Credit Crisis Driven Changes to Asset Allocation and Spending Rates for College Endowments

Transcription

1 Credit Crisis Driven Changes to Asset Allocation and Spending Rates for College Endowments Walter P. Neely Professor of Finance And Bill M. Brister Assistant Professor of Finance Else School of Management Millsaps College Abstract The Credit Crisis of 2008 has caused institutional and individual investors alike to rethink their spending and investment strategies and our finding is that asset allocation, expected returns, and variability in correlations are all important factors, but the withdrawal rate may be the most important. Withdrawal in excess of 4% or 5% can cause an endowment to be underwater from historical cost for many years. When inflation is (as it should be) considered, excess withdrawal causes endowments to be even more underwater. Colleges can reduce their budgets thereby reducing withdrawal rates, or colleges can raise funds (not endowment) to meet spending needs. Other factors such as lower equity risk premiums increases the risk of failure to meet spending needs, especially under the real world assumption that 5% is withdrawn for spending by the educational institution. Higher equity risk premiums reduce failure risks, but if returns are lower the risk of failure in greater. Endowments cannot control the magnitude of the equity risk premium, but the net withdrawal rate is controllable. The effects of lower returns on all asset classes as is the current expectation, puts extra pressure on portfolios because spending 5% when average returns are lower than 8% reduces real values of portfolios. These results appear to hold for the historical period that includes several major periods of market stress. Treasury bonds, both long and intermediateterm, are less correlated with equities especially during stress periods, and adding intermediate term government bonds reduces the risks during stress periods, so increasing the allocation to fixed income (government and corporate bonds) offers protection during periods of stress like the credit crisis of However the best protection is fiscal discipline.

2 Credit Crisis Driven Changes to Asset Allocation and Spending Rates for College Endowments I. Introduction The Credit Crisis of 2008 has caused institutional and individual investors alike to rethink their spending and investment strategies including their asset allocation strategies (Lauriella, 2009). Asset allocation did not protect investors, including college endowment portfolios, from dramatic declines in market values of their portfolios. Investors are questioning whether the relationships between asset class returns have broken down (Lauriella, 2009). Diversification seems to have failed with US and foreign stock markets moving downward together, along with commodities, private equity, and corporate bond markets. Almost all other asset classes fell along with stocks in Relationships between risk and return seem to have turned upside-down, with bonds (especially government) beating stocks for the holding period. Liquidity has also been a concern of endowment managers as markets for private equity and toxic debt holdings seem to have dried-up. Investment and spending policies face more and more scrutiny by state regulators as markets have dealt with the 2001 and 2008 market crises. Endowment managers are concerned that reliance on the old asset allocation conventional wisdom may not be sufficient for today s portfolios, and colleges must plan spending needs while preserving the values of their endowments. The behavior of asset class returns during periods of market stress may be different than the average historical behavior of asset classes, and asset allocation theories may fail if they are based on these long-term historical studies. In this paper we examine long-term historical returns. One difference between history and the future is the possibility of lower future returns. Another issue we examine is the effect on the portfolio of spending requirements under various equity risk premium scenarios. Returns will vary from the historical averages, and the possibility of lower future returns affects the asset allocation required to meet spending needs of colleges. We examine all aspects of risk including failure to meet spending needs of the college. Another issue we consider is liquidity risk in the sense of government versus corporate bonds. Liquidity risk can be reduced with asset allocations tilted toward government bonds, so we compare outcomes with both corporate and government bonds. Our focus is on the asset allocation required to preserve the spending needs of colleges in light of the lessons of the 2008 credit crisis and the new laws governing endowments. 1

3 II. Spending by College Endowments Endowment funds are donated funds held by an institution like a college. Donors may restrict their gift to make it permanent, and these gifts are the type of endowments which we address in this paper. Some gifts to colleges are not donor restricted, but the gifts may be restricted as to purpose or as to ability to spend the corpus. We do not address these board restricted parts of endowment funds, but we do address endowment funds expected to be held to perpetuity. Educational endowments often target annual spending of 5 to 5.5% of their endowment fund s value, according to the Common Fund Institute (2001), and some colleges in need of operating funds spend more than these targeted percentages in order to balance their budgets. The governing laws for endowment spending are state laws that until recently restricted spending to amounts exceeding the original or historic values of the endowment gifts. These state laws are based on the Uniform Management of Institutional Funds Act (UMIFA) which date back to the 1970s. Due in part to the volatility of financial markets, these laws are in the process of change. The National Conference of Commissioners on Uniform State Laws (NCCUSL) approved in 2006 the Uniform Prudent Management of Institutional Funds Act (UPMIFA, 2006) and recommended certain changes including changes to endowment spending. The UMPIFA eliminates the historic value concept replacing it with seven criteria providing guidance on prudence. Among the seven criteria that directors should consider are the effects of inflation, the expected total return, other resources of the institution, and general economic conditions. Since these criteria may be considered less binding, the UPMIFA included an optional provision that creates a presumption of imprudence if an institution spends more than seven percent of the endowment fund. Since approval of the UMIFA by 35 states in the period after 1972 (Rowland, 2009) the historic value rule was not particularly binding to older endowments. Market values rose well over historic nominal values over the period. Purchasing power was not considered in the older UMIFA laws. The general rules followed by most endowments may have considered the effects of inflation. The new UPMIFA based laws provide guidance to maintain purchasing power. Rowland (2009) quotes the drafting committee saying that colleges must establish a spending policy that will be responsive to short-term fluctuations in the value of the fund. The new rules also allow flexibility during down markets in restoring historic value by means of additional gifts (presumably not endowments but unrestricted gifts). Spending might be allowable in rates exceeding norms if additional (non endowment) gifts or income make up for the spending out of endowment. 2

4 The financial crisis of 2008 has caused a spending crisis for colleges (and other nonprofits); especially those located in the 24 states that have not passed the new UPMIFA based laws (Hechinger and Levitz, 2009). As of June 2008, 39% of colleges and other schools reported underwater endowments, and since the crisis had not reached the low prices of March 2009, many others must have broken historic values. Recently, efforts to pass the UPMIFA based laws have added sense of urgency. The new guidance from UPMIFA allows colleges (and other nonprofit endowments) to spend so much of an endowment fund as the institution determines to be prudent for the uses, benefits, purposes and duration for which the endowment fund is established. Factors specified are general economic conditions, "expected total return from income and the appreciation of investments, and "other resources of the institution that should be considered by colleges as they manage their endowments (UPMIFA, 2006, p. 6). Whether the college endowment is located in a UPMIFA or a UMIFA state, spending policy and investment policy must be integrated, and both policies are under increased scrutiny as a result of the crisis of This paper deals with the risks of not meeting spending needs of endowments, and it deals with the required strategic long-term asset allocation policies needed to meet spending requirements. We examine aspects of the seven criteria established in the UPMIFA. We focus on spending rate, expected return assumptions, inflation impacts, other institutional resources and the investment policy of the institution. College spending needs affect asset allocation, portfolio return, risk, and endowment liquidity result from the interplay of spending and asset allocation. It is this interconnected system that we study in the context of the credit crisis of 2008 and the changes in legal requirements resulting from adoption of the UPMIFA. III. Asset Allocation Conventional wisdom based on long-term studies suggests that asset allocation, the selection of long-term weights assigned to each asset class, not the selection of individual investments, accounts for 85% to 95% of return according to Brinson, Hood, Beebower (1986) and Brinson, Beebower, and Singer (1991). Blake, Lehmann, and Timmermann (1999) who found similar results for UK pension funds, conclude that strategic asset allocation, also called constant mix, accounts for most of the return variability. Perold and Sharpe (1988) examine four asset allocation strategies. Buy and hold or do nothing as a conscious asset allocation strategy is compared to three strategies that require periodic portfolio rebalancing. The most basic or uncomplicated asset allocation rebalancing strategy is called constant mix which involves buying 3

5 under-weighted and selling over-weighted asset classes in order to return the portfolio to its original (constant) mix. Implementation of the constant mix strategy requires buying poorly performing asset classes and selling the best performers. Constant mix requires the purchase of stocks as they fall in value, at least in relative terms. 1 Arnott and Lovell (Maginn and Tuttle, 1990) list the benefits of disciplined rebalancing as: controlling drifting mix, avoiding market timing, a commitment to policy, a way to add value, and avoiding momentum investing. Rebalancing frequently involves buying the asset class whose relative value has declined. When portfolios are not frequently rebalanced the resulting portfolio contains winning asset classes whose asset allocation has expanded. Rebalancing frequently does add to transactions costs including higher commissions and market impact costs. But these models assume that historic correlations between returns of asset classes will be similar to historical averages. A study by Arshanapalli, Coggin, and Nelson (2001) testing tactical versus strategic asset allocation finds that a broker model (tactical) average return for is 13.59% with a Sharpe ratio of The Markowitz model return is 11.84% with a Sharpe ratio of 0.91%. A constant mix of 55/35/10 (also called the Robot Blend) results in a return of 13.38% with a Sharpe ratio of They conclude that a constant mix of 60/30/10 blend is not a bad choice. Asset allocations must be rebalanced periodically in order to adhere to policy and in order to control risk. Neely and Brister (2006) found that rebalancing does control the risk of asset allocation equity creep. They considered annual rebalancing along with 5% and 10% rebalancing rules, both of which require less portfolio monitoring and result in lower transaction costs compared to more frequent rebalancing. They concluded that rebalancing by percent rule or annual rebalancing results in similar risks whether measured by failure rate, variability or asset allocation weight creep. In this paper we assume annual rebalancing in our tests of asset allocation. IV. Risk Premiums and Asset Allocation Black and Litterman (1991) state that the risk premium on the market portfolio is a difficult number to estimate. They analyze asset allocation and the equity risk premium in the context of universal hedging in a multi asset global portfolio. They compare historical risk premiums to investor s expectations of future risk premiums illustrating the impact on asset allocation of lower expected risk premiums. Graham and Harvey (2005) 1 Perold and Sharpe (1988) also examine an asset allocation strategy called constant proportion portfolio insurance (CPPI) which involves the sale of the sale of stocks which fall in value and the corresponding purchase of stocks as they rise in value. The fourth asset allocation examined by Perold and Sharpe is a variant of CPPI using options. 4

6 state that the traditional methodology of estimating the market risk premium has involved the use of historical data from Ibbotson Associates (2008), and they cite a difference of 8.21% in large company returns minus T-bill returns. Fama and French (2002) find that use of historical data is flawed by the large capital gains that were not expected by investors. In fact, Ibbotson and Chen (2003) provide an estimate of 4% using geometric returns and 6% using arithmetic averages for future market risk premiums. Graham and Harvey (2005) estimate the market risk premium using a survey of expectations of chief financial officers, and they find a ten year premium of 2.9% to 4.7%. A recent study by Fernandez (2009) finds that textbook authors have reduced their recommended equity risk premiums from an average of 8.4% in 1990 to 5.7% in After the buoyant markets of the late 90s, expected equity risk premiums (expected future returns) were expected by many to be smaller (Arnott, 2004), but that forecast was prior to the 2008 credit crisis. Arnott (2004) suggests that future returns will not duplicate historical returns. With lower current than past dividend yields and higher current P/E ratios, he expects lower capital gains and total returns. With observable risk free rates on Treasury securities combined with lower expected total returns, he concludes that the equity risk premium that averaged 7-8% from 1926 to 2003 will be closer to 2% in the post 2002 horizon period according to Arnott (2004). Cornell, Arnott, and Moroz (2009) in a more recent paper find a current (2009) expected equity risk premium to be four percent, double the 2003 Arnott (2004) forecast but still much lower than the historical average of at least six to eight percent, for geometric and arithmetic averages, respectively. Dramatically lower equity prices suggest that post 2008 equity risk premiums may have grown at least temporarily from the lows of post 2001 to higher levels in One sure thing is that risk premiums are variable. Equity risk premium variability affects asset allocation decisions, and endowment managers should consider the variability in risk premiums as they consider asset allocation and spending policies. Asset allocation appears straightforward in theory but in practice it is difficult to know expected return, risk and correlations. If future equity risk premiums and total returns on stocks are indeed lower, there are significant implications for asset allocation. V. Hypotheses The seven criteria established in the UPMIFA along with the optional spending cap represent the bases of our hypotheses dealing with permanent endowment funds of colleges. The spending or endowment withdrawal rate is a key risk which is lessened at lower spending levels. To the extent that additional non-restricted funds are raised to provide other institutional resources, spending levels from endowment and nonendowment funds may be sustained. Expected returns are another criterion which must 5

7 be considered by college boards, and achieving higher returns should generate greater spending ability and lower risk of shortfall in keeping the endowment permanent. Lower returns (and expected risk premiums) conversely increase shortfall risk. The impact of inflation impacts the real value of the permanent endowment, and to the extent the endowment nominal return is not sufficient to provide spending and cover inflation, the real value of the endowment fund will decline. Spending in excess of reasonable percentages will reduce the real value of the endowment below its historic or previous values. The investment policy of the institution includes the asset allocation, and an important aspect of asset allocation is the mix between equity and bonds. The greater the allocation to equity, the greater are some risks. However, higher equity allocations should provide for greater spending or withdrawal rates over the long run. It is this interconnected system that we study in the context of the credit crisis of 2008 and the changes in legal requirements resulting from adoption of the UPMIFA. We expect spending moderation to reduce failure risk and equity weight risk. If equity risk premiums fall to 4% or lower, we expect less volatility in portfolio weights. We contrast these measures of risk and return at different asset allocation levels to examine the impact on risk and return for differing asset allocations. We also examine the impact of spending (portfolio withdrawal) requirements on the asset allocation and rebalancing variables and the impact on the values of the portfolio. We develop a measure of failure of the portfolio to meet the spending requirement. Our measure is similar to the success rate developed by Cooley, Hubbard and Walz (1998) for individual investors. Bengen (1996, 1997, 2006) also illustrates the effects of withdrawals and rebalancing for individual's in retirement. Failure of investment returns to meet spending requirements results in a decline in the real (inflation adjusted) value of the portfolio. We expect lower risk premiums and lower equity returns to result in greater risk of spending pressure by educational institutions. Failure to sustain the real values of endowments will result if spending rates are not cut from pre high equity risk premium times. Correlation between asset class returns can vary, and we will examine subperiod correlations including stress (down market) periods. Building portfolios with stress in mind will impact asset allocation needed to meet college spending needs. VI. Data and Methodology Ibbotson Associates (2008) provide returns for US equity, intermediate term government bond, and long-term corporate bond asset classes for the 1926 through 2008 period. We consider rolling three, five and ten year periods during the period, so there are 79, 77, and 72 three, five, and ten year periods, respectively. We 6

8 also consider long-term and periods of market distress to examine the consistency of relationships between asset classes. We examine spending requirements in the context of the risk and return implications of buy and hold, constant mix with annual rebalancing. We test spending rates of 3% to 8% using strategic asset allocations of various stock / bond mixes or portfolio weights with alternative return and rebalancing scenarios. VII. Results Asset allocation policies are meant to control risk and to enhance return. Table 1 shows the risk and return statistics for various equity/bond mixes for the full period. Equities returned an arithmetic average of 11.6% with a standard deviation of 20.6%. The typical endowment may target a 70/30 equity/bond mix, and the portfolio average return is 9.8% using intermediate term government bonds and 10.0% using corporate bonds. Since the standard deviation risk is greater using corporate bonds and the likelihood of shortfall is greater (not shown), we will concentrate on intermediate term government bonds for purposes of this research, although use of corporate bonds and other assets should be considered by college endowment funds. There are meaningful increases in return and risk as the asset allocation increases from 60/40 to 80/20, and the other (in addition to standard deviation) risks associated with these asset allocations will be examined further in this paper. 7

9 VIII. Credit Crisis The Role of Correlation We examine the risk and return implications of changes in the correlations of markets during crisis periods. Correlations are not unchanging between returns of the asset classes of US securities markets. Correlations change both during crisis times and longer-term. We examine these changes especially during crisis times when we show increases in correlations during years of low equity market returns. We make connections between asset allocation, government/corporate bonds, and spending rates. One focus of the research is the effect on returns of holding more US government securities to survive market crises and to provide the liquidity needed during crisis periods. The financial crisis of 2008 which extended into March of 2009 was intense not only because of the depth of the declines but also because diversification seemed to fail as almost all risky asset classes fell in value. The top panel of Table 2 shows the correlations between Ibbotson assets classes for the time period. These correlations are compared to the correlations between the asset classes for down market years. Down markets are defined as years during which large capitalization stocks declined by more than 10%. In 2008 both large and small capitalization stocks fell about 37% while government bonds exhibited the best returns as investors flocked to the relative safety of government securities. Down market years are surely not all alike, but down market returns exhibit significant differences in correlations compared to full period correlations. 8

10 Down market correlations between asset classes are shown in the second panel of Table 2. Correlations are positive between stocks and other assets except for long-term government bonds. As interest rates fall during down markets long government bonds appreciate in price when stocks decline. Intermediate government bonds do not appreciate as much so the return is lower and the correlation is not negative, but slightly positive. Even T-Bills and stocks become quite positively correlated during down markets. The correlations of long term corporate bonds are greater than those of intermediate term governments, but not as much greater as expected, and corporate bond to equity correlations do not seem to change during down markets. Contrast the correlations during down markets to correlations for Negative or near zero correlations are exhibited between T-bills and equities. However, the correlation has changed from negative for the pre 1970 period (not shown) compared to the period when correlations were positive. More striking is the correlation during stress periods with correlations higher for T-Bills than for government bonds. Intermediate government bonds have nearly zero correlations with stocks on average for the full period. Hence long-term and intermediate term government bonds provide better diversification than T-bills during down markets. Corporate bonds provide low correlations with stocks and higher returns than intermediate term governments. Bonds provide a source of safety and liquidity especially during down markets, safety not only in the sense that their values are stable, but safety in the sense 9

11 that correlations are much lower during just the times when lower correlation are needed during down markets. Correlation explains one aspect of risk, but there are other ways to examine risk in the context of asset allocation of college endowments. IX. Risk, Investment Failure, and Spending Rates Risk in modern portfolio theory is most often defined as variability. Variability is an important aspect of risk to college endowments, but because of the competing objectives faced by endowments, other risks are important too. Colleges are required to maintain the historical or real values of funds given to them to perpetuity. At the same time college budgets are dependent on contributions from endowment total returns to meet college spending needs. Failure to meet spending needs competes with the need to maintain value to perpetuity. Failure of the portfolio occurs when the portfolio fails to maintain its nominal value over a specified period of time. True endowments must maintain nominal values, at least under 1972 prescriptions. The two key variables considered on the first panel (historical 6.12% equity premium) of Table 3 are withdrawal (spending) rate and asset allocation. We examine failure (to maintain value) rates for rolling investment periods of 3-years, 5-years, and 10-years for the 1926 through 2008 time period. The failure rate is the percentage of investment periods for which nominal value of the portfolio is not maintained. The failure rates for 60/40, 70/30 and 80/20 asset allocation schemes are shown. 10

12 Ten year investment periods exhibit lower failure rates than do three and five year holding periods, at least when withdrawal rates are low. As withdrawal rates approach 6 to 7%, however, the failure rates tend to rise regardless of asset allocation for the different investment holding periods. Failure rates for the 10-year investment periods tend to increase sharply as withdrawal rates increase from 6% to 8%. Chart 1 illustrates this finding. As shown for the 60/40 asset allocation, the failure rate for the 10-year investment period jumps from 16.22% at 6% withdrawal rate to 31.08% at 7% withdrawal rate. If failures occur during 31% of the ten year holding periods, most college endowments would find it unacceptable. 11

13 Failure rates do not appear to be different for various asset allocations tested, especially for ten year holding periods. When withdrawal rates are 5% or less failure occurs 12-13% of the ten year holding periods assuming historical 6.12% equity risk premiums. When withdrawal rates exceed 6%, failure rates rise unacceptably. Failure rates, perhaps unexpectedly, fall slightly as asset allocations are tilted higher in equity proportions. The higher returns that usually come with higher equity allocations also reduce failure rates. If equity risk premiums are lower, as shown in the bottom half of Table 3, the conclusions are similar as found above for historical premiums. Asset allocation does not appear to significantly increase failure risk, particularly at withdrawal rates of 6% or less. Ten year failure rates are an acceptable 13-15% for withdrawal rates up to 5%, but perhaps not acceptable for 6% and higher withdrawal rates. The impact of lower future return (equity risk premiums) assumptions limits withdrawal rates to lower proportions. There is some evidence that even though withdrawal rates must be reduced in a lower return world, higher equity asset allocations should add to return without increasing failure risk. The maintenance of permanent value is one of the key criteria set by UMIFA and UPMIFA, and it is important to donors to maintain value of endowment funds. Table 4 shows the length of time in years that portfolios fail to maintain their historical values. Episodes of underperformance are periods during which the ending value is less that the beginning value. Since withdrawal for spending reduces the ending value, investment returns must exceed the withdrawal rates. Given market volatility in is not unusual to see periods of time when failure occurs. 12

14 There are periods of several years in duration during which portfolios fall below their historic values. We define a down market episode as a period of at least 3 years until the market value of the portfolio regains its previous high value. These down market periods include the Depression years of the 1930s, the recent double crisis years of 2001 and 2008, and other less significant crisis periods during Various asset allocations are shown on Table 4. The first column shows that even without any withdrawals there are five crisis periods with durations of 7 years ( ), 7 years ( ), 3 years ( , 3 years ( ), and 4 years ( ). So, during the 81 year time span, there are 24 years of underwater episodes even without withdrawals, assuming a 60/40 asset allocation. As the equity allocation is increased to 80/20, the underwater episode durations increase to a total of 30 years. Withdrawal (spending by the endowment fund) increases the number and duration of the underwater episodes. For a 5% withdrawal rate and an 80/20 allocation there are five periods totaling 47 years (including the depression a total length of 23 years) of underwater episodes. As withdrawal rates rise from 3% to 5%, the length of episodes decreases as the equity weight goes from 60% to 80%. Endowments must be prepared for periods during which nominal values are underwater. Maintaining real value (adjusted by the consumer price index) is very difficult when prolonged market declines occur. When the asset allocation is 60/40 and the withdrawal rate is 5%, the real value of the portfolio never recovers to the beginning value in Table 4 Duration of Economic Episodes (in Years) Economic Episode = Period of 3 Years or More When Portfolio Value is Less Than Beginning Value Withdrawal Rate Asset Allocation 0% 3% 5% 7% 60/40 7,7,3,3,4 8,8,4,3,5,4,7 23,3,4,4,8,9 58,3,9 70/30 7,7,3,4,5 8,8,4,3,4,3,7 23,3,3,4,8,9 34,28,9 80/20 8,7,3,3,4,5 18,4,3,4,9 23,3,4,8,9 28,18,9 Real 60/40 5,8,6,4,13,7 7,18,3,27, X. Conclusions We find that endowment asset allocation, withdrawal or spending rates, expected returns, and variability in correlations are all important factors, but the withdrawal rate may be the most important. Withdrawal in excess of 4% or 5% can cause an endowment to be underwater from historical cost for many years. When inflation is (as it should be) considered, excess withdrawal causes endowments to be even more underwater. Colleges can reduce their budgets thereby reducing withdrawal rates, or colleges can 13

15 raise funds (not endowment) to meet spending needs. Other factors such as lower equity risk premiums increases the risk of failure to meet spending needs, especially under the real world assumption that 5% is withdrawn for spending by the educational institution. Higher equity risk premiums reduce failure risks, but if returns are lower the risk of failure in greater. Endowments cannot control the magnitude of the equity risk premium, but controlling the net withdrawal rate is controllable Lower asset allocation equity weights do not control failure (to maintain historical values) as much as hypothesized. When lower allocations to equity are used to reduce risk measured by variability, variability risk is lowered. The higher returns of higher equity (higher variability risk) asset allocations may reduce the risk of failure. The risk of failure is not controlled as well. The offsetting effects of higher return and lower variability must be managed by choosing an asset allocation that matches the needs and tolerances of endowments. We substantiate the maximum of 5% payout for foundations or educational institutions. Payouts in excess of 5% jeopardize the future abilities of such institutions to provide support at inflation adjusted levels. However, the key factor in controlling failure risk or the risk of failure seems to be control of spending or success in fund raising to offset excess spending from endowments. In a world of lower expected returns colleges must contain spending and raise new funds that can be used to offset excess spending needed to achieve the goals of the college. The effects of lower returns on all asset classes as is the current expectation, puts extra pressure on portfolios because spending 5% when average returns are lower than 8% reduces real values of portfolios. These results appear to hold for the historical period that includes several major periods of market stress. Treasury bonds, both long and intermediate-term, are less correlated with equities especially during stress periods. The extra returns from other fixed income or alternative assets must be balanced with the lower risk and greater liquidity provided by these safe Treasury bonds. Adding intermediate term government bonds reduces the risks during stress periods, so increasing the allocation to fixed income (government and corporate bonds) offers protection during periods of stress like the credit crisis of However the best protection is fiscal discipline. 14

16 References Arnott, Robert D. The Meaning of a Slender Risk Premium, Financial Analysts Journal, Vol. 87, No. 2, March-April, 2004, pp Arnott, Robert D. and Lovell, Monitoring and Rebalancing the Portfolio, in Managing Investment Portfolios A Dynamic Process, Second edition, Maginn, J.L. and D.L. Tuttle, ed., Warren, Gorham, and Lamont, Arshanapalli, B., T.D. Coggin, and W. Nelson, Is Fixed-Weight Asset Allocation Really Better? Journal of Portfolio Management, Spring 2001, pp Bengen, William P., Asset Allocation for a Lifetime, Journal of Financial Planning, August 1996, pp Bengen, William P. Conserving Client Portfolios During Retirement, Part III, Journal of FinancialPlanning,December 1997, pp Black, Fischer and Robert Litterman, Golbal Portfolio Optimization, Financial Analysts Journal, Vol. 48, No. 5, September-October 1992, pp Blake, D., B.N. Lehmann and A. Timmermann, Asset Allocation Dynamics and Pension Fund Performance, Journal of Business, Vol. 72, No. 4, October 1999, pp Common Fund Institute, Principles of Endowment Management, Cooley, Phillip L., Carl M. Hubbard and Daniel T. Walz, Retirement Savings: Choosing a Withdrawal Rate that Is Sustainable, AAII Journal, February 1998, pp Cornell, Arnott, and Moroz (February 2009) The Equity Premium Revisited, Retrieved from Fama, E. F. and K.R. French, The Equity Risk Premium, Journal of Finance, 57, July 2002, pp Fernandez, Pablo (September 22, 2009) The Equity Premium in 150 textbooks, Retrieved fromhttp://papers.ssrn.com/sol3/papers.cfm?abstract_id= Graham, John R., and Campbell R. Harvey, Expectations of Equity Risk Premia, Volatility and Asymmetry, Working Paper W58, Duke University, July 7,

17 Hechinger, John and Jennifer Levitz, (2009, February 11) Battered Nonprofits seek to tap Nest Eggs, Wall Street Journal, Retrieved from Ibbotson Associates, SBBI Classic Edition Yearbook, Chicago, Lauricella, Tom, Failure of a Fail-Safe Strategy Sends Investors Scrambling, Wall Street Journal, July 10, Neely, Walter P. and Bill Brister, Endowment Asset Allocation and Rebalancing: The Effects of Lower Expected Equity Risk Premiums, unpublished presented to Southern Finance Association, November Perold, A.F and W.F. Sharpe, Dynamic Strategies for Asset Allocation, Financial Analysts Journal, February 1988, pp Cynthia R. Rowland, UPMIFA, Three Years Later: What s a Prudent Director to Do?, Business Law Today,18, 6, Retrieved from Sharpe, W.F. Asset Allocation, in Managing Investment Portfolios A Dynamic Process, Second edition, Maginn, J.L. and D.L. Tuttle, ed., Warren, Gorham, and Lamont, Uniform Law Commission, UPMIFA, 2006, Retrieved from 16