Diversification, a Dynamo for LDI

Diversification, a Dynamo for LDI

Time to read: Minutes
Brett Cornwell

Brett Cornwell, CFA

Client Portfolio Manager, Fixed Income

Oleg Gershkovich

Oleg Gershkovich, A.S.A., M.A.A.A.

LDI Solutions Strategist

Diversification can relieve long duration corporate bonds from being the only way LDI plan sponsors meet spread or duration targets. Complementing with Investment Grade Private Placements, CMLs and CMOs can help create a more efficient and effective portfolio.

Matching characteristics — not CUSIPs — can improve efficiency of risk compensation

Traditionally, plans seeking to reduce risk have turned to long duration investment grade public corporate bonds. However, simply investing in long-duration credit as an investible proxy for the liability discount rate is likely to cause funded statuses to deteriorate over time. While investors are subject to credit events such as downgrades and defaults, the discount rate is not. So, what does this mean?

The discount rate-setting process, as prescribed by the Financial Accounting Standards Board (FASB), makes it impossible to invest in a way that precisely tracks the movement of liability discount rates. If a plan sponsor invests in the identical group of bonds used to create the discount rate — i.e., CUSIP match — and one of those bonds is downgraded, the portfolio will decline in value due to the downgraded bond’s higher yield, but the discount rate, when reconstituted, will have no memory of that bond. The discount rate will not account for the bond’s higher yield following the downgrade and the liability may in fact increase when a downgraded bond is removed from the universe, unlike the portfolio.

Figure 1. For LDI investors, downgrades can weigh heavier over time
Cumulative total returns
Figure 1. For LDI investors, downgrades can weigh heavier over time

Source: FTSE, Bloomberg, Voya Investment Management. Return data is through 09/30/21. The liability return is calculated based on the FTSE Pension Discount Curve rate.

The good news for plan sponsors is that owning long duration corporate bonds is not the only way to meet spread or duration targets. As they de-risk and increase allocations to fixed income, there are other asset classes that can complement an existing long duration corporate bond allocation. For example, investment grade private placements (IGPP), commercial mortgage loans (CML) and securitized assets can offer a more attractive combination of yield and risk characteristics. These assets can also help investors diversify sector, industry, and issuer concentrations of LDI portfolios that focus on corporate bonds. In addition, a portion of a portfolio’s duration exposure can be obtained via high quality, government-backed, long-duration securities such as agency collateralized mortgage obligations (CMOs). We explore each of these non-traditional hedging instruments in more depth, and demonstrate how they can be used with a foundation of traditional long duration corporate bonds to create a more efficient and effective LDI portfolio.

Objectives and criteria for evaluating LDI diversifying asset classes

Introducing non-traditional fixed income assets can certainly diversify an LDI portfolio from a total return perspective. But if the goal is to minimize the tracking error of the portfolio to the liability, then sponsors need to be aware that there will be trade-offs to consider. While this can be complicated, having a simple framework will reduce complexity and help maintain consistency. Therefore, we believe it is important that any diversifying assets added to the hedging portfolio meet one or more of the following criteria: 

  1. Reduce issuer concentration: As plan sponsors de-risk, the first consideration is often increasing the allocation to public corporate credit. As the size of the hedging portfolio increases, so does the allocation to the same corporate bonds that dominate the public corporate benchmarks, resulting in highly concentrated positions. Introducing assets that are not included in the traditional long duration benchmarks can be advantageous when concentration risk is pervasive.
  2. Provide downside protection: As illustrated in Figure 1, credit ratings migration can have a meaningfully negative impact on funded status over time. Adding hedging assets that may provide a measure of downside protection via covenants, structure or collateral can ease downgrade risk and enhance the efficiency of the liability hedging portfolio over the long run.
  3. Exhibit reasonable correlation to the liability discount rate: Long duration corporate bonds have been a common tool to hedge pension liabilities because of their correlation to the discount rate, which is often a AA-rated corporate discount curve. If plan sponsors move a portion of the portfolio away from corporate bonds, they risk introducing tracking error between the assets and liabilities. Thus, we believe any asset under consideration for inclusion in the hedging portfolio should also exhibit a reasonable correlation to the discount rate in order to minimize unwanted tracking error.
  4. Minimize opportunity cost: Shifting the fixed income allocation away from corporate credit often entails an opportunity cost. Plan sponsors need to consider assets that have a yield or expected return that is equal to, or higher than, long duration investment grade corporate bonds, or meet one or more of the other key criteria listed above.

Take LDI to the next level: introduce non-traditional fixed income assets

Derivatives: can’t live with them, can’t live without them

Derivatives, while often misunderstood, are an indispensable tool in LDI strategies. We have seen time and again a rather brute force approach in which plan sponsors invest in long duration corporates and resign themselves to greater volatility of those long-dated assets for the sake of the liability duration match. But it doesn’t have to be this way. Our view at Voya is that it is better to harvest income in intermediate part of the curve and then use a derivative overlay to achieve the desired interest hedge ratio. The shorter spread duration provides a better Sharpe ratio than that of a long spread duration asset. Remember, the yield advantage of long duration corporate bonds can easily evaporate with a modest spread widening event while creating excess volatility in the portfolio. Shorter spread duration assets allow for more spread capture due to lower spread volatility. 

Investment grade private placements: the consummate surrogate 

Of all the non-traditional asset classes, investment grade private placements, or IGPPs, are the most natural fit within a liability hedging portfolio. They track well to the discount rate required in pension liability valuation because these assets are corporate spread products. Furthermore, IGPPs offer a yield premium compared to similarly rated public bonds of the same duration. The yield premium is bolstered by a higher upfront spread to public bonds for a given level of risk and covenant packages that provide an opportunity for lenders to be protected from downside credit events. Additionally, non-coupon income generated from pre-payment fees, amendment/waiver fees and coupon bumps are features unique to private credit that potentially offer attractive incremental income for investors. Thus, we believe IGPPs are complements to, not replacements of, public corporate credit (Figure 2).

Figure 2. IGPP diversification: higher up-front yields, back-end income and lower potential credit losses
Historical total return advantage over public corporate bonds
Figure 2. IGPP diversification: higher up-front yields, back-end income and lower potential credit losses

Advantaged total return chart reflects 01/01/01 – 12/31/20. Lower losses to comparable public corporate bonds. Source: Voya Investment Management. Performance as of 06/30/21 is based on since inception date 01/01/04. Gross returns are presented after all transaction costs, but before management fees. Returns include the reinvestment of income. Past performance does not guarantee future results.

We observe that generally an allocation to IGPPs of 10% of the hedging portfolio is a reasonable starting place. The allocation can be scaled up from there depending on funded status, desired hedge ratio levels, participant profile and other plan characteristics. 

Below we demonstrate the efficacy of coupling IGPP and public long duration corporate bonds with a duration overlay to match the duration of the Bloomberg Long Government/ Credit index. The outcome is that, over the long term, a portfolio with a mix of private and public corporates outperforms the Long Government/Credit benchmark. We consider IGPP, as a complement to public corporates, to be a compelling asset class in a long duration portfolio. Simply put, IGPP provides higher carry per year of spread duration and a higher carry-to-volatility ratio than the benchmark. 

Results from our analysis are illustrated in Figure 3. A few key attributes of private placements (over comparable public corporates) relative to the Long Government/Credit benchmark include:

  • Higher carry per year of spread duration equivalent to 32 basis points of incremental spread for the portfolio and higher option adjusted spread (OAS)/volatility ratio
  • More efficient utilization of risk budget, i.e., stronger standalone Sharpe ratio and positive information ratio
Figure 3. Coupling IGPP and public long duration with a duration overlay
Figure 3. Coupling IGPP and public long duration with a duration overlay

Source: Voya Investment Management, Bloomberg, Blackrock Aladin. 

1 Benchmark represents Bloomberg Long Government/Credit Index. 
2 As of 09/30/21. 
01/31/04 through 09/30/21; excess return represents total return less duration-related return.

Figure 4 shows that the excess return of the model portfolio significantly outperforms the benchmark, highlighting the efficacy of private placements in the portfolio. The portfolio exhibits a higher Sharpe ratio than that of the benchmark across most three-year rolling periods. Even during the extreme market downturns in 2008 and 2020, the portfolio delivered comparable excess returns and risk-adjusted returns relative to the benchmark.

Figure 4. Over the long term, a model portfolio with a mix of private and public corporates outperforms the benchmark
Cumulative portfolio and benchmark excess returns
Cumulative portfolio and benchmark excess returns
Rolling 3-year Sharpe ratio
Rolling 3-year Sharpe ratio

Source: Voya Investment Management, Bloomberg, BlackRock Aladdin. As of 09/30/21; rolling three-year periods from 01/31/04 through 09/30/21. Benchmark represents Bloomberg Long Government/Credit Index. Excess return is defined as the return attributed to carry and spread change and excludes any change in interest rates, i.e., duration neutral.

Comprehensive covenant packages may provide downside protection and minimize opportunity cost due to the spread premium over investment grade corporate public bonds. A modestly shorter duration may be introduced; however, shorter spread duration reduces spread volatility and allows for more spread capture, facilitating a potentially higher returning and more efficient portfolio over time. The shorter duration profile is also an organic match to key rate duration nodes (KRDs) in the belly of the curve for plan sponsors looking to hedge curve risk.

Finally, IGPPs also address concentration risk. Oftentimes a plan sponsor may have two or more long duration credit managers all benchmarked to the Bloomberg Long Duration Credit index or a combination of market-based public benchmarks. It is not uncommon that such a stable of managers will have overlapping allocations to issuers in those benchmarks. Introducing IGPPs diversifies the concentration risk inherent with those overlaps by providing an alternative set of issuers independent of any benchmark, thereby reducing idiosyncratic risk within the portfolio. With so many notable benefits, IGPPs are crucial for inclusion in a hedging portfolio.

Commercial mortgage loans: a match to pensions made in mortgage heaven

For decades, insurers have used commercial mortgage loans (CMLs) in reserving for their life annuity business. In essence, a defined benefit pension plan also is an annuity business, albeit a nonprofit one. Yet for years plan sponsors have shied away from fully exploring this asset class as a viable, non-traditional fixed income asset worthy of liability hedging portfolios. So why include CMLs? 

There are several features that make CMLs an attractive LDI hedging asset. They potentially provide a yield advantage over the life of a deal, achieved through higher spreads to public corporate bonds with similar ratings and duration profiles, “back-end” economics due to pre-payments and loan modifications, and structural protections that translate into lower losses and higher recoveries than public corporate bonds. 

Another attractive feature is the array of assets to choose from when allocating to CMLs. For example, we have the ability to target short, medium or long duration and fixed or floating rate assets. Thus, plan sponsors have options when optimizing duration needs and return objectives or to hedge KRDs across the curve. 

As shown in Figures 5, 6, and 7, CMLs meet several of the key objectives outlined above by reducing issuer concentration, lower correlation to traditional long duration benchmarks, minimizing opportunity costs through a yield advantage and providing potential downside protection through conservative underwriting and high quality collateral.

Figure 5. Commercial mortgage loans exhibit a total return advantage over corporates
Figure 5. Commercial mortgage loans exhibit a total return advantage over corporates

Source: Voya Investment Management. As of 06/30/21. Proprietary ratings model used to map commercial mortgage loans to a corporate bond rating equivalent. Core lending is for time period 01/01/02 – 12/31/20; transitional bridge is for time period 06/01/14 – 12/31/20. Lower losses to comparable public corporate bonds. Past performance is no guarantee of future results. Investors cannot invest directly in an index.

Figure 6. The favorable return-to-risk profile of CMLs is evident
Risk and return, 01/01/04 – 06/30/21
Figure 6. The favorable return-to-risk profile of CMLs is evident

Source: Voya Investment Management. As of 06/30/21. Past performance is no guarantee of future results. Investors cannot invest directly in an index.

Figure 7. Total CML portfolio return correlations, January 2000 to June 2021
Figure 7. Total CML portfolio return correlations, January 2000 to June 2021

Source: Bloomberg and Voya Investment Management. As of 06/30/21. Past performance is no guarantee of future results. Investors cannot invest directly in an index.

CMOs — an unchained asset class

The differentiated alpha sources inherent to the mortgage-backed securities (MBS) universe make a compelling case for adding agency collateralized mortgage obligations (CMOs) to traditional long duration portfolios that focus primarily on long corporate bonds. This asset class — which transforms U.S. residential and commercial mortgage cash flows into investor-friendly tranches by different weighted average life and duration characteristics — provides a bevy of investment options for LDI investors. Our focus is on long CMOs, i.e., those with an agency guarantee of principal and interest that exhibit low correlation to traditional long duration benchmarks (Figure 8).

Figure 8. CMOs can provide a bevy of investment options for LDI investors
Correlation based on excess return
Correlation based on excess return
Index characteristics
Index characteristics

Source: Aladdin and Bank of America/Merrill Lynch for the period 06/30/10 – 06/30/21. Index characteristics are as of 06/30/21. Correlations were calculated using monthly excess returns: Bloomberg U.S. Long Corporate index, Bloomberg U.S. Long Credit index, Bloomberg U.S. Long Government/Credit index, Bloomberg Long U.S. Treasury, and ICE Bank of America Merrill Lynch 10+ Year U.S. Agency CMO Excluding IO & PO index.

However, agency CMOs have had their share of drama since the pandemic, highlighting the negative convexity of the asset class and reinforcing why active management and security selection are so important in the space. The agency CMO and agency CMBS markets total over $1.9 trillion in outstanding debt. With mortgage rates plummeting to all-time lows in 2020, prepayments accelerated and the CMO universe contracted. By the end of 2020 the BoA/ML 10+ year CMO index consisted of only three securities with a duration of 6.8 years. As interest rates stabilize at higher levels and prepayments slow, we expect cash flows to extend and the creation of longer duration CMOs to expand. While market estimates vary, we believe that the long duration agency CMOs and CMBS of greater than eight years of duration equal approximately 10% of this universe, totaling roughly $190 billion in securities.

Applying the same framework for evaluating this asset class, issuer concentration is easily minimized because these assets are not part of the traditional long duration benchmarks. Since the collateral is government guaranteed, CMOs also provide downside protection and are thus immune from ratings downgrades. The trade-off is accepting a lower spread and yield compared to long-dated corporate bonds due to the government guarantee. While there is some correlation to the discount rate, it is lower than other spread asset classes, which introduces tracking error. However, these government-backed assets do provide spread and yield premiums to U.S. Treasuries of comparable duration, and thus can serve as a surrogate to Treasuries to help meet duration targets.

One final consideration is that, typically, conversations about CMOs tend to be with larger plans that have higher allocations to fixed income. These plans usually have significant exposure to corporate credit events, thus allocating to assets that can diversify the idiosyncratic risk is imperative. Furthermore, while IGPPs are a coveted asset class there are capacity considerations. For larger plans that seek to deploy larger allocations, it could take months, if not years depending on size, to put that money to work in IGPPs. By comparison, the CMO market is larger and able to allow more efficient deployment while still meeting some of the key objectives of diversifying assets. 

One approach we advocate is to combine CMOs with other securitized credit instruments to improve the return expectations — thus minimizing the opportunity cost — without eroding the diversification benefit. And by introducing some of these other structured credit products, correlation to the discount rate can improve compared to a pure-play CMO allocation.

The Dynamo in Action: if the proof is in the pudding, here are two recipes

In considering traditional and non-traditional asset classes within an LDI portfolio, we review two types of pension plans: a closed and frozen plan, and an open and accruing plan. We will use the FTSE AA Pension Discount Curve, an industry standard, to measure the liability. 

While plan sponsors have historically sought to mirror the performance of the AA-rated long duration corporate bond universe, perfectly replicating the cash flow discounting process is impossible. Investors are subject to principal losses from credit events, while the factors used to discount cash flows are not. To address this deficiency, we explore the inclusion of non-traditional fixed income assets to a traditional long government/credit allocation and quantify the impact on yield and tracking error to the liability. 

We begin by reviewing the asset classes under consideration and their characteristics. Figure 9 illustrates traditional long duration asset classes along with the non-traditional hedging assets discussed above. Notice how the duration of the non-traditional asset classes are in the intermediate range and the yield is lower than long corporate. This is because shorter duration products have lower yields since there is no term premium. However, as mentioned above, shorter spread duration reduces spread volatility and allows for more spread capture. Thus, when combined with Treasury futures to achieve a desired duration, shorter duration products can facilitate a higher returning and more efficient portfolio over time.

Figure 9. Traditional and non-traditional asset classes for LDI diversication
Figure 9. Traditional and non-traditional asset classes for LDI diversication

As of 07/31/21. 

Source: Voya IM, Bloomberg, and ICE Bank of America/Merrill Lynch.

Case Study 1 – A plan in run-off: the low liability growth rate

Setting the baseline

Starting with the closed and frozen plan, based on the aforementioned curve, the liability discount rate is 2.6%, i.e., the liability is growing at 2.6% per year, and the duration is 14 years. Using Voya’s proprietary LDI optimization tool, we constructed a correlation matrix of varying asset classes, yields and the plan’s participant cash flows to build an asset allocation that minimizes the tracking error between the assets and liabilities. The initial outcome is shown in Figure 10 as portfolio A and represents a traditional fixed income portfolio of 75% long corporate and 25% long Treasuries. This portfolio matches the growth rate of the liability (2.6%) with a tracking error of only 78 bp. 


Pension Metrics – Plan Without Accruals

Source: Voya Investment Management.

Introducing diversifiers

Once non-traditional asset classes are introduced to the model, such as IGPPs in the case of portfolio B, the balance instantly shifts 16% from long corporate and 3% from long Treasuries into IGPPs for a total allocation of 19% to the asset class. Note that the yield of portfolio B is unchanged but the tracking error is reduced by almost 30 bp. 

Expanding the tool-kit further, CMOs and CMLs are included, resulting in portfolio C, where the optimizer shifts 3% away from IGPPs into CMLs while ignoring CMOs. This is due to the optimization function of the tool, which is to produce the least-risk portfolio that meets the yield objective — in this case, 2.6%. Although CMOs mitigate credit ratings migration, the lower spread of CMOs compared to other credit-oriented assets does not help reduce tracking error to a credit-based liability. The CML, while a good fit, did not change tracking error or yield. However, it is an additional and viable asset class that can be used to diversify the portfolio. 

In the last stage, portfolio D, a 20+ U.S. Treasury futures contract is introduced to the LDI tool kit. In doing so two important outcomes can be achieved: the allocation to IGPPs reverts to 19%; and CMOs can be introduced to the portfolio because the duration can be adjusted via the use of Treasury futures. The duration adjustment to the portfolio improves the tracking error to the liability. Risks such as issuer concentration, idiosyncratic risk, downgrades and defaults are mitigated by including non-traditional assets that provide some sort of benefit in the form of covenants, structure, or collateral.

Figure 10. Closed and frozen plan characteristics

Source: Voya Investment Management. As of 07/31/21.

Case Study 2 — A higher liability growth rate

Next, we consider a pension plan that continues to accrue benefits. Here is a great example of how an accruing and open plan with a well-diversified portfolio coupled with a derivative overlay can reduce tracking error without giving up yield. This active plan has a liability duration of 12.2 years and annual service accruals of 1%. Half of this liability is comprised of in-payment annuities, and another 10% reflects a cash-balance plan structure that has a negative duration of about one year due to interest crediting rate sensitivity. Using the FTSE Pension Discount Curve and this plan’s participant profile we determined a discount rate of 2.85%. Accounting for the service accrual rate, the plan grows at about 3.85%. Thus, Voya’s LDI optimization tool is tasked with finding the lowest tracking error portfolio relative to a credit-based liability that can earn 3.85%. In addition, this plan is 111% funded and has a growth-seeking allocation of 25%. Adjustments will only be made to the remaining 75% of the portfolio, which includes the immunizing fixed income assets. However, the over-funded position of the plan provides added “natural leverage” because every dollar in assets covers 10% more relative to a dollar in the liability. Put another way, a 75% allocation to LDI assets reserves for 83% of the liability. 

Pension Metrics – Plan with Accruals

Source: Voya Investment Management.


Our initial step is to determine which traditional fixed income assets are a good match to the liability, (see Figure 11, portfolio A). The result is 67% in long corporates and 8% in 10–20 year U.S. Treasuries. Two things to note here that are further beneficial to this plan: the funded status is 111%, providing greater flexibility for our assets to keep pace with the liability; the interest rate hedge ratio (IRHR) is 110%. Given that 75% of assets are dedicated to LDI and an IRHR of 110% which, together with the over funding, provides an opportunity to “sweat” these assets and do more to address interest rate risk. However, this may not be the most efficient hedging portfolio, as we will see below.

Introducing diversifiers

As in the prior case study, opening up the hedging tool-kit and introducing IGPPs yielded interesting results; a nearly complete shift away from Treasuries and into IGPPs, as seen in portfolio B in Figure 11. 

Our optimizer preferred IGPPs due to the yield premium — despite its lower duration match to the liability — compared to the combination of 10–20 year Treasuries and long corporates. This reduced the interest rate hedge ratio to about 100%, with a modest improvement in tracking error of 13 bp, resulting in a more efficient hedging portfolio.

Taking this analysis further by introducing CMLs and CMOs (portfolio C), the optimizer shifts 8% from IGPPs and 1% each from long Treasuries and long corporates to CMLs. 

As in the prior case study, the lower yield and shorter duration was not viewed favorably by the optimizer. However, with the help of Treasury futures to adjust the duration profile (portfolio D), a modest allocation to CMOs (2%) can be introduced and also raise the IGPP allocation to 20%. This one step alone achieved a 75 bp reduction in tracking error from the previous stage and 100 bp reduction in tracking error relative to baseline traditional fixed income portfolio (Figure 11, portfolio A). 

Figure 11. Open and accruing plan characteristics
Figure 11. Open and accruing plan characteristics

Source: Voya Investment Management. As of 07/31/21.

Conclusion: the whole is greater than the sum of the parts

While this phrase is attributed to Aristotle with some debate, we can be certain that LDI was not an inspiration. However, it is apropos, as we have demonstrated that a range of traditional and non-traditional fixed income parts, when in combination, create a whole that goes beyond its parts. Through the lens of four key objectives to be considered when introducing diversifiers to a liability hedging portfolio, we established a framework for evaluating non-traditional fixed income asset classes such as IGPPs, CMOs, and CMLs and their applicability relative to a plan specific profile. In doing so, we were able to quantify the benefits of a well-diversified fixed income portfolio. And one of those key benefits is the reduction of tracking error to the liability without giving up yield; better yet, yield may even increase. 

Recent improvements in funded status have many sponsors nearing the end of their glide paths — increasingly closer to their “end state.” Hibernation strategies — whereby sponsors lock down their well-funded pension plan’s investment strategy to eradicate risk within the plan, secure their funded position, and “hold the line” as they wait to make their next move — are gradually becoming more prevalent. Once in these upper echelons of LDI investing and hedging, many sponsors invest like insurance companies; after all, a corporate pension plan is a nonprofit insurance subsidiary. In doing so they seek the very same fixed income duration products we have explored above.

It is clear to insurers, and a growing number of corporate plan sponsors, that diversification relieves corporate bonds of being the sole provider of duration by allowing other asset classes to complement them and contribute to the portfolio’s duration exposure. What’s more, with these combinations of diversifying assets, other benefits such as a reduction in issuer, sector and industry concentrations are realized. All of this comes together to demonstrate how diversification can convert mechanical energy into electrical energy, making it the dynamo LDI investing needs as it evolves in the next decade.


1  Benchmark represents Bloomberg Long Government/Credit Index. 

2  As of 09/30/21. 

3  01/31/04 through 09/30/21; excess return represents total return less duration-related return.

4 Long CMO is represented by the ICE Bank of America Merrill Lynch 10+ Year U.S. Agency CMO Excluding IO & PO index total return.


Past performance does not guarantee future results. All investing involves risks of fluctuating prices and the uncertainties of rates of return and yield inherent in investing. All security transactions involve substantial risk of loss. Voya Investment Management has prepared this commentary for informational purposes. Nothing contained herein should be construed as (i) an offer to sell or solicitation of an offer to buy any security or (ii) a recommendation as to the advisability of investing in, purchasing or selling any security. Any opinions expressed herein reflect our judgment and are subject to change. Certain of the statements contained herein are statements of future expectations and other forward-looking statements that are based on management’s current views and assumptions, and involve known and unknown risks and uncertainties that could cause actual results, performance or events to differ materially from those expressed or implied in such statements. Actual results, performance or events may differ materially from those in such statements due to, without limitation, (1) general economic conditions, (2) performance of financial markets, (3) interest-rate levels, (4) increasing levels of loan defaults, (5) changes in laws and regulations, and (6) changes in the policies of governments and/or regulatory authorities.