The Sustainable Draw Rates Study

February 8, 2019
Written by Matthew White, CFA®, Investment Analyst

SUSTAINABLE DRAW RATES STUDY: PART ONE

Part one of The Sustainable Draw Rates Study focuses on our Initial Hypothesis and relevant background information.

Summary

  • Adding stocks to portfolios generally improves the probability that an investor will be able to continue a real spending rate over several decades without running out of money.
  • A portfolio composed entirely of bonds is unlikely to sustain real spending rates above 2.5% for investors over longer time periods (30 years +).
  • The addition of high-quality, low-correlation non-traditional investments to portfolios improves the likelihood of maintaining a real spending rate for investors.
  • There is a diminishing value to adding stocks to portfolios beyond ~70% when an investor has access to high-quality, low-correlation non-traditional diversifying investments.

Background

From time to time we evaluate outside portfolios composed entirely of stocks. Being curious investors ourselves we naturally asked: “What could be the rationale for building such concentrated portfolios?”  We know that stocks are generally the highest returning asset class over extended periods of time, so on the surface, it makes some intuitive sense that they might be the most appropriate tool to accomplish investor objectives. Advances in low-cost investment access vehicles, such as Vanguard and iShares ETF, further beg the question of whether a portfolio invested entirely in low-cost stock investments is the best approach to accomplishing long-term investor goals; rendering any other portfolio option a high-fee exercise in futility and/or over de-worse-ification.

Before we can set out on our investigation, we must first define what investor goals may be. This is no miniscule task, as the range of individual investor goals is as broad as the population of the world is vast.  For the purposes of this study we restrict ourselves to an investor who has already accumulated or received assets and is seeking to design a portfolio that maximizes the amount they can spend without running out of money. We also assume that this investor will increase the amount they spend each year by the amount of inflation so that the purchasing power of distributions is not diminished over time.

We think this scenario is applicable to a broad array of interested investors. In the case of private families, the ability to maintain a lifestyle standard over time, whether over one lifetime or over multiple generations, is of great concern. In the case of many organizations, such as foundations or endowments, the directive of providing a sustainable and growing spending rule to beneficial parties is an irrevocable mandate which must be met by current stewards of the organization.

So, we set out to prove that an approach incorporating anything other than stocks would diminish the likelihood investors would achieve this goal.

Designing Our Study

We set up our study by establishing a portfolio of low-cost stock investments as well as six comparison portfolios to compare results with.

The All Stock portfolio was constructed roughly in line with global market capitalizations. The All Bond portfolio consisted of investment grade US bonds. The mix for each of the four “Diversified” portfolios is based on a Mean-Variance-Optimization exercise to obtain optimally allocated portfolios for various degrees of risk.

Our test portfolios allow us to compare a range of traditional and non-traditional mixes to ascertain two answers related to our hypothesis:

  • Does an “All Stock” portfolio provide better results than portfolios that include bonds, including an “All Bond” portfolio?
  • Does an “All Stock” portfolio provide better results than portfolios with incremental mixes of bonds and diversifying investments?

We then define our specific investment goal. Given that investors have a wide range of spending requirements, an even wider array of tax circumstances, and potentially even greater differences in attitudes towards risk, we simplify our target goal as such:

“Spend 5% of the initial portfolio value in year 1 and increase it with inflation each year, while limiting the potential of running out of money”

We select 5% as a spending rule knowing that some investors spend less out of their portfolio, and some spend more. However, we know from other studies that a 5% spending rule is a potentially unsustainable rate into perpetuity; thus it should provide us a dispersion of probabilities of success across our sample portfolios as we advance forward in time. We make the logical leap that the portfolio which is most likely to sustain a 5% spending rate is also most likely to sustain a 3% spending rate, as well as a 6% spending rate, and so on and so forth. We assume inflation to be 2.2% based on current consensus estimates. For the lay investor, when we refer to a real spending rate, we are saying it adjusts upward each year by this inflation figure.

For purposes of the simulation, we assigned expected returns (and associated volatility) to each portfolio based on the asset classes it holds, as well as an annual deduction from the portfolios for “cost to access” the respective asset classes.

Expected returns and volatility for each asset class are based on the most recent forward-looking consensus estimates from our 2018 Capital Market Assumption study. The cost to access deductions set the expectation that each investment vehicle will exactly match its market asset class return, before fees. While imperfect, it is the most even way to assume managers will add no value, nor extract any value on a gross basis. We do note that this assumption gives the benefit of the doubt to the traditional asset classes since it assumes they may be accessed at a lower cost than non-traditional investments.

We assume that stocks and bonds may be accessed with low-cost mutual fund or ETF shares for less than 10 basis points. In the case of non-traditional asset classes, we used all-in operating expense ratios from funds which provide diversified exposure to their respective classes, which typically ranged from 100 to 200 basis points. For additional detail on assumed returns, volatility, and costs, please see the notes at the end of this series or contact us for additional information.

With our portfolio range of annual returns defined and our required spending cashflows setup, we conducted a 2,000 simulation Monte-Carlo analysis over 40 years to give us a deep distribution to test our hypothesis. We advanced forward 10, 20, 30, and 40 years and inspected each of the 2,000 paths that a portfolio might have taken based on our return and volatility assumptions, with the constant draw of 5%, increasing by inflation. If we inspect a portfolio path at one of our time intervals and it still has at least $1, we will say that it was successful up to that point.

We then aggregate each portfolio’s number of successful paths divided by 2,000 to come up with a “probability of success” for each decade. We will say a portfolio’s probability of success at each point in time is its likelihood of providing an investor with 5% of the initial portfolio value, every year, increased for inflation, and still have at least $1 left.

Coming in part two: we examine results by decade and asset mix.