations in sector-specific mandates has no excess return (or alpha) implications. That thesis argues that because active management of long government bonds in isolation has limited potential to generate excess returns (or alpha), then carving them out of the actively managed bundled long duration or LDI portfolio will have little to no alpha implications and therefore no overall return implications. We strongly dispute that assertion.
Opportunity costs associated with unbundling far outweigh potential fee savings
It may be true that the prospects for adding value when managing long government bonds in isolation are limited. With the lack of credit risk and a limited set of issuers, the manager certainly has fewer levers to pull. However, when those bonds are instead part of a bundled portfolio that includes a significant share of non-government holdings, like a long government/credit portfolio or a customized LDI portfolio, an active manager like PIMCO can seek to achieve an alpha target commensurate with that of an actively managed LDI strategy on the entire amount invested in that portfolio, including the dollars invested in long government bonds. In other words, we can implement largely the same top-down and bottom-up active positioning, both directionally and in magnitude, whether the actively managed LDI portfolio benchmark is 50% credit/50% government, 75% credit/25% government, or 100% credit. That is, if we want to be 2% overweight a certain issuer, 1% unaderweight another, and run a longer-duration posture by 0.5 years, we can implement that positioning in any of the three portfolios above and ultimately seek the same excess return.1 On the other hand, when a plan sponsor moves long government bonds from a bundled approach to an unbundled one – where long government bonds are managed passively or in a low-discretion fashion – they forgo a large chunk or all of the alpha potential on the dollars moved.
This opportunity cost can be significant. Take for example a plan sponsor with a $1 billion bundled LDI portfolio actively managed with a 100 basis point (bp) alpha target. The portfolio is composed of 55% long credit bonds and 45% long government bonds (roughly equivalent to a portfolio benchmarked against the long government/credit index). In this situation, the expected annual alpha in dollars is $10 million ($1 billion portfolio value multiplied by the 100 bps alpha target). Let’s now suppose that the plan switches to an unbundled approach and carves out roughly $450 million from the bundled portfolio and moves it into a passive long Treasury or Treasury STRIPS strategy. This would amount to a $4.5 million annual opportunity cost ($450 million multiplied by the forgone 100 bps alpha target), which would reduce the plan’s expected alpha on this portfolio from $10 million to only $5.5 million (see Figure 1). In most circumstances that opportunity cost is far higher than the potential fee savings associated with unbundling (which we estimate at about $0.8 million in the scenario above). In other words, for every $1 of potential fee savings achieved through unbundling, this plan would give up more than $5.5 in excess return potential.
Burden of rebalancing introduces operational hurdles and higher transaction costs
Because the realized performance difference between long credit bonds and long government bonds or long STRIPS can be quite large even over short periods of time, unbundled LDI portfolios will need to be rebalanced frequently to maintain the desired credit versus government mix (see Figure 2).
This explicit rebalancing requirement introduces a number of disadvantages and challenges compared with a bundled approach. Indeed, plan sponsors or their advisors will need to dedicate resources to continuously monitor the credit versus government exposure at the overall fixed income portfolio level and frequently execute reallocations between the different pieces. In addition to the associated operational burden, those reallocations will generate incremental transaction costs that can offset the potential savings in investment management fees resulting from carving out the management of long government bonds. Also, unbundled approaches often require monitoring a larger number of investment managers and accounts, which could further reduce potential net fee savings.
Finally, for those with meaningful STRIPS allocations, the liquidity and transaction cost considerations linked to rebalancing needs may be magnified given the unique nature of the long-dated STRIPS market. Indeed, because of its relatively small size ($158.5 billion in market value eligible for the Bloomberg Barclays U.S. STRIPS 20+ Year Index) and the potential to explicitly create or eliminate new securities (through stripping or reconstitution), the long STRIPS market imparts a number of inefficiencies. Passive management may leave investors on the wrong side of those inefficiencies.
Timing of implementation appears challenging
The recommendation to carve out long government bonds is often accompanied by a suggestion to convert them, at least partially, to a long-dated STRIPS allocation. While STRIPS may offer an efficient means to achieve significant duration exposure with a limited capital commitment and improve duration hedge ratios, it is paramount to consider the timing of implementation. Indeed, because long STRIPS constitute a very concentrated exposure, they can experience substantial underperformance (both in the absolute and relative to liabilities) if implemented at an inopportune time.
With 30-year U.S. Treasury rates currently lower than they have been for 93% of the time over the last 10 years and 95% of the time over the last 15 years, and with a curve that is relatively flat, there are a number of plausible scenarios that could lead to significant long STRIPS underperformance.
For example, in a scenario where 10-year rates rose by as little as 50 bps, and the curve moved from its current level of steepness (about 60 bps between 5-year and 30-year rates) to just its average level over the last 10 years (about 160 bps between 5-year and 30-year rates), the 20+ year STRIPS index could lose about 23% to 28% depending on the horizon over which this happens (see Figure 3). This represents a sizeable loss in the context of a relatively mild rate increase/curve steepening environment.2
That very pronounced negative performance in absolute terms would also likely exceed the decline in the value of liabilities under the same circumstances by a wide margin (because duration exposure of liabilities is less concentrated at the long end of the curve and also because the credit spread exposure embedded in the liability discount rate may provide some buffer).
Minimizing investment management fees may be a legitimate objective for plan sponsors. However, with unbundled LDI approaches, the potential reduction in expenses comes with a number of less desirable implications, such as significant opportunity costs, diminished excess return potential, and increases in operational burden or implementation complexity that far outweigh the limited fee savings.
While there can be no assurance that the investment approach outlined above will produce the desired results, we believe that the interest of plan sponsors and participants would be best served by maintaining a bundled approach for their long duration or LDI mandates. Doing so could achieve a number of advantages, including:
- Maintain excess return potential on the entire long
- Efficient automated rebalancing embedded in the bundled approach potentially reduces both the plan management burden as well as transaction costs
- Help avoid increasing allocations to long-dated STRIPS at an inopportune time, given the current relatively low level of interest rates and flat interest rate curve posture
- Avoid passive management in a market sector that is more likely to be inefficient given its limited size and ability to explicitly create or remove securities
We believe plan sponsors who are among the large crowd that is already employing a bundled approach should stay the course. On the other hand, we encourage those who currently use an unbundled structure to consider, analyze, and quantify the potential benefits of converting to a bundled structure.