Can mutual funds limit pension risk as effectively as SMAs?

August 2, 2018

A/AText size:AAA
Brett Dutton

Brett Dutton
View bioRead blog

Separately managed accounts (SMAs) allow for more precise matching of key rate durations¹ than collective investment trusts or mutual funds, but do the benefits of this incremental improvement in precision outweigh the extra complexity and potential cost?

Brett Dutton, lead Investment actuary and head of Vanguard's Pension Strategy and Analysis team, explains, "A skilled investment advisor or manager could successfully use any of these vehicles to construct a portfolio suitable for most pension sponsors' liability-hedging objectives."

Yet SMAs are often seen as the premier choice for liability hedging. According to Mr. Dutton, that's likely due to some commonly held misconceptions.

One such misconception is that a liability-hedging portfolio should match the cash-flow timing of the liability—as closely as possible, and at any cost. This view leads to a preference for SMAs, which allow pension investors to buy and sell specific quantities of individual bonds, giving them the potential to match the liability's key rate durations, and even year-by-year cash flows, more precisely than they could with a blend of mutual funds.

In formulating this view, pension sponsors may assume that a closer match of any degree will have an appreciable effect on funding status volatility. But this is not necessarily true.—"For any investor, there is likely a point where the benefits of additional liability-hedging precision are marginal at best," Mr. Dutton explains. "One reason for this is that segments of the yield curve don't act independently of nearby segments. Rather, the curve tends to transition smoothly between maturities. For this reason, we often find that matching key rate durations with pinpoint precision isn't necessary for effective control of yield curve risk."

Risk reduction: Broad vs. granular custom liability benchmarks

Consider the two custom liability benchmarks in Figure 1. Benchmark A uses a mixture of fairly broad credit and Treasury indices. This benchmark could be easily implemented (with passive or active management, or a combination of both) using mutual funds, collective investment trusts, or an SMA. Benchmark B uses more granular allocations to Treasury STRIPs and, as such, it could only be fully implemented by selecting individual securities in an SMA.

Figure 1: Custom liability benchmark weights


Benchmark A (Broader allocations) Benchmark B (More specific allocations)
Long (10+ Year) A+ Credit 60.0% 60.0%
Intermediate (5–10 Year) Credit 15.0% 15.0%
Long (10+ Year) Treasury 7.0%  
Intermediate (5–10 Year) Treasury 18.0%  
0–5 Year Treasury STRIPS   9.7%
5–10 Year Treasury STRIPS   1.5%
10–15 Year Treasury STRIPS   9.4%
15–20 Year Treasury STRIPS   2.5%
20–25 Year Treasury STRIPS   1.9.%

Note: Benchmark B includes two public credit benchmarks and a custom Treasury STRIPs benchmark, which serve to fine-tune key rate duration exposures. This is a common construction for a liability-hedging SMA benchmark, which will often use public credit or corporate benchmarks as the portfolio's core and add customized exposures for precise liability hedging.

As you can see in Figure 2, the risk characteristics of both custom liability benchmarks are similar. However, the key rate duration match by maturity bucket is noticeably tighter to the liability in Benchmark B.

Figure 2: Sample benchmark characteristics

Sample benchmark characteristics chart

The difference in downside risk is slight

How much value does Benchmark B add because of its closer key rate duration match to the liability? Using an industry-recognized pension risk model, Mr. Dutton measured the value at risk (VaR), which is the amount a portfolio may decline in one year because of rising liabilities or declining assets, or both. The model indicates that the downside risk associated with either portfolio is less than 1% of the total pension liability: 0.58% for Benchmark A and 0.44% for Benchmark B.²

It's important to note that this VaR model primarily focuses on interest rate and credit spread risk, and it excludes risk factors such as: demographic or actuarial changes, discount curve measurement risk, and payment of expenses from plan assets. Each of these affects downside risk; so when accounting for these factors the overall risk would likely be greater than 1% for each benchmark.

"However, the important point remains true—the difference in risk between the portfolios would remain modest," Mr. Dutton says. "Furthermore, getting the overall risk level much closer to zero is practically impossible, regardless of the investment vehicle used, because the liability itself is not directly investable."

SMAs can add cost and complexity, not risk control

Achieving the closer key rate duration match that SMAs potentially offer likely comes at a cost. To trade specific quantities of bonds, a portfolio manager may need to navigate high bid-ask spreads (i.e., incur high transaction costs), especially for issues that are thinly traded and for segments of the market that may be capacity constrained, such as 10–20 year maturities. Additionally, the complexity of this approach requires the portfolio manager to oversee what could be hundreds of individual holdings for each client.

When compared with managing a mutual fund-based portfolio, which would typically have five or fewer funds, each with known sector composition and interest rate sensitivity, the costs and complexity of using an SMA can be significantly higher.

Furthermore, the relative simplicity and transparency of mutual funds offers substantial benefits for governance and fiduciary oversight.

"In my experience, many plan sponsors would view a marginal loss in precision as a perfectly acceptable exchange for the diversification, flexibility, and lower costs of using mutual funds to hedge pension liabilities," Mr. Dutton says.

¹ Key rate durations are specific levels of interest rate sensitivity at different points on the yield curve. Matching key rate durations helps protect a plan's funding status when nonparallel interest rate changes occur, for instance, if long interest rates fall while short interest rates rise, as happened in 2017.

² Value at risk (VaR) is the amount by which the plan's funding status may decline over a one year period, at the 5th percentile (i.e., in a once-in-20-years downside event). This analysis estimated value at risk using each benchmark and PFaroe Risk Analytics as of March 31, 2018. For illustrative purposes only. Not based on any particular portfolio.


  • All investing is subject to risk, including the possible loss of the money you invest. There is no guarantee that any particular asset allocation or mix of funds will meet your investment objectives or provide you with a given level of income. Past performance is no guarantee of future returns.
  • Diversification does not ensure a profit or protect against a loss.
  • Bonds and bond funds are subject to the risk that an issuer will fail to make payments on time, and that bond prices will decline because of rising interest rates or negative perceptions of an issuer's ability to make payments.
  • Costs are one factor impacting total returns. There may be other material differences between products that must be considered prior to investing.
  • Separately managed accounts (SMAs) are not mutual funds. These investments are available only to tax-qualified plans and their eligible participants. Investment objectives, risks, charges, expenses, and other important information should be considered carefully before investing.
  • Advice services offered through Vanguard Institutional Advisory Services are provided by Vanguard Advisers, Inc., a registered investment advisor.
  • Certain projections or other information regarding the likelihood of various investment outcomes are generated by modeling functionality (the Model) made available through PFaroe®, an application licensed by Vanguard from Risk First. These projections are hypothetical in nature, do not reflect actual investment results, and are not guarantees of future results. Model results will vary with each use and over time.
  • The Model's projections are based on a statistical analysis of historical data. Future returns may behave differently from the historical patterns captured in the Model. Moreover, the Model may underestimate extreme negative scenarios unobserved in the historical period on which the Model estimation is based.
  • The Model forecasts distributions of future projections for a wide array of risk factors. The risk factors modeled include interest rates, credit spreads, equity markets, inflation, foreign exchange, and a selection of alternative investment classes. The stochastic projections of assets and liabilities are then generated based on the underlying economic scenario generator and calibration thereof. Assets and pension liabilities are projected and repriced consistently. Future liability values are projected in line with the applicable underlying assumptions and future asset values are projected based on the defined investment strategy, where both allow appropriately for benefit payments and contributions.
  • Risk First maintains a Model Risk Management Policy and associated procedures to which any changes to the Model will be subject.