For better pension liability matching, consider adding Treasuries

For better pension liability matching, consider adding Treasuries Vanguard research December 2012 Executive summary. When pension plan sponsors think about reducing risk, their first inclination is usually to add long-term corporate bonds to their portfolio, since a plan s liability is discounted using high-quality corporate bond rates. Long corporate bonds match the liability well, but not perfectly; among other factors, they can suffer defaults and downgrades that do not affect the value of the pension promise. U.S. Treasury bonds have lower yields than the pension liability, but because their default risk is minimal, adding them may mitigate the downgrade risk in corporate bonds. Authors Paul M. Bosse, CFA R. Evan Inglis, FSA, CFA Can a combination of Treasury securities and long corporate bonds in an all-bond portfolio reduce pension risk better than long corporate bonds alone? Vanguard s research suggests it can. This is important, as hedging the pension liability is steadily displacing return-seeking as pension plans primary objective. Connect with Vanguard > vanguard.com

Over time, a portfolio of corporate bonds held by a pension plan faces a headwind relative to the pension liability as a result of credit-rating downgrades and defaults within the portfolio. Particular bonds can be downgraded out of the universe of bonds used to determine the discount rate for the liability. Although such downgrades represent losses in an asset portfolio, the measured liability value actually increases when higheryielding bonds are eliminated from those used for liability measurement. That means even a precisely matched portfolio of corporate bonds such as the hypothetical portfolios put together to justify discount rates for accounting purposes will not ultimately provide a return equal to the yield used for measuring liability and will fall behind the liability value. Vanguard research suggests that combining longterm U.S. Treasury and corporate bonds can reduce pension risk more than using long corporates only. Such a strategy can be important, given the current trend toward hedging a pension s liability versus seeking a higher return. Downgrade drag in corporate bonds The life cycle of the corporate bond headwind is shown in Figure 1. Note that the actual loss to the pension plan s funded status doesn t happen until the downgraded bond drops out of the liability universe. The real event isn t the decline in the asset values; rather, it is the upward revaluation of the liability. The far-right panel in Figure 1 shows this: When a rating agency drops the bond from the universe (for example, AA bonds), the liability reverts to its original value, leaving the asset portfolio behind. This downgrade headwind tends to depress returns in times of economic crisis (as in downturns in 2000 2002 and 2008, reflected in Figure 2). Adding Treasuries can mitigate this downgrade effect and improve the tracking of the pension liability. Our study suggests that, over time, and owing to credit downgrades and defaults, a typical corporate bond portfolio suffers an annual return hit of 60 80 basis points relative to the liability. (This hit tends to occur not regularly, but episodically during times of economic crisis, as Figure 2 shows). Although Treasury bonds carry lower yields than corporate bonds of comparable maturity, that difference is less dramatic once this headwind is accounted for. Notes on risk: All investing is subject to risk, including possible loss of principal. Past performance does not guarantee future results. When interest rates rise, the price of a bond or bond fund will decline. s are subject to credit risk and inflation risk. Credit risk is the risk that a bond issuer will fail to make timely payments of interest and principal. Inflation risk is the possibility that increases in the cost of living will decrease or eliminate the returns of an investment. Because high-yield bonds are considered speculative, investors should be prepared to assume a substantially greater level of credit risk than with other types of bonds. 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. The performance of an index is not an exact representation of any particular investment, as you cannot invest directly in an index. U.S. government backing of Treasury or agency securities applies only to the underlying securities and does not prevent share-price fluctuations. Unlike stocks and bonds, U.S. Treasury bills are guaranteed as to the timely payment of principal and interest. Although the income from the U.S. Treasury obligations held in a fund is subject to federal income tax, some or all of that income may be exempt from state and local taxes. In a diversified portfolio, gains from some investments may help offset losses from others. However, diversification does not ensure a profit or protect against a loss in a declining market. 2

Figure 1. Impact of corporate bond headwind on liability Two-bond liability universe Market devalues B Rating agency downgrades B; B drops out of liability universe Yields 6.0% B B 5.0 A B A Yield increases A Discount rate = 5.0% Discount rate = 5.5% Discount rate = 5.0% Note: Using an approach similar to that proposed in Ransenberg and Hobbs (2011), we estimated the impact of corporate bond downgrades by taking the difference in yields between the statistics universe and the returns universe of the Barclays U.S. Long Credit A or Better Index. Figure 2. Impact of corporate bond downgrades on pensions funded status 105% 100 95 90 85 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Funded status Notes: The Barclays Long Credit A or Better Index was used to represent both the typical pension liability and its underlying assets. To calculate funded status each month, this index was used to represent the return for both the pensions assets and their liabilities, but with a different return calculation. Each month, the assets were credited with the return from the index s returns universe. The same was true for the liability, but the liability also grew with an extra return component because of the difference in yield between the statistics universe and the returns universe, multiplied by the duration of the index. The extra return component caused the downgrade headwind. That downgrade impact was captured by comparing the yield from the index s returns universe, whose underlying basket of bonds remains static each month, and the yield from the statistics universe, whose underlying basket adjusts as bonds are upgraded and downgraded. This approach was referenced from Ransenberg and Hobbs (2011). 3

Arguments for Treasury bond exposure Since Treasury yields still remain below those of corporate bonds even after accounting for the downgrade headwind better tracking may not be reason enough to add Treasuries to the mix. 1 Other reasons, though, argue for including them: Some investors may not consider a downgrade to be a problem if the intent is to hold bonds to maturity. But many pension plan sponsors, particularly of frozen or closed plans, will not wait for long bonds to mature before terminating their plans, in which case the lower price of a downgraded bond will affect the cost of termination. In addition, the downgrade may be just the first step toward default. Long investment-grade corporate bonds are sparse enough that full duration coverage along the yield curve is hard to achieve. In particular, pension obligations in the range of 15 to 20 years are difficult to cover fully with corporate bonds. Also, pension plans typically have payments at 30 years and beyond, and long Treasury STRIPS can be an excellent vehicle for getting the longest duration exposure. Better liquidity in the Treasury market makes it less expensive to adjust a portfolio to keep the asset-liability match close as the liability changes. Lack of liquidity is less of an issue with bond funds than with individual securities, but it is a consideration that will grow increasingly important as new regulations make trading corporate bonds more challenging. 2 Treasury exposure is especially important as a plan looks to terminate in the near term. At that point, the trade-off in yield with corporate bonds is not as significant as tracking the liability closely. In normal corporate spread environments, the discount rate used for an annuity purchase will be similar to Treasury yields, even if the rate moves more in tandem with corporate bonds. Key terms R-squared (R 2 ). A measure of how much of a fund s past returns can be explained by the returns from a given index. STRIPS. An acronym for Separate Trading of Registered Interest and Principal of Securities. STRIPS are created when a broker-dealer or the U.S. Treasury separates ( strips ) the interest and principal of a Treasury note or bond into separate components, which are then traded as zero-coupon securities. Investors buy them at a price below the face value of the securities and receive the full amount when the STRIPS mature. In portfolios with significant equity exposure, Treasuries are a better diversifier because they don t add to the corporate risk in equities. Allocations to long STRIPS are effective (Bosse and Vincent, 2010). How much to invest in Treasuries? Putting portfolios to the test Adding Treasury bonds will not address the return loss from the downgrade/default headwind, but it may lower the pension liability risk for a 100% bond portfolio. To check this, we tested various portfolios relative to a pension liability (see Figure 3). The liability is represented by the Citigroup Pension Liability Index (CPLI) series. Using monthly data from January 1994 (the inception of the CPLI) through March 2012, we first looked at how well the Barclays U.S. Long Corporate Index tracked the liability. The first column of Figure 3 shows the regression s R-squared and the calculated correlation between the Barclays index and the CPLI. Moving across the table, we added the Barclays U.S. Long Treasury Index allocation and again calculated the R 2 and the correlation of the resulting portfolios to the CPLI. 1 The wider corporate bond spread may be due to the tax advantage afforded to Treasuries (no state income tax is levied on Treasury interest payments), compensation for expected default, an illiquidity premium, or a general risk/uncertainty premium. 2 The Volcker rule, which was part of the Dodd-Frank Wall Street Reform and Consumer Protection Act, signed into law by President Barack Obama in 2010, may make it harder to trade corporate bonds because it limits banks ability to hold corporate bond inventory to facilitate trading. 4

Figure 3. Effect of allocation on liability tracking: January 1994 March 2012 Shaded area highlights the strongest correlations between a bond portfolio s Treasury exposure and its fit to the CPLI. Percentage in long Treasuries 0% 10% 20% 25% 30% 35% 40% 50% Regression fit: R-squared 0.829 0.857 0.876 0.880 0.881 0.878 0.872 0.849 Correlation to CPLI 0.910 0.920 0.936 0.938 0.939 0.937 0.934 0.922 Notes: Pension liability is represented by the Citigroup Pension Liability Index (CPLI) series; long corporates are represented by the Barclays U.S. Long Corporate Index; long Treasuries are represented by the Barclays U.S. Long Treasury Index. See text for further explanation of this analysis. Figure 4. Comparing correlations of CPLI to a long corporate bond portfolio and to a 25% Treasury mix: December 29, 1995, through March 30, 2012 1.2 1.0 0.8 0.6 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 CPLI versus long corporates CPLI versus 25% Treasury mix Note: Figure compares correlation of CPLI to Barclays U.S. Long Corporate Index and to Barclays U.S. Long Treasury Index; rolling 12-month periods. Indeed, adding long Treasury bonds improved the fit to the CPLI. Tracking peaked at a 30% portfolio allocation to Treasuries (calculations are shown in Figures A-1 and A-2 in the Appendix, on page 6). Given that the test period was relatively short, we drew no conclusion about a precise optimum amount for a long Treasury bond allocation. A Treasury allocation of 20% to 35% provided a benefit over a portfolio with no Treasury bonds. As shown in Figure 4, a 12-month rolling correlation over 16 years shows that a 25% allocation to long Treasury bonds improved the liability match, primarily at key distress points. (A shorter, more typical liability was tested to confirm these findings; see Figure A-3 in the Appendix, on page 7.) During a market crisis, when the downgrade headwind is strongest, Treasuries often command a flight-toquality premium a valuable diversification effect. 5

Bottom line If reducing pension risk is your primary objective, then a 100% corporate bond allocation is an excellent way to accomplish it. For a better fit, and for the other benefits that Treasuries can provide, allocating 20% to 35% of the fixed income portfolio to Treasury bonds can make your portfolio even more effective. References Bosse, Paul M., and Francis Vincent, 2010. Pension Risk Control: Is There a Better Way? Valley Forge, Pa.: The Vanguard Group. Ransenberg, Daniel, and Jonathan Hobbs, 2011. Overcoming Credit Downgrades: Four Ways to Improve Your Liability Hedge. New York: BlackRock. Appendix. Linear regression results and alternate liability test Linear regression results for explaining CPLI with long-term corporate bonds and with a blend of 70% long corporate bonds/30% long Treasuries Figure A-1. Linear regression fit for CPLI with Barclays U.S. Long Corporate Index Figure A-2. Linear regression fit for CPLI with 70% Barclays U.S. Long Corporate Index/30% Barclays U.S. Long Treasury Index 20% 15 15% 10 5 0 5 10 y = 0.6222x + 0.0015 R 2 = 0.8285 15 20% 15 10 5 0 5 10 15 20 25% 10 5 0 5 10 20% 10 0 10 20 30% y = 0.6075x + 0.0017 R 2 = 0.8808 6

Figure A-3. Comparing correlations of a typical client liability to a long corporate bond portfolio and to a 25% Treasury mix: December 29, 1995, through March 30, 2012 1.2 1.0 0.8 0.6 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Client liability versus long corporates Client liability versus 25% Treasury mix Note: Figure compares correlation of typical client liability to Barclays U.S. Long Corporate Index and to Barclays U.S. Long Treasury Index; rolling 12-month periods. An alternate liability test The CPLI, as of October 31, 2012, had a duration of 19.4 years, whereas most pension liabilities are in the range of 11 to 15 years. We wondered whether this difference might mean that a typical pension plan would not lower risk by adding Treasury bonds to its portfolio. To check, we ran the same test as we did for Figures 3 and 4, except we used a typical pension liability for an actual Vanguard client. The result: Adding Treasury bonds actually improved the liability match a bit more than it did for the CPLI. The R 2 and correlations both rose by similar small amounts for all the tested allocations. Figure A-3 shows the resulting 12-month rolling correlation. 7

P.O. Box 2600 Valley Forge, PA 19482-2600 Connect with Vanguard > vanguard.com Vanguard research > Vanguard Center for Retirement Research Vanguard Investment Counseling & Research Vanguard Investment Strategy Group E-mail > research@vanguard.com For more information about Vanguard funds, visit vanguard.com or call 800-662-2739 to obtain a prospectus. Investment objectives, risks, charges, expenses, and other important information about a fund are contained in the prospectus; read and consider it carefully before investing. CFA is a trademark owned by CFA Institute. 2012 The Vanguard Group, Inc. All rights reserved. Vanguard Marketing Corporation, Distributor. ICRBPL 122012