To predict how well endowments can weather economic volatility, investment managers can conduct modeling of the risk spectrum. What they learn can allow them to consider inherent interconnections and trade-offs that affect results.
By Lucie Lapovsky
Risk is present in nearly every decision we make. Consider a traveler who is anxious about flying because of safety concerns. She drives to her destination instead, believing it to be safer. Missing from this traveler's assessment of risk is data on the probabilities of various outcomes from her choices. She has actually chosen the riskier vehicle, increasing her chance of being in a serious accident by more than 50 times. In the absence of data, we often believe we are avoiding risk when in fact we are incurring unseen and potentially greater risk. Institution endowment committees must be aware of a similar trap that exists in investment decision making.
Achieving endowment goals requires careful assessment of available data to understand the risks associated with various asset allocation and spending policies. One must look no further than the recent market downturn to underscore the importance of acknowledging the full spectrum of potential risk in any given strategy. The equity market collapse that began in late 2008 put many institutions in a quandary with regard to their policies.
On investment policy, responses of institutional endowment committees have ranged from doing nothing (i.e., maintaining the precollapse asset allocation), to putting everything in equities and equity-like investments—since in theory, the only way to go is up—to putting everything in cash to preserve the endowment corpus and not risk further losses. On spending, institutional responses have included sticking with current policy (resulting in a reduced endowment draw because of a smaller corpus), to increasing the spending from the endowment to compensate for other revenue reductions in the budget, to eliminating spending from the endowment altogether for a year or two to allow the corpus to recover more quickly. While few endowment management committees responded at these extremes, many have adjusted their asset allocations and spending policies to move in one of these directions.
How can institutions know if they are making the best decisions? Like the traveler described above, institutions may unknowingly assume more risk than necessary if they fail to consider the data available to them. By using model simulations, institutions can forecast the possible impact of their asset allocation and spending policies. To illustrate the use of simulation modeling, this article models two asset allocation and two spending policies to provide a backdrop for discussing the risks involved in different policy decisions.
Thanks to widely available computer software, modeling a variety of scenarios is now fairly straightforward. Use of these techniques requires investment committees to consider all relevant variables and to engage in valuable discussions about risk and the trade-offs involved, thereby affording a framework for disciplined decision making. For other stakeholders, this approach culminates in rich documentation that can clarify the rationale behind specific endowment policies and enhance the transparency of the investment committee's decision making.
One limitation of modeling is that it is based on a set of economic assumptions. In addition to making certain assumptions about inflation, future gift flows, and administrative expenses, rates of return and the probability of achieving them must be estimated for each asset class. For this reason, it is advisable for most institutions to use investment consultants to help make appropriate inputs to such models.
Regardless of the models used, three primary goals must be considered in assessing endowment risk:
- Supporting the students of today and tomorrow (intergenerational equity).
- Maintaining the predictability of support from the endowment for the operating budget.
- Maintaining liquidity of the endowment so that money is available to meet spending goals.
Each goal has one or more associated risks that give rise to the chance of not achieving the goal, and each adverse result diminishes the chance of attaining the goal. For example, for the goal of intergenerational equity, an institution can underspend on the current generation of students to ensure that the endowment corpus is preserved long term. This could actually weaken the institution in such a way that although the endowment will theoretically be preserved, the institution may essentially wither in the short term.
It is also important to acknowledge the interrelated nature of these risks. Higher spending today can favor the current generation over future generations. Reaching for higher returns can call for investing in illiquid asset classes, as well as the transfer of too much market risk to the spending regime. More aggressive asset allocations provide the best chance of maintaining the value of the endowment corpus in perpetuity, while an allocation that favors cash and fixed income may have little volatility but will not maintain its value in real (inflation-adjusted) terms. Because these risks are not only interrelated, but to some degree are also competing, they cannot adequately be considered in isolation.
For a baseline example, let's consider an institution with a $350 million endowment, an asset allocation of 70 percent in equities, and an expected return of 8.25 percent. The baseline spending formula assumes 4.75 percent of rolling, 24-month average asset values. In addition, let's assume that inflation will average 3.5 percent over the long term, administrative expenses will be 0.5 percent of endowment assets, and new gifts of $1 million each year will increase annually by inflation.
By using Monte Carlo simulation (a computational technique used to measure uncertainty), we can forecast the impact of these investment and spending policies, considering both the expected (median) outcome as well as the range of possible outcomes. Figure 1 is a Monte Carlo simulation of what will happen to the endowment over each of the next 20 years in nominal dollars. Each bar summarizes the results of the thousands of trials for which their values have been ranked. The white line shows the median outcome, the value at which half the trials fall below and half are above. The dark blue section spans the 25th to 75th percentile, thus comprising 50 percent of the trials. The top of the green bar represents the 95th percentile, so 5 percent of the trials had higher results. Similarly, 5 percent of the trials fell below the bottom of the tan bar (the 5th percentile). Using this model, overall we can anticipate having our result land within the boundaries of the bar with about 90 percent confidence, leaving a 1 in 10 chance of experiencing an outlier.
In this example the graph shows that, with the set of assumptions defined above, this endowment is expected to grow over time (approximately double over 20 years for the median case) and carries substantial volatility. For instance, in 2029 the endowment range from 5th to 95th percentile is $269 million to $1.545 billion with a median value of $639 million. In terms of risk, there is a chance, albeit very low, that in 20 years the endowment value will be less than the original $350 million. On the other hand, there is also a small chance that the endowment, in nominal dollars, will have increased more than 400 percent.
Impact on Endowment Goals
To evaluate this set of baseline policies on each of the three primary goals, we can review appropriate metrics from the simulation looking at real (inflation-adjusted) returns.
Intergenerational equity. While Figure 2 has the same data as Figure 1, the data are adjusted for inflation. This graph shows the probability of maintaining the purchasing power of the endowment over this period of time. The tan and blue bars meet at the current fund value of $350 million, the level required to keep the purchasing power constant. The percent on top of each bar indicates the chance of maintaining the purchasing power of the endowment after 20 years. In this example, 46 percent of the thousands of trials ended up with a real market value equal to or greater than $350 million at the 20-year mark, while 54 percent ended up with a real market value of less than $350 million.
This asset allocation and spending policy poses the risk of not maintaining the endowment's purchasing power at least 50 percent of the time, something that may prompt a committee to discuss how to increase the certainty of maintaining intergenerational equity. For example, a committee might choose to lower the spending rate and/or change the asset allocation to one that has a higher expected rate of return.
Operating budget support. With regard to spending goals, we can review projected spending levels to see how the endowment can support the operating budget. To more fully reveal the trend, Figure 3 shows a deterministic forecast, a single path representing expected/median outcomes. The baseline spending policy (4.75 percent of the endowment corpus of the previous 24 months) produces a precipitous decline in spending for the next year and produces lower spending for the next several years.
This picture represents a situation currently being experienced by many institutions, where lower spending has triggered actual program cuts, layoffs, and tuition hikes as institutions deal with the reality of significant operating budget reductions. In this case, the spending formula inherits volatility from the investment policy. Due to the severe market losses of the fourth quarter in 2008, nominal spending is slated to decrease by 15 percent to 20 percent from 2009 to 2010 and then slowly rebound. Given the assumptions used in the model, the endowment spend will not return to the 2009 level until 2015.
The concern of some college and university leaders is that if their institutions do not increase the funds available from their endowments to support their operating budgets during these financially stressful times, they will be severely weakened, perhaps to the point that their future survival may be at stake. Other institution leaders may make a different assessment and decide to reduce their spending rate because of the risk that there will be insufficient endowment corpus to support future generations.
Liquidity. An institution can calculate a liquidity ratio comparing cash available to cash needed. Figure 4 shows that even the least likely probability comfortably exceeds the threshold of 1.0, which indicates that the institution will have the cash it needs to support its spending policy. Although interpretations may vary, the odds of a liquidity crisis seem quite low in this example. Several large endowments that were heavily invested in illiquid investment classes did face liquidity issues in 2009, but this is a risk that had rarely if ever been of concern before the recent market drop.
Quantitative Policy Setting
To this point, we've considered the baseline case and current investment and spending policies only to gain insight about the trajectory of key metrics and the volatility of those metrics. Next we will explore how this quantitative approach is invaluable in the context of policy setting. For example, let's imagine that the same investment committee considers new spending and investment policies to mitigate the current results. To make it tangible, we will introduce three discrete scenarios into our model and evaluate their effectiveness compared to the baseline (current) policies:
- New investment policy (New IP): 100 percent equity allocation resulting in an expected return of 9.0 percent.
- New spending policy (New SP): use of hybrid policy with 70 percent weight on prior-year spending and 30 percent weight on current formula, as well as a 94 percent floor (relative to prior year's nominal spending amount).
- New investment policy and new spending policy (New IP + New SP), as indicated above.
Figure 5 shows a comparison of the probability of maintaining the real value of assets in the endowment over 10 years and 20 years by using different combinations of the asset allocation and the spending policies described above. When looking ahead to 2029, the blue scenario (new investment policy and current spending policy) has the highest chance of maintaining intergenerational equity, at 53 percent. The dark green scenario (new investment policy and new spending policy) has about the same chance as the baseline policy (47 percent versus 46 percent, respectively), but with much greater volatility. And the tan scenario (baseline investment policy and the new hybrid spending policy) has only a 39 percent chance of maintaining the real value of the endowment's assets. For a clear example of why committee members must consider the range of outcomes possible for any set of policies, note that the worst-case forecast for the dark green scenario (new investment policy and new spending policy) approaches the zero line. This represents some meaningful chance (about 5 percent) of losing the entire fund. To the extent that the endowment committee had not explicitly considered this possibility, this process reveals a lurking risk that could prompt a revision in stated investment goals and objectives. Because many investment committees view the risk of loss as much more serious than the potential of greater gains, they are often willing to forgo the upside of gains for protection against downside loss.
As Figure 6 shows, scenarios using the hybrid spending approach (tan and dark green) have higher spending levels during the first five-year period than the traditional spending policy, which takes a fixed percentage of the endowment averaged over the previous 24 months. The higher spending in the hybrid policy (though still less in real dollars than the spending in 2009 in the baseline spending policy) provides greater support for the operating budget from the endowment and affords an institution time to adjust more gradually to the real decline in the endowment. While this mitigates some of the institutional risk that accompanies the need to make budget reductions swiftly, it does result in a lowered probability of maintaining the real value of the endowment for the long term.
Risk Comfort Zone
In summary, the higher spending provided by the hybrid spending policy in the short term results in a slightly lower endowment corpus in the long term. The new investment policy provides a higher probability of growth in the corpus over time, yet also has greater volatility and more downside risk. The combination of the new spending and investment policies may appear to be the best choice (as they suggest higher short-term spending and carry a similar chance of intergenerational equity), until the committee considers the trade-off that these new policies carry a 5 percent probability of losing virtually the entire fund.
While this analysis provides no clear answer for the best course of action for any institution, these scenarios do provide leaders with a range of choices to consider based on the best available data and assumptions rather than relying on the average outcome only, which is where many investment committees tend to focus. Risk is inevitable, but by undergoing these kinds of simulations, an endowment management committee can better determine which risks it is more comfortable accepting—a good first step in policy setting and decision making.
LUCIE LAPOVSKY, New York City, is a consultant in higher education finance and a former college president.