Beyond The 4% Rule

by Abraham Okusanya

  Investors may be tempted to use investment strategies designed to produce relatively low volatility (and accept the typically lower returns that come with them) during the early part of their retirements in order to minimise the probability of an adverse sequence of returns. But this is not necessarily effective. A strategy that produces low volatility may nonetheless deliver a highly disadvantageous sequence of returns because SOR risk and volatility, although not unrelated, are not the same thing. A poor SOR can just as easily be the product of persistence in returns (or autocorrelation) as of volatility.

  A tale of two sisters

  Take two sisters: Mrs Unlucky, who started her retirement in 1906, and her much younger sister, Mrs Lucky, who started her retirement in 1924. They both invested in a portfolio comprising 50% UK equities and 50% UK bonds.

  Fig. 13: Mrs Lucky Vs. Mrs Unlucky

  Over the subsequent 30-year period, they both enjoyed good average return of 5.7%pa (nominal) and very similar levels of volatility (about 10.4%) in their portfolio. Their real returns were also very similar, at over 4%pa. But in terms of sustainable income from their portfolios? It’s night and day! For some strange reason, Mrs Lucky’s portfolio would have supported twice that of her sister over a period of 30 years.

  Fig. 14 shows what would have happened if they both took no income from their £100,000 portfolio over the subsequent 30-year period.

  Fig. 14: Year-end portfolio balance over 30 years with no withdrawal

  But suppose they both took an income of £5,000pa from their portfolios, without adjusting for inflation over a 30-year period? As Fig. 15 shows, Mrs Unlucky (1906) ran out of money while Mrs Lucky (1926) ended up with more than £100,000!

  How could that be? Same average returns. Same volatility. But very different income from their portfolios.

  It’s very simple: Mrs Unlucky got stitched up by sequence risk. The order of return was unfavourable.

  In the first decade, Mrs Unlucky got a return of -3.28%pa compared to Mrs Lucky’s 9.82%pa. And it didn’t matter that Mrs Unlucky’s portfolio gave a decent return of 7.04%pa and 10.48%pa in the second and third decades of her retirement. The damage had already been done.

  So what?

  Sequence risk is the primary investment risk in retirement. Not volatility.

  From a portfolio point of view, many asset managers are barking up the wrong tree. Volatility-managed solutions designed for retirement income most likely won’t work. In fact, taking volatility off the table may end up being dangerous. Add in the impact of high fees that these products typically charge, and they may amplify the dangers of sequence risk rather than reduce it.

  Fig. 15: Year-end portfolio balance over 30 years with £5,000pa withdrawal

  Fig. 16: Annual portfolio withdrawal of £5,000

  Controlling volatility doesn’t necessarily control sequence risk. This is because sequence risk is exacerbated by withdrawals from a portfolio, not by volatility.

  Natural yield: a totally bonkers retirement income strategy

  I’m often asked, how realistic is it to rely on natural yield to meet retirement income needs?

  The rationale is that by relying on the natural yield from their portfolio, retirees can avoid drawing on their capital or selling fund units, so they avoid the dangers of sequence risk.

  The natural yield approach contrasts with the total return approach. The total return approach seeks to draw income from capital growth and dividends in a sustainable way.

  Some asset managers promise stable natural yield to retirees. So, it’s no surprise that the consumer financial press is awash with articles on this approach. One by my friend Sam Brodbeck in the Telegraph10 sums up the approach rather nicely: ‘The idea is to ignore the fluctuating capital value of a portfolio and only take the natural yield. An original £100,000 investment might dip to £80,000 or rise to £120,000 in terms of value, but investors should resist the urge to touch the capital.’

  The trouble is, this approach is bonkers for several reasons.

  Dividend and bond yields fluctuate significantly over time. This means that a retiree’s income will change from year to year. Volatile income makes budgeting nearly impossible.

  Natural income is highly unlikely to meet the spending pattern of most retirees once it’s adjusted for inflation. Except for the very wealthy.

  Even if yield appears stable in percentage terms, the income received in monetary terms is in relation to the outstanding capital. This invariably fluctuates over the retirement period.

  Proponents of a natural yield retirement strategy have offered little empirical evidence to back their theory. They erroneously focus on the percentage yield of the FTSE 100 or FTSE All Share. Most retirees are more likely to have both bonds and shares in their portfolio.

  The test

  I set out to examine the natural yield approach to retirement income using empirical data. This time, I used the Barclays Equity Gilt Study (BEGS), which runs from 1900 to 2015. Unlike the DMS database which I used for my other research, BEGS decomposes equity and bond returns into capital growth and income yield.

  I created a portfolio consisting of 50% UK equities and 50% gilts, which is rebalanced annually.

  I examined the inflation-adjusted natural income on a £100,000 portfolio for retirement periods starting in 1900, 1905, 1910, 1915 …. to 2005 and 2010. I also included a retirement period starting in 2008, just to take my total number of scenarios to 20!

  Based on the dataset, I looked at a retirement period of 30 years. Retirees starting after 1985 haven’t completed the full 30-year period yet, so I’ve presented their results for the period covered so far. For instance, a retirement period starting in 1990 (Class ’90) has had 25 years so far and a retirement period starting in 2000 (Class ’00) just 15 years!

  The real natural income is the inflation-adjusted natural income from the portfolio each year. I work on the basis that no withdrawal is taken from the capital and the retiree relies entirely on the natural dividends and coupon on their portfolio.

  The result

  Fig. 17 opposite shows the real income yield from the portfolio over the subsequent 30-year period (or less for retirement dates starting after 1985).

  Fig. 17: Real (inflation-adjusted) natural income from £100,000 portfolio over subsequent 30 years from various start dates

  This chart shows that a retiree relying only on natural income experiences significant income fluctuations from year to year.

  For example, our Class of 1900 started their retirement with a natural income of £4,550. By the second year, their inflation-adjusted income fell to £3,897 and by the fifth year, it was £3,005. But their troubles had only just begun; by year 20, their real income had fallen to £1,024!

  I don’t know anyone who would find this level of income fluctuation acceptable.

  In Fig. 18, I show the first year’s natural income, the lowest, the mean and the highest real income over the entire retirement period. I also show the income volatility in percentage terms. This is the standard deviation from the mean income over each retirement period.

  So, a retiree living off natural yield may or may not start off with a decent income. But they should expect their income to go up and down each year like a yo-yo. And once you add in the effect of inflation, this yo-yo effect (income volatility) is simply unacceptable for most people.

  Chasing yield

  Of course, an advocate of natural yield will argue that a natural yield portfolio will specifically overweight high-yield assets such as equities, commercial property, real estate investment trusts (REITs) and high-yield bonds. A paper by Vanguard11 exposes the flaws in this thinking. I want to pick out these key drawbacks:

  Fig. 18: First year’s, lowest, mean and highest real natural income over the entire 30-year period for various start dates

  Fig. 19: Cumulative total returns during the global financial crisis (12 Oct.,2007 to Mar., 2009)

  a) High-yield asset classes such as commercial pr
operty/ REITs, equities and high-yield bonds tend to have large drawdowns, particularly during stressful market conditions. Fig. 19 is taken from the Vanguard paper referred to earlier and it shows the total returns of major asset classes, including high-yield ones during the financial crisis of 2008. As you can see, income-yielding asset classes experienced larger losses.

  b) Overweighting high-yield asset classes invariably increases concentration risk and reduces diversification in the portfolio.

  c) There’s some empirical evidence to suggest that high dividend stocks tend to outperform over the very long term. This is essentially known as ‘value premium.’ But this is more likely to be down to a low price-to-dividend ratio (or frankly low price-to-anything ratio). Fortunately, there are better ways to capture value premium than price-to-dividends. Using price-to-earning or price-to-book is a far more effective way to capture the value premium.

  d) The strategy relies heavily on the ability of a manager to select high-yield stocks and bonds. Good luck with that one.

  I stand by my assertion that a natural yield approach is a bonkers retirement income strategy for virtually all but very wealthy retirees, who have other sources of steady income to rely on.

  Cash is trash

  Another common approach used by financial planners to manage sequence risk in retirement portfolios is to hold a cash reserve. The strategy typically involves holding between one to three years’ worth of income in cash. This helps them to avoid having to sell down portfolio holdings to pay income during protracted market declines.

  The question is, does a cash reserve enhance portfolio longevity better than a fully invested portfolio that sells down holdings to pay income? And with varying practices among planners, what is the optimal amount of cash holding in drawdown? I’ve seen one, two, three and even five years’ worth of income held in cash reserves!

  Holding lots of cash in a portfolio can be a drag on performance in the long run. For instance, based on 5%pa withdrawal rate, holding two years’ of income equates to 10% of the portfolio (ignoring inflation and cash held to cover platform and advice fees), which could have been invested.

  US-based financial planner Harold Evensky and his partner Deena Katz are perhaps the most prominent architects of the CFR strategy (Evensky-Katz Cash Flow Reserve Strategy). They’ve used it with clients and written about it for many years.

  Fig. 20: Evensky- Katz Cash Flow Reserve Strategy

  But does CFR strategy really work?

  In a 2013 paper titled The Benefits of a Cash Reserve Strategy in Retirement Distribution Planning12, Evensky et al used Monte Carlo simulations to compare a one-year CFR strategy with a fully invested 60/40 portfolio with no cash reserve. In the CFR, income is paid from the cash reserve and replenished if it dips below two months of income. The 60/40 portfolio is rebalanced annually, and the monthly income taken directly from the portfolio.

  Based on a 4% withdrawal rate, they found that ‘the fully invested portfolio is slightly superior to the cash reserve approach, assuming that there are no transaction costs and taxes.’ But when you take into account the transaction costs of selling down the portfolio monthly to pay income, the cash reserve approach produces a better outcome – around a 5% improvement in the success rate.

  So, the cash reserve method doesn’t help reduce the effects of sequence risk if there’s no cost and tax drag for selling down the portfolio. Also, holding cash on most platforms results in negative return once you consider the platform fees. So, this diminishes the effectiveness of a cash reserve even further.

  This finding is consistent with similar but more extensive research by Walter Woerheide and David Nanigian13. This used 83 years of historical data to test the success rate of one, two, three and four years’ cash reserve vs. fully invested portfolios across several asset allocations and withdrawal rates.

  In the CFR portfolios, they assumed that withdrawals were taken from the portfolios when return was positive and from the cash reserve when the portfolios were down. The overwhelming conclusion is that the fully invested portfolios produced better outcomes than corresponding cash reserve portfolios in about 80% of scenarios.

  These studies appear to tell us that cash flow reserves don’t necessarily help to mitigate sequencing risk. Using cash produces sub-optimal outcomes when compared to a fully invested portfolio. The reason is clear – markets tend to be up more than down, so trying to completely avoid volatility or being overly conservative doesn’t compensate for the drag on returns. In any case, with a consistent rebalancing strategy, more of the withdrawals are taken from the asset class that’s performed the best, assuming the withdrawal is pro-rated based on the portfolio allocation.

  Sequence risk vs. stupidity risk

  A cash reserve may not help with mitigating sequence risk but it could be very effective in reducing stupidity risk. It’s hard to argue against the positive impact it has on managing the cognitive and behavioural biases that damage returns if retirees panic during market stress.

  The knowledge that they always have six to 12 months’ income in cash, which is not subject to the whims of the market, is an effective ‘framing’ and ‘mental accounting’ technique to help people sleep better at night during market declines. To quote Wall Street Journal columnist Jason Zweig, when markets fall, ‘an investor who has courage but lacks cash is as powerless as one who has cash but no courage.’

  Yet it’s important not to be overly conservative, as a lower allocation to equities invariably reduces a portfolio’s longevity. Keeping more than one year of income in cash does more harm than good, and the trade-off in lost return is too high.

  More importantly, this trade-off can be offset by taking the cash reserve from bond allocations, rather than the equity allocations. Suppose a 60/40 (equity/bond) portfolio is recommended for a client. If they keep 5% of the total portfolio in cash and invest £95,000 in a 60/40 portfolio, the overall allocation to equities is 57%. Instead, consider a 60/35/5% allocation to equities/bond/cash. Woerheide and Nanigian‘s research indicates that if the cash reserve is taken from the bond allocation, no real harm is done (although no significant advantages are achieved). In other words, the cost of obtaining the behavioural benefits of a cash flow reserve is negligible.

  It’s important to think carefully about how frequently the cash reserve is replenished. Starting out with one year’s income and replenishing the reserve every six months means you’ll always have at least six months’ worth of income in a cash reserve.

  Kenigsberg M, Mazumdar P and Feinschreiber S (2014): Return Sequence and Volatility: Their Impact on Sustainable Withdrawal Rates. The Journal of Retirement 2014; 2: 81.

  Sam Brodbeck (October, 2016) ‘Can I live off the natural yield of my portfolio?’ http://www.telegraph.co.uk/investing/funds/can-i-live-off-the-natural-yield-of-myportfolio/

  Schlanger T., Jaconetti C., Westaway P., Daga A., (2016): Total-return investing: An enduring solution for low yields. Vanguard Research

  Pfeiffer, Shaun, John Salter, and Harold Evensky. 2013. “The Benefits of a Cash Reserve Strategy in Retirement Distribution Planning.” Journal of Financial Planning 26 (9): 49–55.

  Woerheide, Walter, and David Nanigian. 2012. “Sustainable Withdrawal Rates: The Historical Evidence on Buffer Zone Strategies.” Journal of Financial Planning 25 (5): 46–52. Available at SSRN: https://ssrn.com/abstract=1969021 or http://dx.doi.org/10.2139/ssrn.1969021

  CHAPTER 4

  Safe withdrawal rate: how safe?

  The key framework for managing sequence risk from a drawdown portfolio originated from Bill Bengen. Bill was an engineer who later became a financial adviser. His seminal paper in 1994 transformed the conversation around retirement income planning. The paper has been peer-reviewed and referenced by both academics and practitioners.

  Sadly, the SWR framework has been misinterpreted and misapplied far too often. If I didn’t know better, I’d say Bill would be cringing if he read some of the nonsense that’s been written.r />
  Bengen’s approach was to examine how to sustain spending. He didn’t simply look at average rates of return, he examined actual historical sequences of market returns. He established a safe withdrawal rate (SWR) – a percentage of the initial balance, with the ongoing withdrawal amount adjusted for inflation. It was capable of surviving any 30-year sequence in history. It’s crucial to understand that the percentage of withdrawal only relates to the capital in the first year of retirement. So, for instance, a withdrawal rate of 4% from a £100,000 portfolio gives an income of £4,000 in the first year. This £4,000 a year then increases or decreases in line with inflation each year, regardless of the value of the outstanding portfolio.

  SWR is based on the most severe economic and market conditions in 100 years of market history. And it’s designed to survive these extreme conditions.

  Adapting Bengen’s approach to the UK, the SWR is 3.1%. This is based on someone who started their 30-year retirement in 1900. But other 30-year periods starting in 1901, 1902, 1903, 1936 and 1947 were equally bad.