The answer lies in the industry’s misguided focus on volatility as the primary measure of risk. Higher allocation to equities is associated with higher volatility. In theory, this should matter less to the retiree, if the higher equity allocation gives them a greater chance of meeting their income objectives. But, in practice, investors contend with volatility on a daily basis. If the portfolio volatility is higher than the retiree is willing or able to accept, then there’s a greater chance they might cave in and abandon the investment strategy.
The problem with high equity allocation in retirement isn’t really an investment problem, it’s an investor problem. Working with a financial adviser can be hugely beneficial here. An adviser can help reconcile the conflict between a client’s willingness to tolerate volatility and the chances of meeting their income objective. A good adviser also provides an invaluable handholding service, particularly in extreme market conditions.
Ultimately, if retirees aren’t prepared to accept at least 50% equity allocation, they’ll need to accept a very low withdrawal rate or seriously consider a safety-first approach. In other words, annuitising some of their retirement funds to secure their income.
Small-cap and value equities
Is there a role for small-cap and value equities in a retirement portfolio?
Bengen explored the role of small-cap and value equities in research published in 199743, 200644, and then 201645.
His earlier articles noted that introducing small-cap and value equities into the portfolio mix significantly increased the SWR. But Bengen was cautious about the higher volatility associated with small-cap stocks and whether the outperformance of small-cap equities when compared large-cap equities (so called, small-cap premium) will persist.
When Bengen revisited the subject again in 201646, he was far more bullish about the improvement that small-cap and value equities added to the safe withdrawal rate.
Bengen’s findings were corroborated by Tomlinson47 (2014) who considered the impact of small-cap and value stocks on withdrawal rates.
But does this hold up for UK retirees? I took a look at the impact of UK small-cap and UK value equities on withdrawal rate.
I constructed three portfolios:
50/50 UK small/bond portfolio
50/50 UK value /bond portfolio
25/25/50 UK small/UK value/bond portfolio
I compared the withdrawal rates over every 30-year period rolling since 1900 from these portfolios, to those of the baseline 50/50 UK equities/bond portfolio. You can see the results in the table below.
Fig. 58 shows that including value and/or small-cap in a retirement portfolio improved the sustainable withdrawal rate. Replacing all the equity allocation with value or small-cap improved the withdrawal rate for most of the time period. The median improvement is 0.4% for small-cap and 0.9% for value.
Fig. 58: Sustainable withdrawal rate with small-cap and value equities
Fig. 59 below provides a summary of the withdrawal rates for each of the four portfolios, including the baseline portfolio.
Fig. 59: Percentile rate of sustainable withdrawal rate with small-cap and value equities
It shows that the gross withdrawal rate in the worst case is 3.1% for our baseline allocation, compared to 3.2% for small-cap, 3.8% for value and 3.5% for an equal blend of value and small.
Replacing large-cap equities with small-cap and value improved the sustainable withdrawal rate in virtually all rolling 30-year periods.
Fancy blending value and small-cap equities? No problem, you’re in luck here too! Equally blending value and small-cap equities gives a withdrawal rate of 3.5%, an improvement of over 10% even in the worst-case scenario. Remarkably, we’ve achieved this uplift in withdrawal rate without increasing the overall equity allocation within the portfolio.
The positive results notwithstanding, I must sound a note of caution. A reason value and small-cap equities improved the withdrawal rate is simply because they deliver higher returns than the overall equity market.
Fig. 60 below summarises the annualised return, standard deviation, maximum gain and loss between 1900-2016.
Fig. 60: Annualised return, standard deviation, maximum gain and loss for large, small and value equities (1900-2015)
This dataset shows us that value and small equities do outperform, but with slightly higher volatility. They do seem to deliver higher risk-adjusted return though.
I’ll leave the debate about exactly why value and small premium exists to financial academics. But there’s extensive empirical research to show that value and small-cap premium does exist.
Dimson, Marsh and Staunton48 addressed the question by noting that, ‘The key question is whether the size premium will continue in the future. This is unanswerable since we cannot travel forward in time. However, we can at least travel farther back and look at virgin data for earlier decades, before 1955. As a result of the research for this book, we now have reliable UK stock returns data for the first half of the twentieth century, but unfortunately only for the largest one hundred stocks. This may nevertheless be enough to provide a pointer. Given that the size effect tends to operate across the entire size spectrum, if there were a size effect, we would expect to find evidence even among the biggest stocks.’
They looked at the difference between equally weighted and capitalisation-weighted return of the top 100 stocks and concluded that this, ‘provides some tentative evidence on the existence of a size effect over the first half of the twentieth century.’
Dimson et al also attempted to answer the question of whether value premium existed pre-1956 and concluded that it did.
They noted that, ‘Over the long term, the historical record of value investing has been positive in the United Kingdom as well as the United States. We now know that value stocks did better than growth stocks in the earlier as well as the later parts of the twentieth century. And the value-growth premium appears to be relatively robust to alternative definitions of value.’
So, the key point is, including value and small-cap equities increases the SWR. The exact gain on withdrawal rate will depend on the intensity of the value and small-cap blend.
To go global or not? This is the question
One common question is around the impact of global diversification on SWR. We’ve known for some time that most investors exhibit a strong degree of home bias and allocate a higher proportion of their portfolio to their home market. But, the benefit of global diversification is generally accepted in the investment community.
So, what’s the impact of global asset allocation on SWR?
Pfau 201449 conducted perhaps the most extensive study on the impact of global diversification on SWR. He calculated the local currency returns on stocks and bonds in 20 different nations for the same global portfolio. The portfolio consisted of the 20 countries in the dataset. This allowed for 20 different perspectives on the role of international diversification. Previous studies on withdrawals rates with international diversification have only looked within the context of US-based investors. Pfau’s studies looked at SWR from the perspective of retirees in 20 different countries, investing globally.
Pfau concluded that global diversification helps more often than not – although it doesn’t provide a complete panacea for market risk in retirement.
Fig. 61: Sustainable withdrawal rate with UK-centric vs. global asset allocation
My research shows that a global asset allocation (ie 50/50 global equity/global bond split) results in a slightly lower sustainable withdrawal in 58% of historical periods when we compare it to a UK asset allocation. So, in 42% of historical periods between 1900 and 2016, global asset allocation results in higher SWR.
A closer look at the results reveals some particularly interesting patterns. Fig. 62 below shows the historical worst case, 10th, 50th and 90th percentile withdrawal rates for UK and global portfolios.
Fig. 62: Sustainable withdrawal rate for global and UK-centric portfolios at various percentile ranking
s
In the worst-case scenarios, SWR for a global allocation is 0.3% lower than that of a UK-centric portfolio. This is because the SWR in these very bad scenarios is primarily due to the aggressive inflation during World War One. The average annualised inflation rate between 1914 and 1920 is 15.3%.
Because the main cause of a low SWR during this period was aggressive inflation, a global asset allocation did nothing to improve the SWR! Global equities returns were actually heavily weighted down by World War One, and fared worse than UK equities.
It’s interesting that the median SWR is 0.4% higher for a global portfolio than a UK-centric portfolio. This suggests that in mild market conditions, a global portfolio produces a slightly better outcome. But, in extreme conditions – either very poor or very good sequence of returns – UK portfolios tend to support higher withdrawal rates.
Portfolios consisting of an equal split between global equities and UK bonds seem to fare slightly better than a global equities/bond portfolio. This is probably because UK bonds are a better hedge for UK inflation than global bonds!
Nonetheless, there’s a strong case for global allocation in retirement portfolios. The annualised return on UK equity between 1900 and 2016 is 11.2%, with a volatility (standard deviation) of 21.2%. This compares to annualised return of 10.6% for global equity and a volatility of 16.26%. The lowest annual return on UK equity is -51.62%, compared to -31.4% for global equity.
Fig. 63: Performance of equities & bonds in £ (1900 -2016)
The UK equity portfolio may give a higher return, but it’s also more volatile. For a retiree, a slightly lower SWR but with less volatility is probably preferable. That’s what they get with the global allocation. They may sacrifice some upside and residual wealth, but they’ll probably sleep more easily!
Commodities and alternatives
What about other asset classes?
Many people have questioned whether including other asset classes like commercial property and commodities improves withdrawal rates. We can’t make an assertion one way or the other, because historical records for these asset classes don’t go back far enough.
Cassaday (2006)50 indicated that adding other asset classes, such as REITs, commodities and international exposures, potentially has a positive impact on sustainable withdrawal rates. However, this research was limited in its scope; the authors only used US historical data since 1972 and assumed 3% static inflation. So, the period covered in the research excluded some of the more severe periods like the early 1900s and 1936. It’s not clear whether these asset classes would add much value under such stressful market conditions.
My take on this is that the likely benefit of adding these asset classes to a retirement portfolio are theoretical and at best, modest. These asset classes should only be used sparingly, if at all, (less than 10% of the portfolio allocation) and at very low cost.
Impact of alpha on SWR
One common question about the SWR framework is what’s the impact of superior investment return, also known as alpha, on sustainable withdrawals?
The logic goes that sustainable withdrawal rates are calculated using market returns (ie beta), so a retiree should be able to improve SWR by achieving higher risk-adjusted returns (ie alpha).
Alpha has the opposite effect that fees and charges have on sustainable withdrawal rate. A 1% improvement in risk-adjusted return over and above an index-based portfolio, results in a 0.5% improvement in withdrawal rate. Bengen (2006) framed this as the impact of the ‘super investor’ who generates portfolio alpha.
In practice, alpha is rare. If it does exist, it’s a very shy animal. It’s not like investment costs, which we know in advance and where we can see the impact. For all the energy spent chasing alpha, its relative impact on a successful retirement is minimal, compared with the impact of having a robust withdrawal strategy in place.
Fig. 64: Relative contribution of alpha, beta and cashflow (withdrawal rate) for a retiree
Research by Blanchett (2013)51 considered the relative impact of alpha, beta and cashflow (ie withdrawal) on a retiree’s overall outcome. The research notes that, ‘while Alpha and Beta are important, a 1% change in the initial withdrawal rate (Cash Flow) has a greater impact than a 1% increase in return, either Beta or Alpha. Therefore, Cash Flows are the most important determinant of retirement success!’
So, regardless of a retiree’s wealth, market return (beta) and the withdrawal strategy (cashflow) explains over 90% of the success!
There are two ways to approach this from a SWR point of view.
Ignore the impact of alpha when you calculate the initial sustainable withdrawal rate. You can increase withdrawal if alpha is delivered within the first few years of retirement.
Take a higher initial withdrawal rate based on estimated alpha. Reduce the withdrawal rate if no alpha is delivered in the first few years of retirement.
It’s best to ignore the impact of alpha when you calculate initial withdrawal rates. If you use rules-based spending methods, like Guyton’s guardrail or Kitces’ ratcheting, withdrawal will naturally be adjusted upwards if alpha is delivered in the first five years of retirement.
What low bond yield means for sustainable withdrawal rates
There have been concerns – voiced by several leading retirement income experts – that the era of low interest rates and low bond yields makes the SWR framework unlikely to hold up. Their argument is based on evidence that current bond yields are a strong predictor of future bond returns. Bond yields are currently low, so this suggests low returns in the next 10 years. And since the first decade of retirement largely predicts the sustainable withdrawal rate over the entire 30-year retirement, this implies a lower sustainable withdrawal rate.
We should take these warnings seriously. But it’s important to remember that the SWR framework is based on expectations of low return in the first place.
While current bond yield is a strong predictor of future nominal bond return, it’s a poor predictor of real (inflation-adjusted) bond return. Using data between 1900 and 2016, the coefficient of determination (R2) between current bond yield and real bond return over the subsequent 10 years is 25%! This means that there is a lot about future real bond return that is not predicted by current bond yield.
We often hear that the so-called ‘bond bubble’ is about to burst. Some argue that this makes the sustainable withdrawal rate framework unlikely to hold up in the future. But the real question is, is a downturn likely to be as bad as some of the worst bond returns in history?
Fig. 65: Some of the worst historical real annual return on bonds
The reality is that large losses on bonds are actually a fairly common occurrence. The SWR framework has survived periods of multiple negative double-digit bond returns. That’s what it’s been designed to do.
According to Professor Elroy Dimson, UK bond investors lost half their wealth in real terms in the inflationary period from 1972 to 1974! In the period between 1914 and 1920, UK bonds lost over 60% in real terms over seven consecutive years. But the SWR framework would have held its own during this period.
This is one important advantage of using extensive historical data to stress-test a withdrawal strategy. We can stress-test some of the most severe return environments and calibrate our plan accordingly.
What high equity valuation means for sustainable withdrawal rates
Another concern about the SWR framework is the likely impact of market valuation. Some of the concerns are valid, but many are misguided.
If we consider high equity valuations, they suggest lower-than-average expected return. But this is not necessarily the case.
The data available on UK equity valuation is somewhat limited, going back to 1926 for PE ratio and 1935 for Shiller PE10 (also known as the CAPE ratio). The data shows clearly that there is an inverse relationship between the PE ratio in a given year and the annualised real return on equities over the subsequent 10 years.
However, we should be extremely cau
tious if we use current equity valuations to predict future equity returns, in the medium to long term. Using UK data between 1926 and 2016, the R2 between current PE ratio and subsequent 10 years’ real return on UK equities is 20%. For Shiller PE10, the R2 is 22.5%. In other words, the predictability of future return, using current PE or CAPE ratio, is lower than the flip of a coin.
Nonetheless, given current high equity valuation, future returns are likely to be lower than the historical average. Thank goodness, the SWR framework isn’t based on historical average. It advocates a strategy that survives some of the most severe market conditions in history.
This highlights the need for flexible withdrawal strategies that will adjust withdrawals in the event of poor market conditions. Add to this a global asset allocation approach, rather than one that’s heavily dependent on a single country.
Bengen, William P. “Asset Allocation for a Lifetime”. Journal of Financial Planning, August 1996.
Bengen, William P. (“Conserving Client Portfolios During Retirement, Part IV”. Journal of Financial Planning, May 2001.
Bengen, William P. (2006) “Baking a Withdrawal Plan ‘Layer Cake’ for Your Retirement Clients”. Journal of Financial Planning, August 2006.
Bengen, William P, (2016) Small-Cap Withdrawal Magic. Financial Advisor Magazine http://www.fa-mag.com/news/small-cap-withdrawal-magic-28553.html
Tomlinson, J (2014) Do Small Cap-Value Stocks add Value in Retirement Portfolios? Advisor Perspectives. 4/8/14 by https://www.advisorperspectives.com/articles/2014/04/08/do-small-cap-value-stocks-add-value-in-retirement-portfolios
Beyond The 4% Rule Page 10