Evaluating the Selection Process for Determining the Going Concern Discount Rate

Transcription

1 By: Kendra Kaake, Senior Investment Strategist, ASA, ACIA, FRM MARCH, 2013 Evaluating the Selection Process for Determining the Going Concern Discount Rate The Going Concern Issue The going concern valuation is meant to be an aid in the budgeting process toward full funding and, in most cases, works alongside the solvency valuation to promote the long term viability of the plan. The Standards of Practice that guide plan sponsors through the process of choosing the going concern discount rate for actuarial valuations are not highly restrictive and have historically resulted in a broad range of assumptions among pension plan sponsors. So, how does the selection of the going concern discount rate influence the funding process and what effect should it have on the investment decision? A Different Perspective The going concern discount rate, although an important element of the funding decision, is not something that should have any effect on the investment decision. After all, it s just an assumption, whose primary use is to control the strength/weakness of the funding regime. It influences how a plan sponsor manages two tradeoffs: the tradeoff between pay now or pay later, and the tradeoff between low cost and stable cost. Evaluating the Selection Process for Determining the Going Concern Discount Rate p 1

2 Background Financial economists and actuaries have been debating best practice for valuing pension liabilities for decades. Financial economists have historically argued that liabilities should be discounted using current market yields and that those yields should reflect the high credit quality of entitlements being promised and the duration of projected plan cash flows. Actuaries have supported that view with the caveat that an actuary should be afforded the flexibility to apply judgment on the basis of each particular situation. In addition, some actuaries have argued that, over the long term, the discount rate used to value liabilities for budgeting purposes should be consistent with the long term return expectation of the plan s asset portfolio. My colleague Bob Collie points out that nothing is certain in life, and the provision of retirement income played out as it is over decades is an especially uncertain game. 1 Further, he outlines three aspects of uncertainty: 1. Funding. How much money is set aside today? 2. Investment. How much risk is being taken in the investment of those assets? 3. Backstop. What happens in the case of a shortfall The investment risk taken on should evolve as a balanced tradeoff between the upside return potential and the downside risk of increased contributions The key takeaways from his paper were (1) that those three aspects are inter-dependent and (2) that the backstop role is an intrinsic component of security, all the more so when funding is weak, and/or investment is aggressive. For some plans, Canadian regulators have prudently adopted a balanced solution, ensuring that plans are funded on a going concern basis (the projected value of future benefits discounted relative to the expected return on plan assets) and on a solvency basis (the market value of existing accrued entitlements). As a natural extension, the investment risk assumed should then evolve, in concert with each particular plan s governance process, as a balanced tradeoff between the upside return potential and the downside risk of increased contributions. The key advantage of having both requirements is that the flexibility of the going concern approach is not allowed to go so far as to leave a consistent shortfall on a solvency basis. 1 Bob Collie A perspective on retirement security: Who stands behind America s pensions? Russell Research Evaluating the Selection Process for Determining the Going Concern Discount Rate p 2

3 Other plans (such as public, multi-employer and target date) are not required to fund on a solvency basis. In those cases, the going concern discount rate becomes the key assumption in the funding exercise. 2 In these cases, ignoring the solvency position neglects an essential aspect of uncertainty, and hence, increases the likelihood that a backstop may be required. Going Concern Discount Rate The liability discount rate is based on the long term expected return on portfolio assets. It is determined by the plan actuary 3, to some extent discretionary, and typically calculated using a building block approach composed of component factors such as inflation and expected real returns for each broad asset class. The regulations also allow for a risk adjustment range in this calculation to account for a variety of uncertainties. The discount rate decision is meant to represent a best-estimate assumption and be a reflection of the long-term expected return on portfolio assets. The actuarial professions role is to provide guidance to the process of selecting assumptions for funding valuations. 4 In cases where minimum statutory contributions are not affected by a solvency valuation, the Canadian Institute of Actuaries (CIA) recommends that the process for determining the going concern discount rate begin with the selection of a best estimate assumption and further introduce a plan specific margin of adverse deviation for conservatism, or a Provision for Adverse Deviation (PfAD). 5 Intended as a budgeting tool, the rate is a reflection of the long-term expected return on portfolio assets 2 In general, the pension liability will change over time as a result of benefit accruals and improvements, actual versus expected plan experience, plan closure or freezing, changes in plan demographics and regulations. However, the pension liability will be the most sensitive to changes in the liability discount rate and inflation Source: Asset Allocation and Risk Management for Defined Benefit Pension Plans: A Canadian Perspective Kaake, K Actuarial Standards of Practice (ASOP) #27 - Selection of Economic Assumptions for Measuring Pension Obligations: Best- Estimate Range - For each economic assumption, the narrowest range within which the actuary reasonably anticipates that the actual results, compounded over the measurement period, are more likely than not to fall. 4 ASOP 27 General Selection Process The general process for selecting economic assumptions for a specific measurement should include the following steps: a. identify components, if any, of each assumption and evaluate relevant data; b. develop a best-estimate range for each economic assumption required for the measurement, reflecting appropriate measurement-specific factors; and c. further evaluate measurement-specific factors and select a specific point within the best-estimate range. With respect to some (or all) of the components of an economic assumption, the actuary is not required to identify the explicit bestestimate range before selecting the specific point, provided that the actuary is satisfied that the selected point would be within the best-estimate range had such range been explicitly identified. After completing steps (a) through (c) for each economic assumption, the actuary should review the set of economic assumptions for consistency. 5 Provision for Adverse Deviation (PfAD) that is appropriate for the plan s risks and consistent with Plan s the risk management process. A margin of 50 basis points is common and may result in a PfAD of 8% of plan liability for a fully funded plan with a typical maturity profile. Canadian Institute of Actuaries, Task Force on the Determination of Provisions for Adverse Deviations in going Concern Valuations. January Provisions for Adverse Deviations in going Concern Actuarial Valuations of Defined Benefit Pension Plans Evaluating the Selection Process for Determining the Going Concern Discount Rate p 3

4 Budgeting versus Pricing Although most pension plans in Canada have the obligation to fund benefits on both a solvency and a going concern basis (which improves the stability and consistency of contributions), many plans rely on the going concern assumptions to determine the pricing of future benefit accruals. A return assumption that is higher (lower) than the achievable return associated with the asset allocation decision may lead to an under (over) pricing of future benefit entitlements. Moreover, for plans that have altered the level of risk exposure within their portfolio (typically when the funded status diverges from a desired range), the pricing of future benefits often diverges from the desired long-term risk-return profile for the plan. When a plan falls below an acceptable funded level and does not have the desire or the ability to fund through contributions alone, the plan sponsor may choose to increase their exposure to risk (and hence their return assumption) in an effort to improve the funded position. Without a corresponding increase in the PfAD, the impact of a higher portfolio risk profile with its associated higher return assumption is two-fold: 1. A decrease in the incremental price of benefits going forward. 2. An immediate improvement in the going concern funded status of the Plan as the present value of future benefits is decreased. Note that in both instances, entitlements are discounted with a return assumption, ahead of the associated risk. To highlight the implications of this, consider an organization with the following two Plans (A and B): A risky investment strategy can make a given benefit promise appear less expensive Plan Comparison Plan A Plan B Payment due in 30 years $250,000 $250,000 Current Investments Stocks $15,000 $15,000 Bonds $15,000 Total $30,000 $15,000 Expected Return 6% 8% Expected Liability $43,528 $24,844 Unfunded Liability $13,528 $9,844 Evaluating the Selection Process for Determining the Going Concern Discount Rate p 4

5 While both plans have a $250,000 payment due in 30 years, Plan A has $30,000 and is allocated equally into stocks and bonds, for an expected return of 6%. However, Plan B invests only $15,000 exclusively in stocks expected to earn 8%. Ironically, the expected and unfunded liabilities for Plan A (at $43,528 and $13,528 respectively) end up being larger than that of Plan B (at $24,844 and $9,844 respectively). That is to say, a risky investment strategy can make a given benefit promise appear less expensive, which comes perilously close to assuming that a dollar invested in equity is worth more than a dollar invested in fixed income. This can lead to significant intergenerational inequity among participants over the life of a plan. To help resolve the issues associated with relying on the going concern discount rate to price benefits, a more appropriate solution to pricing might be to use a conservative return expectation. This expectation would be consistent with the long-term level of risk exposure appropriate for the Plan, in a fully funded position. Given that individual plan dynamics may vary significantly from plan to plan, regular asset-liability studies should be employed as an aid in this process. 6 Similar intergenerational inequity can also evolve when management adopts a short-term focus in a manner inconsistent with the long-term nature of the assumptions and/or the asset allocation decision. 7 Of course, although the potential for concern is much more pronounced in plans that are not required to fund on a solvency basis 8, an unattainable going concern discount rate assumption will nonetheless distort gains and losses over time. Therefore, the discount rate choice, if relied upon for budgeting purposes, should be a reflection of the investment strategy but not the reverse. Put another way, it is important to take care that the discount rate choice does not affect the investment decision. It is important to take care that the discount rate choice does not affect the investment decision 6 Note that the CIA recommends actuaries address the following question when setting the going concern discount rate: With a PfAD of x% in a fully funded plan, what is the probability that the plan will be fully funded at some future date? Canadian Institute of Actuaries, Task Force on the Determination of Provisions for Adverse Deviations in going Concern Valuations. January Provisions for Adverse Deviations in going Concern Actuarial Valuations of Defined Benefit Pension Plans 7 Typical examples would include: granting (removing) benefit entitlements as a direct result of short-term economic gains (losses); or, setting contribution rates based on the cost of an additional year of benefits and pricing the cost using a discount rate assumption that does not reflect the long-term risk budget of the plan; or even if the assumption turns out as it inevitably will to be different from actual experience. 8 Plans such as public, multi-employer or target date have the option to fully omit or partially omit (post-retirement cost of living adjustments, for example) entitlements from the Solvency valuation process. Single employer corporate plans are typically required to fully fund on a Solvency basis. Evaluating the Selection Process for Determining the Going Concern Discount Rate p 5

6 Observed Asset Returns To help us understand the limitations inherent in choosing a single best estimate discount rate assumption to value and price the liability it is important to assess the distribution of observed returns and volatility. The actual observed return distribution is commonly different from the theoretical normal distribution. Observed returns typically have fat tails (extreme events occur more often) and skewness (the distribution has one tail that is longer than the other). In addition, the variance of returns is characterized by variability over time, mean reversion and clustering. Forecasted Asset Returns Ultimately, the best estimate return assumption should be set so that there is an equal likelihood of the actual value being greater or less than the expected value. To isolate the practical implications of the best estimate assumption, we ve utilized Russell s capital market assumptions as an aid in exploring the impact of market risk, active risk and diversification on the long term funding viability of a typical Canadian pension plan. To this end, we ve looked at distributions of forecasted outcomes. Static return assumptions are not consistent with either financial theory or capital market history. Moreover, to help us understand why the average return expectation might not reflect economic reality, we ve generated a range of forecasted returns, using Russell Capital Market assumptions. Exhibit 3 shows the evolution of cumulative annualized returns over a 20-year horizon. For this traditional portfolio (60% equity / 40% bonds), we ve forecasted the distribution of possible outcomes across a variety of horizons. In each year the results are ranked from worst to best case outcomes to produce a range of financial outcomes (uncertainty), worst case outcomes (downside risk) and expected or average outcomes (trends). Ultimately, there should be an equal likelihood of the actual value being greater or less than the expected value Interestingly, over the mid and long term, average return expectations are greater than median return expectations. For example, our simulated outcomes produced an average annualized return of 5.7%; however, if we rank the simulations and choose the median path (the 2,500th path in a simulated set of 5,000) we find that the cumulative annualized return is only 5.2%. This spread between average and median is a reflection of the distribution of asset returns across time, where skewness, fat tails and volatility play an important role in forecasting future outcomes. Evaluating the Selection Process for Determining the Going Concern Discount Rate p 6

7 Total Return Note that by increasing (decreasing) the allocation to return seeking assets we would see an increase (decrease) the dispersion of results which, by and large, would result in an increased (decreased) exposure to economic downside risk. Exhibit 3: Evolution of cumulative annualized returns over a 20-year horizon for a passively managed portfolio. 9 25% 20% Passive Portfolio 15% 10% 5% 0% -5% -10% -15% 1-year 3-year 5-year 10-year 20-year 95% 19.0% 13.0% 11.2% 9.8% 8.9% 75% 10.3% 7.6% 7.1% 6.8% 6.7% 50% 4.2% 4.0% 4.1% 4.7% 5.2% 25% -2.1% 0.0% 1.0% 2.5% 3.6% 5% -12.6% -6.3% -3.8% -0.7% 1.3% Average Return 3.9% 4.1% 4.4% 5.1% 5.7% Importantly, Russell s current assumptions result in both the average and the median cumulative annualized returns being comparatively less than the typical discount rate assumption for going concern valuations in Canada. Additional concerns arise when we take this analysis further to look at the probability of exceeding the cumulative going concern return (Exhibit 4). When comparing forecasted results to a static annual going concern discount rate of 6.25%, the probability of exceeding our estimate for a passively managed portfolio (60% equity / 40% bonds) is less than 35% over the long term. 9 Russell strategic planning assumptions at December 31, Model contains 30% Canadian Equity, 30% Global Equity and 40% DEX Universe Bonds. Forecast produces 5, year paths of portfolio returns and interest rates Assets are calculated along each path. Model assumes annual rebalancing. The model portfolio does not represent an actual portfolio and is shown for illustrative purposes only. Refer to Exhibit 7 for a more detailed summary of the 20-year distribution statistics. Evaluating the Selection Process for Determining the Going Concern Discount Rate p 7

8 Total Return Exhibit 4: Probability of exceeding the typical long-term going concern return expectation. 1-Yr 3-Yr 5-Yr 10-Yr 20-Yr Going Concern Discount Rate 6.25% 6.25% 6.25% 6.25% 6.25% Cumulative Going Concern Return 106% 120% 135% 183% 336% Number of forecasted paths with returns exceeding the Going Concern discount rate 2,064 1,715 1,556 1,530 1,617 Probability of exceeding the Cumulative Going Concern Discount Rate 41% 34% 31% 31% 32% Active Management When making forecasting adjustments to reflect assumptions for active management, the picture becomes more positive. 10 Our simulated outcomes produce an average return of 6.8% and, for the median path, a return of 6.3% at a 20-year horizon. As a caveat, this invariably assumes that the plan sponsor can implement an effective active management program to contribute positive excess returns. Exhibit 5: Evolution of cumulative annualized returns over a 20-year horizon for an actively managed portfolio % 20% 15% 10% Active Portfolio 5% 0% -5% -10% -15% 1-year 3-year 5-year 10-year 20-year 95% 20.5% 14.2% 12.4% 10.9% 10.1% 75% 11.5% 8.9% 8.2% 7.9% 7.8% 50% 5.3% 5.1% 5.2% 5.7% 6.3% 25% -1.1% 1.0% 2.0% 3.5% 4.6% 5% -12.0% -5.3% -2.8% 0.4% 2.4% Average Return 5.0% 5.2% 5.4% 6.1% 6.8% 10 Russell offer rigorous, robust, logically consistent alpha forecasts for all asset classes that have monthly pricing and returns data. These forecasts allow us to make more accurate and appropriate recommendations to clients and products for clients and are consistent with Russell s mission of Improving Financial Security for People. A fundamental starting point for forecasting alpha is a method for explicitly assuming a modest and reasonable level of manager selection skill. 11 Russell strategic planning assumptions at December 31, Model portfolio contains 30% Canadian Equity, 30% Global Equity and 40% DEX Universe Bonds. Forecast produces 5, year paths of portfolio returns and interest rates Assets are calculated along each path. Model assumes annual rebalancing. Forecast assumes Alpha of 1.1% and Tracking Error of 1.6% Evaluating the Selection Process for Determining the Going Concern Discount Rate p 8

9 Notwithstanding, when comparing forecasted results under this paradigm we do see an improved probability of exceeding the cumulative going concern 6.25% return assumption, to as much as 50% at the long end of our 20-year horizon (exhibit 6). Exhibit 6: Probability of exceeding the typical long-term going concern return expectation. 1-Yr 3-Yr 5-Yr 10-Yr 20-Yr Going Concern Discount Rate 6.25% 6.25% 6.25% 6.25% 6.25% Cumulative Going Concern Return 106% 120% 135% 183% 336% Number of forecasted paths with returns exceeding the Going Concern discount rate 2,268 2,096 2,027 2,218 2,513 Probability of exceeding the Cumulative Going Concern Discount Rate 45% 42% 41% 44% 50% Summary The going concern best estimate assumption is a very important tool in the budgeting process. Nonetheless, it is a forecast and, by definition, does not come without error. Extending the use of this assumption beyond the general budgeting exercise can lead to poor governance, the under-pricing of future benefit entitlements and intergenerational inequity, especially in cases where the PfAD fails to reflect an adequate level of conservatism. Moreover, the going concern discount rate should not have any influence on the investment decision. The real world peculiarities in the distribution of observed returns, combined with the timing asymmetry of contributions and benefit payments, can have important implications to the long-term funding viability of the Plan. These considerations, among others, will not be reflected in a static discount rate assumption. Plan sponsors should consider relying on a conservative best estimate assumption and include a plan specific adverse margin for uncertainty to ultimately determine the going concern discount rate. This is particularly important if the full solvency position is not included in the funding process. Regular asset-liability studies should also be considered as an aid in setting an investment allocation strategy, setting risk tolerance budgets and pricing future benefit entitlements. on a total portfolio level, net of fees. The model portfolio does not represent an actual portfolio and is shown for illustrative purposes only. Refer to Exhibit 7 for a more detailed summary of the 20-year distribution statistics. Evaluating the Selection Process for Determining the Going Concern Discount Rate p 9

10 REFERENCES Collie, Bob A perspective on retirement security: Who stands behind America s pensions? Investments and Wealth Monitor by Investment Management Consultants Association (IMCA), Actuarial Standards Board Actuarial Standard of Practice No. 27 Selection of Economic Assumptions for Measuring Pension Obligations Canadian Institute of Actuaries, Task Force on the Determination of Provisions for Adverse Deviations in going concern Valuations Provisions for Adverse Deviations in going concern Actuarial Valuations of Defined Benefit Pension Plans Collie, Bob Corporate bonuses and the error of EROA Russell Research Appendix Our research has found that forecasts based on the sample statistics of historical return data are not the strongest indicators of future market outcomes. These forecasts fail to incorporate any of the lessons from academic research on asset class behavior. Russell s approach to setting risk/return assumptions is based on a combination of financial theory and capital market history. Russell s capital market forecasts are revised every quarter or more frequently if market circumstances change materially. Our forecasting process is designed to project market outcomes over strategic planning horizons and is constructed by identifying and estimating structural relationships among market elements. Projections of financial outcomes are then combined with investor risk preferences in the development of asset allocation and liability management strategies. The forecasting system is structured around a core set of macro-economic variables. The following are the distinctive features of our forecast model: Inputs that are forward-looking, yet consistent with history (we consider historical means as good reflections of history but poor predictors of future returns) Inputs that dynamically adjust by time-period (static inputs are not consistent with either financial theory or capital market history) Relationships between asset classes that are modeled directly and consistently across each scenario, rather than indirectly through correlations alone. Evaluating the Selection Process for Determining the Going Concern Discount Rate p 10

11 Exhibit 7: Detailed Summary of Distribution Statistics (20-year) 12 Passive Portfolio Active Portfolio RETURN SUMMARY STATISTICS Average 25th 75th Average 25th 75th Number of Periods th Percentile Return 27.0% 28.4% 2nd Percentile Return -17.4% -16.5% Average Return 5.7% 6.8% Annualized Return 5.7% 6.8% Cumulative Return 202% 270% Excess Return Over Risk-Free 2.6% 3.7% Average Excess Return Over Risk-Free 2.8% 3.9% Excess Return Over Market -2.0% -0.9% Average Excess Return Over Market -2.0% -0.9% RISK SUMMARY STATISTICS Average 25th 75th Average 25th 75th Beta Beta T-Stat Upside Beta Downside Beta Upside Semi-Variance 1.6% 1.2% 2.0% 1.8% 1.4% 2.2% Downside Semi-Variance 0.9% 0.4% 1.2% 0.9% 0.4% 1.1% Upside Semi-Standard Deviation 12.6% 11.1% 14.0% 13.4% 11.8% 14.7% Downside Semi-Standard Deviation 8.8% 6.4% 10.8% 8.5% 6.0% 10.7% RISK/RETURN STATISTICS Average 25th 75th Average 25th 75th Annualized Standard Deviation 10.3% 9.0% 11.6% 10.5% 9.1% 11.7% Variance 1.1% 4.1% 7.2% 1.1% 5.2% 8.3% Sharpe Ratio Excess Return Over Market/Risk -19.0% -8.6% Sortino Ratio CORRELATION STATISTICS Average 25th 75th Average 25th 75th R-Squared Standard Error Correlation Footnotes for this analysis: Model contains 30% Canadian Equity, 30% Global Equity and 40% DEX Universe Bonds. All statistics that show average, 25th and 75th are calculated for all 5,000 paths and the summary statistics show are based on the statistics calculated individually on the paths. All beta and market relative statistics are versus Global Equity. All semi-variance and semi-standard deviations statistics are based on the total return of the portfolio in relation to a neutral return of 0%. Standard deviation calculations are done by calculating the standard deviation of all 5,000 forecasted paths and showing the average, 25th percentile and 75th percentile of those potential standard deviations. This is a different measure than what is analyzed in the placemat and other analysis (where the standard deviation measures the potential dispersion in 10 or 20 years, rather than the volatility of the path) and therefore will not match. R-Squared is the percentage of the excess return of the portfolio over cash that is explained by the excess return of Global Equity over cash. The correlation is the correlation of excess return of the portfolio over cash with the excess return of Global Equity over cash. The standard error is a measure of the excess returns of the portfolio over cash that are not explained by the excess return of Global Equity over cash. It is calculated as the square root of the average of the squared errors based on the calculated betas for each path. Evaluating the Selection Process for Determining the Going Concern Discount Rate p 11

12 Important information Nothing in this publication is intended to constitute legal, tax securities or investment advice, nor an opinion regarding the appropriateness of any investment. This is a publication of Russell Investments Canada Limited and has been prepared solely for information purposes. It is made available on an as is basis. Russell Investments Canada Limited does not make any warranty or representation regarding the information. Forecasting or other forward-looking information is inherently uncertain, maybe incorrect, and is not representative of a projection of the stock market, or of any specific investment. There is no guarantee that any stated expectations will occur. References to future returns for either asset allocation strategies or asset c lasses are not promises or even estimates of actual returns a client portfolio may achieve. Asset classes are broad general categories which may or may not correspond well to specific products. Russell Investments Canada Limited is a wholly owned subsidiary of Frank Russell Company and was established in Russell Investments Canada Limited and its affiliates, including Frank Russell Company, are collectively known as Russell Investments. TRADEMARKS The Russell logo, and any Russell indices are either trademarks or registered trademarks of Frank Russell Company. Copyright Russell Investments Canada Limited All rights reserved. The contents of this report are intended for the recipient of the report only and are not be reproduced, transferred or distributed in any form without prior written permission from Russell Investments Canada Limited. Date of publication: April 2013 INST (EXP-04-14) Evaluating the Selection Process for Determining the Going Concern Discount Rate p 12