Pension risk: How much are you really taking?

Pension risk: How much are you really taking? Vanguard research June 2013 Executive summary. In May 2012, Vanguard conducted the second of a planned series of surveys of corporate defined benefit (DB) plan sponsors. 1 One question new to the 2012 survey was, What downside variation on the plan funding ratio is acceptable to you? The majority of plan sponsors (90%) responded with a figure between 0% and 10%. Sponsors also listed uncertainty in the plan funding ratio as a top concern about pension risk. Yet, average asset allocations in the DB plans we analyzed were roughly 60% equity/alternatives and 40% intermediate-term bonds, indicating a likely large mismatch between assets and liabilities and, consequently, exposure to large declines in funding ratios. To confirm this disconnect, we analyzed historical and projected annual funding-status volatility for a typical plan and found that both historical and projected funding-status volatility were well beyond the 10% variability ceiling with which sponsors reported being comfortable. Author Kimberly A. Stockton 1 For a full analysis of the survey results, see the Vanguard research paper titled Survey of Defined Benefit Plan Sponsors, 2012 (Stockton, 2012). Connect with Vanguard > vanguard.com

A key finding of Vanguard s Survey of Defined Benefit Plan Sponsors, 2012 (the second in a series of corporate-sponsor surveys, reported in Stockton, 2012) related to an apparent disconnect between sponsors risk expectations and their actual portfolio asset allocations. Specifically, our 2012 survey asked sponsors a new question: What downside variation on the plan funding ratio is acceptable to you? Most sponsors (90%) responded with a figure between 0% and 10%, while also listing plan funding ratio as a major risk concern. Nevertheless, average asset allocations in the plans we analyzed were roughly 60% equity/ alternatives and 40% intermediate-term bonds, which suggests a disconnect between assets and liabilities and potential exposure to large declines in funding ratios. This brief paper highlights the results of our recent deeper analysis into survey respondents reported asset allocations and risk tolerance. The first step in attempting to confirm the perceived discrepancy between sponsors reported tolerance for risk and the typical risk profile of defined benefit plans was to review historical funding ratios for a typical plan from January 1998 through January 2012. We compared results for a hypothetical plan with a 50% equity/40% aggregate bond/10% alternative allocation (Portfolio 1) with those for a plan with an LDI (liability-driven investing) allocation of 80% long credit bonds/20% long-duration U.S. Treasury STRIPS (Portfolio 2). As Figure 1 shows, in Portfolio 1, annual downside variation of 10% or more in the funding ratio was a relatively common event during the 13-year analysis period. In contrast, for Portfolio 2, which was more closely matched to liabilities, annual variation in funding ratio would have remained within 10% over the entire period. Note that there is still some volatility even with the LDI portfolio, because it is not perfectly matched to the liability; a perfect matching to liability is not possible for a pension liability measured with corporate bond discount rates. Also, please see the box on page 3, Four key considerations in this analysis. Important: The projections or other information generated by the Vanguard Capital Markets Model regarding the likelihood of various investment outcomes are hypothetical in nature, do not reflect actual investment results, and are not guarantees of future results. VCMM results will vary with each use and over time. The VCMM projections are based on a statistical analysis of historical data. Future returns may behave differently from the historical patterns captured in the VCMM. More important, the VCMM may be underestimating extreme negative scenarios unobserved in the historical period on which the model estimation is based. Notes on risk: All investing is subject to risk, including the possible loss of the money you invest. Past performance is no guarantee of future returns. 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. Investments in stocks or bonds issued by non-u.s. companies are subject to risks including country/regional risk and currency risk. Diversification does not ensure a profit or protect against a loss. Stocks of companies based in emerging markets are subject to national and regional political and economic risks and to the risk of currency fluctuations. These risks are especially high in emerging markets. Although the income from the U.S. Treasury obligations held in the fund is subject to federal income tax, some or all of that income may be exempt from state and local taxes. Funds that concentrate on a relatively narrow market sector face the risk of higher share-price volatility. 2

Figure 1. Historical funding ratios: Traditional versus LDI hypothetical portfolios 130% 120 110 Annual funding ratio 100 90 80 70 60 50 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Portfolio 1 (traditional) Portfolio 2 (LDI) Notes: The traditional portfolio is 36% U.S. equity/14% non-u.s. equity/40% U.S. aggregate bonds/10% commodities. The LDI portfolio is 80% U.S. long credit bonds and 20% 20 30 year Treasury STRIPS. Source: Vanguard. Four key considerations in this analysis Four aspects of this analysis deserve highlighting here: Downside funding ratio risk could be even higher in a different interest rate environment. In practice, Vanguard suggests a glide-path approach to derisking, which involves a dynamic approach to asset allocation that increases duration match and decreases equity risk over time. The LDI portfolio results discussed here are based on one static allocation to illustrate the difference in risk between the LDI portfolios and the typical DB portfolio. A measure called pension stabilization was recently passed by the U.S. Congress as part of the Moving Ahead for Progress in the 21st Century Act (MAP-21). MAP-21 introduces an extreme level of smoothing to the determination of contributions under the rules of the Pension Protection Act of 2006 smoothing that will artificially increase funded levels and decrease required contributions. Simulations in our analysis assumed a contribution level that would be more responsive to market interest rates than would be required under MAP-21. Clearly, some sponsors may view forecasted downside risk differently. Another approach, for instance, would be to compare actual funded status at the beginning of a year with 5th-percentile downside results at the end of that year. Evaluated this way, our findings still held. 3

Figure 2. Forward-looking distributions of assets and pension metrics for typical, hypothetical DB plans a. Funding ratio distribution forecast Portfolio 1 185% Annual funding ratio 165 145 125 105 85 95th percentile 50th percentile 5th percentile 45% U.S. equity 15% International equity 40% Aggregate bonds 0% Long credit bonds 0% Extended duration 65 Year 1 Year 2 Year 3 Year 4 Year 5 b. Funding ratio distribution forecast Portfolio 2 145% Annual funding ratio 135 125 115 105 95 95th percentile 50th percentile 5th percentile 25% U.S. equity 5% International equity 0% Aggregate bonds 60% Long credit bonds 10% Extended duration 85 Year 1 Year 2 Year 3 Year 4 Year 5 c. Funding ratio distribution forecast Portfolio 3 115% Annual funding ratio 110 105 100 95 90 95th percentile 50th percentile 5th percentile 0% U.S. equity 0% International equity 0% Aggregate bonds 80% Long credit bonds 20% Extended duration 85 Year 1 Year 2 Year 3 Year 4 Year 5 4

Figure 2 (continued). Forward-looking distributions of assets and pension metrics for typical, hypothetical DB plans d. Downside risk Portfolios 1 3 40% Annual downside risk in funding ratio 35 30 25 20 15 10 5 Portfolio 1 Portfolio 2 Portfolio 3 0 Year 1 Year 2 Year 3 Year 4 Year 5 Notes: Results presented in Figures 2a 2d are based on five-year forecasts from the VCMM. Distribution results are drawn from 10,000 VCMM simulations based on market data and other information available as of September 30, 2012. Because reliable forecasts for each alternative asset class were not available, this analysis assumed that the 10% allocation to alternative assets (which a typical DB plan holds) were in equities. Contributions were assumed to be zero in year 1. After year 1, contributions were based on a proxy for minimum contribution requirements, as stipulated in the seven-year amortization of funding shortfalls defined by the Pension Protection Act. See Appendix I for assumptions used in Figures 2a 2d. See Appendix II for further description of the VCMM. Source: Vanguard, from VCMM forecasts. Next, using our asset simulation model, the Vanguard Capital Markets Model (VCMM), we generated forward-looking distributions of assets and pension metrics for a typical DB plan. 2 We forecasted annual funding ratio distributions for a typical portfolio (Portfolio 1), a more liability-sensitive portfolio (Portfolio 2), and an all-bond portfolio with initial asset durations roughly matched to liabilities (Portfolio 3). We assumed a 100% starting funding ratio and contributions after year 1. The results are shown in Figures 2a 2d. Our projected results paralleled the historical results. The downside risk, as measured by 5th-percentile results, 3 was more than 10 percentage points from the expected result (50th percentile) in each year for the typical portfolio, as shown in Figure 2a. For example, at the end of year 1, funding ratios with this allocation were expected to be 100% but downside results were 21 percentage points lower. As shown in Figure 2b, even with the more liability-sensitive portfolio, downside risk was more than 10 percentage points in all years but the first. Only with the duration-matched portfolio (Figure 2c) was funding-level variation within the 10% level acceptable to the majority of plan sponsors who responded to our survey. For example, at the end of year 1, the expected funding ratio with the duration-matched portfolio was 100%, with downside results of only 4%. Figure 2d shows the difference between the 50thand 5th-percentile results for each of the portfolios. The results for Portfolio 1 indicated high risk and significant increases in the variability of results as the time horizon lengthened, while those for the duration-matched Portfolio 3 indicated relatively low downside risk over all five years. And even Portfolio 2, the more liability-sensitive portfolio, had downside results greater than 10% in all but year 1. 2 See Appendix I for analysis assumptions; see Appendix II for further description of the VCMM. 3 Note that 5% of outcomes will be worse than the 5th-percentile number. 5

Conclusion Based on the historical and forward-looking results reported here, it appears that asset allocations for DB plans may be out of line with plan sponsors reported risk tolerance. Although more than twothirds of respondents to Vanguard s 2012 Survey of Defined Benefit Plan Sponsors indicated they are using one or more LDI strategies, most still had large asset and liability mismatch. We believe that risk to the funded status could be brought within the tolerance level expressed in our survey findings with more aggressive LDI strategies. Reference Stockton, Kimberly A., 2012. Survey of Defined Benefit Plan Sponsors, 2012. Valley Forge, Pa.: The Vanguard Group. It should be noted that although the example in this analysis was for a fully funded DB plan, many plans today are actually underfunded. It may be appropriate for underfunded plans that follow a glide-path derisking approach to assume higher levels of risk and liability tracking error in exchange for higher potential returns and resulting deficit reduction. However, sponsors should be aware of the potential downside risk that they are assuming with this approach. Asset-liability modeling (ALM), especially when it incorporates glide-path derisking strategies, can help plan sponsors reconcile their risk tolerance with the asset allocation in their DB plan. 6

Appendix I. Analysis background and assumptions for funding ratio variation forecast These results are based on Vanguard s Capital Markets Model, a vector autoregression model that uses Monte Carlo simulation techniques. Plan information Starting assets Liability duration $2.9M 15.5 years Normal cost $0 Plan status Frozen Initial funding ratio 100% Time horizon for study 5 years Asset classes U. S. extended duration bonds (long Treasury STRIPS) U.S. long-term credit bonds U.S. aggregate bonds Non-U.S. equity U.S. equity Note: We assumed a 100% starting funding ratio to lessen the impact of assumed future contributions and highlight the impact of asset allocation on funding-ratio volatility. Appendix II The Vanguard Capital Markets Model (VCMM) is a proprietary financial simulation tool developed and maintained by Vanguard s Investment Strategy Group. The VCMM uses a statistical analysis of historical data for interest rates, inflation, and other risk factors for global equities, fixed income, and commodity markets to generate forward-looking distributions of expected long-term returns. The asset, return, and metric distributions shown in this paper are drawn from 10,000 VCMM simulations based on market data and other information available as of September 30, 2012. The VCMM is grounded in the empirical view that the returns of various asset classes reflect the compensation investors receive for bearing different types of systematic risk (or beta). Using a long span of historical monthly data, the VCMM estimates a dynamic statistical relationship among global risk factors and asset returns. Based on these calculations, the model uses regression-based Monte Carlo simulation methods to project relationships in the future. By explicitly accounting for important initial market conditions when generating its return distributions, the VCMM framework departs fundamentally from more basic Monte Carlo simulation techniques found in certain financial software. The primary value of the VCMM is in its application to analyzing potential client portfolios. VCMM asset-class forecasts comprising distributions of expected returns, volatilities, and correlations are key to the evaluation of potential downside risks, various risk-and-return trade-offs, and diversification benefits of various asset classes. Although central tendencies are generated in any return distribution, Vanguard stresses that focusing on the full range of potential outcomes for the assets considered, such as the data presented in this paper, is the most effective way to use VCMM output. 7

P.O. Box 2600 Valley Forge, PA 19482-2600 Connect with Vanguard > vanguard.com Vanguard research > Vanguard Center for Retirement Research Vanguard Investment Strategy Group E-mail > research@vanguard.com 2013 The Vanguard Group, Inc. All rights reserved. Vanguard Marketing Corporation, Distributor. ICRHMR 062013