Liabilities and LDI: How Un-Investable Is Un-Investable?

Transcription

1 leadership series INVESTMENT INSIGHTS October 2014 Liabilities and LDI: How Un-Investable Is Un-Investable? For plan sponsors who wish to preserve funded status in a defined benefit (DB) plan, we strongly advocate the adoption of liability-driven investing (LDI) strategies, including a glide path approach toward continued risk reduction over time. However, as plans progress along these paths, it becomes important to properly calibrate the levels of tracking error (TE) a plan sponsor should expect at each stage. In this article, we present a qualitative review of the factors that can contribute to TE, and suggest that these factors combine to generate a realistic floor of 3% to 4% TE in a direct-to-the-liability mandate that is designed to keep pace with the liability. It is well known that the un-investability of the discount rate curve used to value the liability may contribute to TE for an asset portfolio seeking to match that liability. The most significant reason for this is the differing treatment of downgrades and defaults, which can create volatility and permanent loss. In addition, concentration issues, liquidity and availability of seasoned bonds, trimming procedures to enhance liability discount yields, and vendor pricing basis can also contribute to higher tracking error. However, plan sponsors should not be discouraged by this potential TE. In fact, a meaningful portion of a pension plan s funded-status risk can be reduced through straightforward LDI measures, and many plan sponsors have experienced the benefits of these strategies over the past few years. The risk management issues discussed here may be most noticeable at the advanced stages of an LDI implementation, when plan sponsors are seeking to achieve the tightest fit of assets to the liability. The challenges of un-investability are manageable, but we recommend that plan sponsors, members of the plan committee, and other stakeholders understand the realistic levels of TE that may remain even after moving to a high LDI allocation. Pension risk can be reduced before un-investability issues surface To begin, the concerns discussed in this article should be put in perspective. In our experience, a pension allocation that is split between 70% return-seeking assets and 30% core bonds typically results in tracking error to the liability of 13% to 17%. A meaningful reduction in the return-seeking allocation (shifting the portfolio from a 70% return-seeking and 30% hedging split to something more like 35% and 65% respectively), combined with a duration extension of the bond portfolio, could reduce this TE to between 5% and 7% a significant reduction in funded-status risk even before implementing the later stages of an LDI strategy. How is this risk reduction achieved? Because the return-seeking allocation is typically the largest source of tracking error in the plan, TE can be reduced through an LDI implementation simply by lowering that exposure. 1 Hedging active interest-rate risk by extending dollar duration in the bond portfolio can effectively lower TE as well. However, tightly hedging the credit portion of the liability risk can prove more challenging as we describe later in this article, this is where we believe the majority of the TE floor is generated. As a result, plan sponsors encouraged by the considerable degree of risk reduction in the early and middle stages of an LDI implementation should not be surprised if TE stabilizes (at a low level) in the later stages. Dan Tremblay, CFA Director of Institutional Fixed Income Solutions & LDI Strategist François Pellerin, CFA, FSA, EA, CERA LDI Strategist Ben Tarlow, PhD Portfolio Manager KEY TAKEAWAYS LDI strategies can greatly reduce the tracking error of a DB plan s assets to its liability, but the un-investable nature of the liability itself tends to add structural TE. Plan sponsors should expect a minimum of 3% to 4% TE in a well-matched LDI portfolio, and more may be acceptable depending on circumstances. Credit events may have a different effect on the liability valuation than on asset portfolios, contributing significant amounts of TE. Other factors, such as concentration risk mitigation, varying supply and demand dynamics across the curve, the use of custom yield curves, and differences in pricing sources may also contribute to TE. For later-stage LDI implementations in particular, a reasonable level of TE may be natural for an asset portfolio designed to keep pace with the liability while managing risk. For institutional use only

2 Liability curves are not designed for investing We would highlight that a pension plan s liability may be generally un-investable, because it is valued using a yield curve designed for accounting and funding purposes, not for investing. Each liability yield curve has its own calculation methodology for specifying a yield at each maturity node in the curve that is needed to discount future pension obligations. However, a liability yield curve is often based on yields that are theoretically available in the market, rather than exclusively on the bonds that are actually available to an asset manager. As a result, the difference in returns between the liability yield curve and an actual asset portfolio seeking to match that liability may be significant. 2 To quantify this dynamic, we developed a case study to calculate monthly changes in performance for liabilities and assets over the past 10 years (120 months), ending with Aug Exhibit 1 (right) shows the frequency of one-month performance differences of various magnitudes during that period. The liability performance was calculated by discounting the cash flow obligations for a hypothetical fully funded pension plan, using the Citigroup Pension Discount Curve (a full spot curve of liability discount rates based on a universe of AA-rated corporate bonds), where the overall duration was comparable to our asset proxy, the Barclays Long Corporate AA or Better Index (which represents the diversified universe of bonds most asset managers could invest in for LDI-based mandates). As shown, the most common frequency of observations fell within plus or minus 25 basis points. However, this tight fit occurred only 30% of the time (36 out of 120 observations). Notably, there were seven occasions when the difference was greater than 200 basis points of asset underperformance, with three observations ranging from 450 to 500 basis points of monthly underperformance! Four of the six worst observations occurred during the credit crisis, when credit spread changes were well above average, but there are still instances where the differences were material well after the credit crisis abated. At the other end of the spectrum, there were four instances when assets outperformed liabilities by more than 200 basis points (with the largest outperformance greater than 700 basis points), and 35 instances of outperformance by more than 50 basis points. Although outperformance of assets is typically a welcome event, the frequency here reinforces the point that tightness of fit can be challenging to maintain, regardless of the direction of the difference. EXHIBIT 1: The frequency with which typical plan assets have underperformed and outperformed (out of 120 observations) a typical liability by 50 or more basis points suggests that very low TE might be an unrealistic goal. ESTIMATED ONE-MONTH PERFORMANCE DIFFERENCE: Number of Observations ASSETS MINUS LIABILITIES (120 OBSERVATIONS) < to to to to 50 Assets represented by Barclays Long Corporate AA or Higher Index; liabilities represented by a hypothetical plan liability discounted by the Citigroup Pension Discount Curve. Past performance is no guarantee of future results. All indices are unmanaged. It is not possible to invest directly in an index. Data from Aug. 1, 2004, to Aug. 31, Source: Barclays POINT, Fidelity Investments. One potentially positive aspect of this short-term volatility is that return differences may tend to be mean-reverting, summing closer to zero over time and thus helping to offset some of the more severe monthly discrepancies. Nevertheless, despite the potential for mean-reversion over longer periods, a moderate short-term performance differential is possible, even for a suitable LDI portfolio. Exhibit 2 (see page 3) shows the rolling 12-month performance differential using the same assumptions as the case study above. Note the volatility between the two indices, with performance gaps ranging from 6.5% to 10.7% (in early 2009, following the financial crisis). If the timing of any 12-month discrepancy were to coincide with the end of a reporting status, a plan could experience a reduction in mark-to-market funded status, with all the potential consequences a drop may entail (financial statement volatility, required contributions, PBGC fees, etc.). 3 Even if the extreme monthly differences seem less likely to occur in the current environment of lower volatility and tighter spreads, consider the meaningful number of observations that were between 50 and 200 basis points in either direction 49 instances, or 40% of the time and remember that these are monthly measurements. The cumulative effect of several months of moderate underperformance, even in a quieter market environment, could affect funded status. Although such swings are likely to be much smaller than a plan may experience without an LDI implementation, plan sponsors should be aware of the possibility. However, this potential for short-term TE makes the evaluation of an LDI strategy s effectiveness more challenging. Faced 36 +/ to to 100 Basis Points Difference to to > 401 2

3 with a large single-month deviation, a plan sponsor may need to ask whether the disparity is due to a poor fit of the hedging assets to the liability, or to some manager-specific issue, or to an idiosyncratic liability calculation in the face of a particular event. For example, in Nov. 2011, the downgrade of a single financial issuer that represented more than 6% of the bond universe in the Citigroup Pension Discount Curve had a significant effect on the curve. In June 2012, several other financial issuers were downgraded as well. Both of these events contributed to material price dispersions between assets and liabilities (in our case study above, the estimated asset underperformance in those months was 360 basis points and 280 basis points, respectively). For a plan pursuing a middle-stage or late-stage LDI strategy, dispersions of this magnitude could move the needle on funded status, yet be due primarily to the structural dissimilarity between liabilities and asset portfolios. However, if a month of similar underperformance occurs without a liability-specific event, then perhaps it is time to evaluate the effectiveness of the hedging strategy or some other manager-specific issue. (See our Leadership Series article Balancing Risk and Return for Custom LDI Strategies, Oct. 2014, for a fuller description of this process.) Why downgrades can produce performance dispersions Many LDI practitioners may already be familiar with the reasons downgrades (and more rarely, defaults) can have a stronger detrimental impact on an asset portfolio than on a liability discount curve. Simply put: If a bond or issuer is downgraded below the minimum credit rating for the curve, it simply drops out of the discount rate calculation, with no negative consequence to the liability s return. Unfortunately, the consequences for an asset portfolio are more direct: A credit downgrade will likely reduce the market value of the bond, creating the potential for short-term underperformance of an asset portfolio holding that bond. Perversely, even an asset manager who avoids owning the downgraded bond could feel its effect. Because the pre-downgrade bond may have been EXHIBIT 2: Performance differences between an asset and a liability can be volatile over shorter periods, including the annual cycle used for reporting a pension s funded status. Percentage Points Difference 8% 6% 4% 2% 0% -2% -4% -6% -8% -10% -12% 12-MONTH PERFORMANCE DIFFERENCE: ASSETS MINUS LIABILITIES Aug-05 Feb-06 Aug-06 Feb-07 Aug-07 Feb-08 Aug-08 Feb-09 Aug-09 Feb-10 Aug-10 Feb-11 Aug-11 Feb-12 Aug-12 Feb-13 Aug-13 Feb-14 Aug-14 Assets represented by Barclays Long Corporate AA or Higher Index; liabilities represented by a hypothetical plan liability discounted by the Citigroup Pension Discount Curve. Past performance is no guarantee of future results. All indices are unmanaged. It is not possible to invest directly in an index. Data from Aug. 1, 2004 to Aug. 31, Source: Barclays POINT, Fidelity Investments. trading at a higher yield due to the wider credit spread, its removal could actually increase the liability value if the average yield of the liability curve universe drops when the bond is removed. Liability curve concentration risk is also a factor The high issuer concentration in most liability yield curves may amplify mismatches between liability and asset returns. High concentrations in a liability curve may make it extremely volatile, EXHIBIT 3: Generally, indices used to proxy pension liabilities may hold higher issuer concentrations than broader investable indices. Issuer % of BOA ML 15+ AAA-AA General Electric 18.7 Wal-Mart 13.4 Pfizer 6.0 Florida Power & Light 4.7 Shell International 4.5 Berkshire Hathaway 4.4 Johnson & Johnson 3.6 Microsoft 3.3 Merck 3.0 Roche Holdings 2.8 Top ten total 64.3 Issuer % of Barclays Long Corp. A+ AT&T 4.4 General Electric 3.8 Comcast 3.5 Wal-Mart 3.4 HSBC 2.4 Wells Fargo 2.4 JP Morgan 2.1 Duke Energy 2.0 ConocoPhillips 1.9 Pfizer 1.6 Top ten total 27.4 Issuer % of Barclays Long Corp. Index Verizon 4.0 AT&T 2.3 General Electric 2.0 Comcast 1.9 Wal-Mart 1.8 Goldman Sachs 1.7 Citigroup 1.5 Wells Fargo 1.4 MidAmerican Energy 1.3 HSBC 1.3 Top ten total 19.1 Numbers may not sum due to rounding. All indices are unmanaged. It is not possible to invest directly in an index. Index data as of Aug. 31, Source: Barclays POINT, Fidelity Investments. 3

4 and in practice, an LDI asset portfolio with similar concentrations may be undesirable. For example, the BOA ML 15+ AAA-AA curve (a commonly used liability discount curve) currently has more than 64% concentration in the top 10 securities, with nearly 19% in the single top issuer (see Exhibit 3, page 3). In comparison, the Barclays Long Corporate A or Better Index (a liability-aware bond universe), has less than 30% in the top 10 issuers and not more than 5% in any single issuer. In the full long-credit universe that an LDI portfolio manager might consider, the diversification is even greater. Thus, if an asset manager attempts to construct an LDI portfolio relative to a pure liability-based index, the concentration risk may be high. A more prudent approach such as using a broader, strategically defined universe of securities to appropriately hedge the liability curve s credit risks may reduce concentration risk at the cost of increased TE, as even minor credit events experienced by issuers that are disproportionally represented in the liability curve may lead to meaningful return differences. Bond supply (and demand) varies along the curve The availability of tradable investment securities in the liabilitybased universe can also be a challenge. Liability discount curve providers use various techniques to smooth yields at maturity nodes for which there is not a deep liquid market in corporate securities. These techniques often amount to using some process for interpolating the spread over the Treasury curve from adjacent nodes. As a result, the yield ascribed to any given node within a liability yield curve may not be attainable in an asset portfolio. Exhibit 4 (right) demonstrates the mismatch between the liability yield curve and the available bonds at that node in the Barclays Corporate A or Better Index. At the 17-year node, for example, the liability curve uses a smoothed option-adjusted spread over Treasuries to come up with a yield of 3.73%. The apparent good news for asset managers is that many of the bonds at that node (which are 30-year bonds issued 13 years ago) seem to have higher yields than the liability. The bad news is that these bonds, and others like them, are generally unavailable in the secondary market they have already been bought and tucked away by other pension funds and insurance companies. Whenever securities at that node even come to the market, the low liquidity and specific demand often leads to a wide bid-ask spread, resulting in high transaction costs relative to the on-the-run securities used as a basis for the yield curve, and thus a lower yield for the bond in practice. Note that something different happens at the 30-year node. There, because bonds are closer to issue date, asset managers may find greater liquidity, lower bid-ask spreads, and higher supply. However, with greater liquidity and availability, these bonds are assumed to trade with tighter spreads to Treasuries, which is reflected in the yield curve s calculation. As a result, the liability yield curve at this node is such that the liability will very likely outperform a diversified asset portfolio made up of available bonds, EXHIBIT 4: Yields listed at various nodes in liability yield curves are often different from yields available for actual bonds in the market. 17-Year Node Yield Rating Discount Curve 17-Year Yield 3.73% CONOCO FNDNG GLB7.2510/15/31DT 3.78% A+ BURLINGTON RSC 7.2% 8/15/ % A+ BURLINGTON RSC 7.4% 12/01/ % A+ PACIFICORP 7.7% 11/15/ % A+ AT&T CORP 8/8.5% 11/15/ % A BELLSOUTH GLBL 6.875% 10/15/ % A 30-Year Node Yield Rating WI ELEC PWR 4.25% 06/01/ % A+ 3M CO 3.88% 06/15/ % AA NORTHWESTRN UNI 4.643% 12/1/ % AAA PUBLIC SERVICE ELEC 4% 6/1/ % AA- PPL ELEC UTILS CORP 4.12% 6/ % A MIDAMERICAN ENRGY /15/ % AA WALT DISNEY CO 4.125% 06/01/ % A WISCONSIN PUB SER 4.752% 11/ % AA DTE ELEC CO 4.3% 07/01/ % AA- Discount Curve 30-Year Yield 4.08% LILLY (ELI) & CO 4.65% 6/15/ % AA- AMEREN IL CO 4.3% 07/01/ % A UPAC 4.85% 06/15/ % A ORACLE CORP 4.5% 7/8/ % A+ GEORGE WASH UNI 4.3% 09/15/ % A+ MONSANTO CO NEW 4.4% 07/15/ % A- OGLETHORPE PW 4.55% 06/01/ % A AT&T INC 4.8% 06/15/ % A GOLDMAN SACHS 4.8% 07/08/ % A Rating: Assigned by Fidelity based on S&P, Moody s, and Fitch ratings. Discount Curve: Citigroup Pension Discount Curve. Yields shown of bonds available on Aug. 31, 2014 that are within the Barclays A or Better Index. Source: Barclays POINT, Bloomberg Finance L.P., Fidelity Investments. as many of the readily available bonds in that node have yields lower than that of the discount curve. Managers with experience in constructing an LDI portfolio will strive to balance these market dynamics, knowing when to accept some degree of mismatch to the yield curve s assumptions in order to optimize cost, availability, and return potential. The tradeoff, however, is again likely to be higher tracking error. A curve s trimming criteria can increase risk Discount curves may have additional selection criteria that make them even more un-investable. Some plans calculate their liabilities using custom discount rates, in which bond yields from certain tails of the distribution are eliminated from consideration. For example, a 10/90 approach would drop outliers from the yield calcula- 4

5 tion by excluding bonds in the highest and lowest deciles of yields in the universe. Another popular choice is the use of an asymmetric exclusion process such as the 60/90 approach, in which the lowest-yielding 60% of bonds are excluded while only the highest-yielding 10% are dropped. These approaches often lead to higher discount rates (bringing down the value of the liability) but, needless to say, can create a challenge for an LDI asset portfolio in matching liability returns. Indeed, we have found that keeping pace with a liability valued in this manner typically requires a large allocation to BBB-rated securities or other issues that can outyield an AA-rated universe, compounding the portfolio s TE to the liability. Moreover, this customized universe of bonds can shift quite substantially from one period to the next, and the movement of a particular bond into the cutoff tail may not be as transparent as a ratings downgrade, because it depends on the yield of the individual security relative to the rest of the bonds included in the curve. An asset manager attempting to invest relative to the securities in a custom liability discount curve may have a very difficult time even tracking the constituents from month to month, much less matching them in a bond portfolio. Despite these complications, the decision to use a custom liability curve may be entirely appropriate for a plan sponsor implementing an LDI strategy, particularly in the later stages. However, we would recommend making sure members of the plan committee and other relevant stakeholders are aware of the additional TE that may be introduced, which may remain in the plan even after moving to a high LDI allocation. Pricing basis gaps are likely larger than expected In case the preceding factors are not enough, we also want to highlight the possibility that vendor pricing basis gaps may also EXHIBIT 5: Yields listed at various nodes in liability yield curves are often different from yields available for bonds in the market. Annualized Tracking Error between Bond Indices Long-Duration Corporate (BofA ML vs. Barclays) Intermediate-Duration Corporate (BofA ML vs. Barclays) 3-Year 5-Year 10-Year 1.13% 1.06% 1.75% 0.45% 0.46% 0.72% Past performance is no guarantee of future results. Long-Duration Corporate: BOA ML 15+ Year AAA-AA U.S. Corporate Index and Barclays U.S. Corporate 15+ Year AA or Better Index. Intermediate-Duration Corporate: BOA ML U.S. Corporate Index and Barclays U.S. Corporate Index. All indices are unmanaged. It is not possible to invest directly in an index. Index data as of Aug. 31, Source: Barclays POINT, Fidelity Investments. add to tracking error. Two similarly-constructed bond indices may have meaningful performance dispersions due primarily to vendor pricing differences (perhaps in conjunction with modest inclusion rule differences, such as minimum issue size). Exhibit 5 (below) shows the annualized TE between two long-duration asset indices with similar durations (currently 14 years). TE between these two very similar indices has been higher than one might expect ranging from 1% to 1.75% over the past three, five and ten years. Note also that the TE between similar intermediateduration credit bond indices has been meaningful as well, spanning from 45 to 72 basis points. The implication is that the TE present between the intermediate-duration indices is most likely at least partly attributable to pricing issues, and is amplified by the longer-duration nature of the other indices. If two well-matched intermediate-duration indices can produce TE in excess of 45 basis points and two similar long-duration bond benchmarks can produce a multiple of that it may be unreasonable to expect an LDI asset portfolio to generate tracking error anywhere close to those levels when measured against a liability. Implications for plan sponsors We believe that when comparing the returns of an asset portfolio to those of the liability, plan sponsors may benefit from recognizing several ingrained sources of TE. As mentioned above, we have found that in the early stages of an LDI implementation, TE may be generated primarily by the risk inherent in a return-seeking allocation of the pension assets, and by a bond portfolio that differs in duration from the liability. Accordingly, many plans may see their TE decline from 15% to 17% down to mid-single digits as they pursue the early stages of an LDI strategy and reduce the return-seeking allocation. The cost of success for an LDI strategy, however, is that as the plan moves toward the later stages of fully hedging the liability, the TE that remains becomes more noticeable. Indeed, there may in fact be a practical floor to the level of TE, even for successful LDI implementations. TE may arise because credit events, concentration risk, supply and demand, and selection criteria are all handled differently by liability discount curves than by asset portfolios, and because pricing source differences may introduce routine levels of TE even between similar asset indices. In aggregate, these discrepancies can increase the amount of volatility one should expect when comparing a liability discount curve to an investable index, adding up to a reasonable TE expectation of 3% to 4%. Understanding the sources of this remaining TE may allow plan sponsors to more confidently evaluate the performance of their assets versus their liability, and may help in setting appropriate expectations for a successful risk management program. 5

6 Authors Dan Tremblay, CFA Director of Institutional Fixed Income Solutions & LDI Strategist Daniel Tremblay is senior vice president, director at Fidelity Institutional Asset Management SM (FIAM SM ), an investment organization within Fidelity Investments asset management division that is dedicated to serving the needs of consultants and institutional investors, such as defined benefit and defined contribution plans, endowments and financial advisors. In this role, Mr. Tremblay oversees the Liability Driven Investment (LDI) Solutions team and is responsible for developing custom hedging strategies for LDI clients, providing perspectives on de-risking solutions, and representing the investment process in the marketplace. François Pellerin, CFA, FSA, EA, CERA LDI Strategist Ben Tarlow, PhD Portfolio Manager Dr. Ben Tarlow is a portfolio manager at Fidelity Institutional Asset Management SM (FIAM SM ), an investment organization within Fidelity Investments asset management division that is dedicated to serving the needs of consultants and institutional investors, such as defined benefit and defined contribution plans, endowments and financial advisors. In this role, he is responsible for long duration, long credit and custom Liability Driven Investing (LDI) portfolios. Dr. Tarlow has been a portfolio manager since He is also a member of the Liability Driven Investing Solutions team, where he is responsible for portfolio quantitative risk and return modeling, and drives the team s pension asset/liability risk modeling and drives. François Pellerin is an LDI strategist in the Fixed Income division at Fidelity Institutional Asset Management SM (FIAM SM ), an investment organization within Fidel ity Investments asset management division that is dedicated to serving the needs of consultants and institutional investors, such as defined benefit and defined contribution plans, endowments and financial advisors. In this role, he is a member of the Liability Driven Investment (LDI) Solutions team and is responsible for developing and implementing pension risk management solutions. Investment Director Michael Jarasitis contributed to this article. Fidelity Thought Leadership Vice President and Senior Investment Writer Vic Tulli provided editorial direction for this article. 6

7 Information presented herein is for discussion and illustrative purposes only and is not a recommendation or an offer or solicitation to buy or sell any securities. Views expressed are as of the date indicated, based on the informa tion available at that time, and may change based on market and other conditions. Unless otherwise noted, the opinions provided are those of the authors and not necessarily those of Fidelity Investments or its affiliates. Fidelity does not assume any duty to update any of the information. Past performance is no guarantee of future results. Unless otherwise disclosed to you, in providing this information, Fidelity is not undertaking to provide impartial investment advice, or to give advice in a fiduciary capacity, in connection with any investment or transaction described herein. Fiduciaries are solely responsible for exercising independent judgment in evaluating any transaction(s) and are assumed to be capable of evaluating investment risks independently, both in general and with regard to particular transactions and investment strategies. Fidelity has a financial interest in any transaction(s) that fiduciaries, and if applicable, their clients, may enter into involving Fidelity s products or services. Neither asset allocation nor diversification ensures a profit or guarantees against a loss. In general, the bond market is volatile, and fixed income securities carry interest rate risk. (As interest rates rise, bond prices usually fall, and vice versa. This effect is usually more pronounced for longer-term securities). Fixed income securities also carry inflation risk, liquidity risk, call risk and credit and default risks for both issuers and counterparties. Lower-quality fixed income securities involve greater risk of default or price changes due to potential changes in the credit quality of the issuer. Foreign investments involve greater risks than U.S. investments, and can decline significantly in response to adverse issuer, political, regulatory, market, and economic risks. Any fixed-income security sold or redeemed prior to maturity may be subject to loss. Stock markets are volatile and can fluctuate significantly in response to company, industry, political, regulatory, market, or economic developments. Investing in stock involves risks, including the loss of principal. Specific securities mentioned are for illustrative purposes only and must not be considered an investment recommendation or advice. Endnotes 1 Shifting some exposure to low-volatility equity strategies may also reduce TE. See the Leadership Series article Low-Volatility Equity Investing for U.S. Corporate Defined Benefit Plan Sponsors, Aug As shorthand, we often refer to liability returns to describe changes in the value of the liability due to shifts in the discount rate yield curve, even though the liability itself does not strictly generate a return. This language fits with the use of TE to measure asset return volatility relative to the value of the liability (as opposed to measuring it relative to the returns of a benchmark index). 3 Short-term asset outperformance could have a positive effect on reporting, but the potential benefits to a company of an idiosyncratically inflated funded status may not be equal to the costs of a similarly likely drop. Our overall context here includes the assumption that lower risk is desirable. Index definitions Bank of America Merrill Lynch (BofA ML) U.S. Corporate Index tracks the performance of U.S. dollar-denominated investment grade corporate debt publicly issued in the U.S. domestic market. Qualifying securities must have an investment grade rating (based on an average of Moody s, S&P and Fitch), at least 18 months to final maturity at the time of issuance, at least one year remaining term to final maturity as of the rebalancing date, a fixed coupon schedule and a minimum amount outstanding of $250 million. BofA ML 15+ Year AAA-AA U.S. Corporate Index is a subset of the BofA Merrill Lynch U.S. Corporate Index including all securities with a remaining term to final maturity greater than or equal to 15 years and rated AAA through AA3, inclusive. Barclays Credit Index consists of publicly issued U.S. corporate and specified foreign debentures and secured notes that meet the specified maturity, liquidity, and quality requirements. To qualify, bonds must be SEC-registered, dollar-denominated, have at least $250 million in par outstanding and be rated investment grade by at least two of the following three ratings agencies: Moody s, S&P, Fitch. Barclays U.S. Corporate Index is a broad-based benchmark that measures the investment grade, U.S. dollar-denominated, fixed-rate, taxable corporate bond market. It includes USD-denominated securities publicly issued by U.S. and non-u.s. industrial, utility, and financial issuers that meet specified maturity, liquidity, and quality requirements. Barclays Long Corporate Index includes all corporate bonds in the Credit Index with at least 10 years until maturity. Barclays Long Corporate A or Better Index includes all corporate bonds in the Credit Index rated above Baa1/BBB+ with at least 10 years until maturity. Barclays Corporate 7-10 Index includes all corporate bonds in the Credit Index with 7 to 10 years until maturity. Barclays Long Corporate BBB Index includes all corporate bonds in the Credit Index rated Baa/BBB with at least 10 years until maturity. Barclays Long Credit Non-Corporate Index includes all bonds in the Credit Index that are not corporate bonds with at least 10 years until maturity. Barclays U.S. Corporate 15+ Year AA or Better Index is a customized subset of the Barclays U.S. Corporate Bond Index that constrains the initial universe to issues having a maturity of 15 years or longer and a quality rating of AA or better. Citigroup Pension Discount Curve (CPDC) is a set of yields on hypothetical AA-rated zero-coupon bonds with maturities from six months to 30 years, calculated based on a universe of AA-rated corporate bonds from the Citigroup U.S. Broad Investment-Grade Bond Index and the yields of Citigroup s Treasury model curve. The yields of the CPDC are often used to discount pension liabilities. Third-party marks are the property of their respective owners; all other marks are the property of FMR LLC. This publication may be provided by Fidelity Institutional Asset Management Trust Company or FIAM LLC, depending on your relationship. 7

8 FMR LLC. All rights reserved.