PART TWO:

Evaluating Spending Policy Results: Withdrawals, Portfolio Value and Risk

We’ve set a spending rate and developed an asset allocation with an expected return and risk that seem reasonable. We believe this approach should allow our organization to continue providing financial support to our beneficiaries over many generations. But what if we face market conditions that don’t meet our expectations?

To evaluate this risk and to understand its monetary implications, we created a large number of potential outcomes, using return scenarios that are higher and lower than our expectations, creating “the good, the bad and the ugly.” Now it’s time to evaluate these results to confirm that we can really live with the downside outcomes that we might encounter.

In this second paper on spending policy, we review a set of 30-year reports that measure the expected withdrawals and ending portfolio value for our spending policy and its asset allocation strategy. This analysis also looks at the various types and levels of risk that the organization faces, and we customize this evaluation to align with the organization’s view of what constitutes “mission failure.” This is a true client-centered, goals-based, fiduciary approach to investing and risk management.

THE BASICS: RETURN AND RISK OF OUR ASSET ALLOCATION

Our asset allocation is a mix of 30% bonds, 50% equity and 20% alternatives. This strategy provides an expected compound return of 7.49% with volatility of 11.67% and a one-year expected mean return (“arithmetic average”) of 8.17 percent. We can picture this strategy in the context of the set of globally-diversified asset classes.

The inflation-adjusted (“real”) returns for our strategy’s one-period and long-term and returns are 5.52% and 4.87% respectively. This shows the “volatility drag” of 68 bps that separates the arithmetic and compound returns. The typical calculation of the long-term geometric return is to subtract half the return variance from the mean return. In this example, our variance is the square of the risk (11.67% x 11.67% or 1.36%) so that our nominal compound return is 8.17% minus half of 1.36% (0.68% or 68 bps.) Our calculation is 8.17% - 0.68% = 7.49 percent.

We can expect that in 4-out-of-6 years, the range of one-year returns (at +/-1 one standard deviation) will fall between -3.5% and 19.8%, a spread of over 23 points. Of course, about 1-in-6 returns will lie outside this range (lower than -3.5% or higher than 19.8 percent.)

SOME PRACTICAL CONSIDERATIONS

In our earlier paper, we noted that it’s a reasonable starting point to align our spending rate with the real (i.e., inflation-adjusted) return of the portfolio. But this is only a starting point; there are a few other factors to consider.

First, we need to allow for the negative effects of market risk, as reflected by the volatility of the portfolio’s return. Downside volatility causes the portfolio to make withdrawals in the face of market losses. This is a potentially significant threat to the goal of maintaining the portfolio’s value.

Second, we must remember that our monetary analysis is a sequence of one-year events. This means that we must test our spending rate using the portfolio’s one-year (mean) return, rather than its long-term return. The wide range of one-year returns we anticipate implies that we might consider lowering our spending rate. Clearly, we need to balance all these factors… but how? The answer is found in our examination of the monetary results from our many sample scenarios.

CREATING OUR SAMPLES OF MONETARY RESULTS

Our focus is on median values as we examine our results. This is a more conservative approach, as compared with simply taking an average of all 3,000 potential outcomes. We recognize that “upside” results tend to be very strong, but our focus as fiduciaries is primarily on preserving the portfolio value - and so we place greater emphasis on the lower half of all possible outcomes (rather than on the strength of the upside outcomes.) There is also an inescapable intuition around basing our expectations on “what is most likely to happen” as represented by the mid-point of outcomes. A comparison of expected vs median outcomes illustrates the effect of excluding the strength of the “upside” scenarios.

Our fiduciary approach drives our decision to focus exclusively on the lower half of the occurrences. This pushes more of the uncertainty of outcomes into the “delightful” category (outperforming the goals) and minimizing the “disappointing” category (mission failure.) Our example uses 5% spending with 5-year smoothing of market value over thirty years.

“Beneath the covers” of our spending rule is the methodology for smoothing the market value of the portfolio that is used to calculate each period’s spending amount. We use the average of the prior five years of portfolio market value, net of spending. This increases the stability of the amounts to be distributed, resulting in more reliable support for the beneficiaries. This also minimizes and delays any decrease in support to the beneficiaries when markets decline.

Here is an example of smoothed spending, calculated at the start of an organization’s withdrawals. The beginning portfolio was discounted by our assumed 2.5% inflation over the prior four years. We then took the average of these five years, reducing the market value used to calculate the withdrawal amount from $10 million to about $9.52 million. This process produces an initial “effective spending rate” of 4.76% relative to the portfolio’s current amount. (This is effectively a 25 bps reduction in the spending rate.) By moderating this spending amount, the unspent value remains in the portfolio, eventually creating a surplus value that can sustain spending during periods of market downturns. Over time, this surplus increases the regular withdrawals amounts, often to levels exceeding the withdrawals achieved with higher initial spending rates. This helps to protect the portfolio’s value so that it remains “fully funded” even after substantial market downturns.

INITIAL MONETARY RESULTS

The median results of our $10 million portfolio over thirty years anticipate spending out almost 150% of the portfolio’s starting value while retaining close to 98% of its value for the next generation. Our conservative approach focuses our attention only on the downside events that lie below the mid-point of all outcomes. While half of all outcomes exceed our median value, we don’t consider them in our decision process; these are a potential “upside surprise.” If we were to include the top half of possible outcomes, our results would be 10% higher in withdrawals and 30% higher in portfolio value. We may interpret those expected results as a potential “upside surprise” - but we will determine our spending rate using median results.

It is helpful to track the pattern of our results over time, with an eye toward any trends (rising, falling or stabilizing.) We focus on our primary fiduciary goal of preserving the portfolio’s real value. While our expected value reflects increasing appreciation, our median portfolio values appear to be stabilizing at levels that are very close to our starting capital.

Our graph of median spending indicates an initial increase followed by a stabilizing trend, with a relatively-smooth decline to a level that exceeds the initial spending amount. This confirms the expectation that spending modestly in the beginning tends to create a “sustaining surplus.” After about 20 years, the portfolio maintains its level of monetary support. This is the result of the interplay and balance between market returns, volatility, spending rate, portfolio return, and the smoothing methodology. The long-term relationship between these factors determines whether the spending policy is in line with the investment strategy. If so, both portfolio value and withdrawals should stabilize over time. These relatively small near-term changes in withdrawals confirm our expectation of withdrawals with much lower volatility than in the investment markets.

We also observe that while real portfolio value declined by about three percent, the annual spending amount increased by the same proportion. These were offsetting trends.

THE TREND OF RISK

We stated that risk means different things to different people. The fiduciaries of a “spending” portfolio like an endowment or a foundation see risk in terms of mission failure, which is defined as losing portfolio value, with a resulting decline in support to the next generation of beneficiaries. This raises two questions:

• How severe is a loss that is considered a failure?
• What likelihood of failure is tolerable?

To answer these questions, we examine the portfolio value relative to the loss severity over time. We must find the combination of severity and likelihood of loss that aligns with the client’s fiduciary standard. It is not enough to simply evaluate the median portfolio value at the end of thirty years; we must also examine the downside possibilities.

One may argue that considering a 10% loss as a sign of failure is unrealistic, since this is a high majority of the initial capital. Perhaps, but this is a client-specific consideration. We acknowledge that the likelihood of failure decreases as the tolerable loss severity goes up. In our example, we see risk leveling off earlier at higher risk tolerances, while risk continues increasing at lower loss tolerances.

One might conclude from these results that the spending rate should be lowered, or that a higher portfolio return is required. These decisions can be tested and evaluated to find the appropriate balance between these factors. This reflects the holistic approach to these decisions, all determined in the context of the mission.

EVALUATING A SET OF SPENDING RATES

Our next test examines the effects of higher and lower spending rates, using our asset allocation strategy and our 5-year smoothing methodology. We used 50 bps increments in moving the withdrawal rate higher and lower than our initial starting rate of 5% spending. This testing range of 4% to 6% spending generally encompasses rates used by endowments and both public and private foundations.

It’s easy to see that spending rates above 5% tend to be “liquidating” levels; this is generally true even with more aggressive asset strategies. Why? Our testing shows that the benefits from a higher return are more than offset by the effects of disproportionate increases in volatility, which bring the higher likelihood of monetizing a more severe downside response to markets. This greater level of short-term risk demands higher liquidity to avoid withdrawing from depleted assets. As a result, the portfolio tends to have the same level of capital at the end of the investment horizon when return is pushed beyond the limits of the spending policy. The critical lesson is this: “Not all points are efficient along the efficient frontier” when it comes to spending policy portfolios and goals-based investing.

Our testing shows that higher spending rates produce slightly higher cumulative withdrawals for the current beneficiaries, but at the expense of the next generation. When spending rates increased from 5% to 5.5% and then to 6% the cumulative withdrawals did not increase proportionately; these 10% and 20% relative increases in spending rate produced withdrawals that were only 3% to 4% higher. In response, we saw ending portfolio value decline by 26% to 30 percent. Clearly the future is sacrificed for the benefit of the current generation, a severe violation of the primary fiduciary responsibility. We also see that total value earned (the sum of withdrawals and ending value) declines as spending rates increase.

We experience the opposite set of outcomes as we cut the spending rate. The 10% decrease in spending (from 5% to 4.5%) produces only a 4% decrease in cumulative withdrawals, while the portfolio value increases by 18% (providing a “margin of safety” against future uncertainty of outcomes.) And a 20% cut in spending lowers cumulative withdrawals by only 8% while increasing expected ending portfolio value by almost 40 percent. What is the driver behind these results? Surplus!

One might conclude from these results that a spending rate of 4% may be too low, since an 8% cut in payments to beneficiaries is significant, while the expected surplus value delivered to the next generation is too large. In making this decision, it helps to incorporate the client’s true risk tolerance, using a target loss severity and examining its likelihood of occurrence across the range of spending rates. If the client’s willingness to sustain a 20% loss in portfolio value is limited to a 1-in-4 likelihood, then a 4% spending rate might be appropriate. This demonstrates the need for judgment in this multi-factor process of aligning asset allocation, return expectations, market risk and spending policy.

THE TREND OF MONETARY RESULTS

We stated that point estimates (for statistics) and single monetary values (for results) are important but incomplete measures of success. We also need to examine the trends that we may experience. Our final analysis examines the trends of withdrawals and portfolio value over time, across our set of spending rates.

Our expected real portfolio value (net-of-spending and adjusted for inflation) seems to establish an equilibrium value at our initial spending rate of 5 percent (the real return of the portfolio.) By testing the spending rate in 50 bps increments above and below our starting point, we established two trends: increasing portfolio value vs eroding it. The declining trend indicates a harmful imbalance between the factors we control: asset allocation, spending rate and smoothing of market value. Understanding the relationship between these decisions is key to establishing reasonable expectations around our three potential outcomes: maintaining, eroding it or increasing portfolio value. These results must match the client’s expectations and fiduciary goals with regard to intergenerational equity and the risk of mission failure.

We find several interesting results as we examine the trend of withdrawals. We see the expected range of initial withdrawal amounts that reflects the 50% difference in relative spending rates from the lowest to the highest rates. In all cases, the withdrawals increase during the first few years; this is the result of building a surplus in portfolio value driven by our smoothing methodology and its moderation of the initial spending amount. With the lowest pair of “surplus-generating” spending rates this trend of increasing portfolio value continues. Our base case of spending settles into an equilibrium, maintaining its real withdrawals indefinitely. The “liquidating” spending rate withdrawals hit an inflection point, and then begin declining continually in response to the continuing erosion of portfolio value. After about 20 years, withdrawals from the modest spending rates catch up with, and then exceed the amounts generated by the “liquidating” spending rates.

COMING NEXT

Our next article presents a case study in performance evaluation using this goal-based approach. We examine a real-life portfolio over several market cycles, as it continues to deliver

financial support by following this spending discipline. Critical “stress tests” of success focus on preserving both financial support and portfolio value throughout the substantial market declines of 2000-2002 and 2008, when many institutions were forced to make changes in their levels of support and in the number of programs that they could continue to maintain.

Written in partnership with Stephen Campisi