Is this a safe way to get higher returns in retirement?
Conventional wisdom tells retirees to seek safety with a conservative asset allocation in retirement. Is there a better way?
In my last article I provided two charts that showed the trade-offs of investing in retirement. The first chart showed the expected future returns from our investment management team for different allocations between shares, bonds and cash. Unsurprisingly a portfolio of 100% shares has the highest expected return at 9.41% annually over the next 30 years. A 30% alocation to shares, 60% allocation to bonds and 10% allocation to cash has an expected return of 6.16%.
The next chart showed the safe withdrawal rate from the same allocations. The highest safe withdrawal rate is from a 30% allocation to shares. This means that retirees get the most bang from their retirement bucks with less shares in their portfolio.
How is this possible? If returns are higher from an all-share portfolio logic dictates that is the portfolio that should support the highest withdrawals. The reason for this seeming contradiction is something called sequencing risk. If a retiree has the misfortune to retire right as markets start to tank while money is being withdrawn from a portfolio it causes a big problem.
During a market meltdown a more volatile portfolio with more shares falls more. Once withdrawals come out this smaller portfolio has less assets to appreciate when markets recover. For more detail on sequencing risk please see my first article.
Is volatility the real issue?
Ultimately it is not the short-term volatility of the portfolio that matters. It is the fact that a retiree is forced to sell after the market falls. The adage of not selling low is pounded into our heads as investors. The problem is that a retiree doesn’t have a choice because the money is needed to pay for life.
If the issue is not the fact that markets will inevitably fall at times but having to sell the solution is just not to sell. That is the problem that needs to be attacked. If a retiree can solve this problem the volatility matters less. It is important to be clear about the differences between the assumptions used in the modelling for safe withdrawal rates and real life.
The models assume that a diversified portfolio is sold off equally. Some of the shares are sold, some of the bonds are sold and some of the cash is withdrawn. In real life some retirees face this situation. For example, if you are in a pre-mixed option in an industry super fund this scenario may reflect reality.
However, retirees in a more flexible industry option or with a SMSF can pick and choose which specific assets are sold. This allows a sensible retiree to sell off bonds and cash if the share market has tanked. Shares could be sold if the market continues to climb.
This is the idea behind the bucket approach for asset allocation for retirees. Assets are grouped into buckets of cash, less volatile investments like bonds and large dividend paying companies, and growth assets like shares. This gives retirees more flexibility to replenish the first two buckets when market conditions warrant.
I’ve gone a bit further with my mother’s asset allocation in retirement with a 2-bucket portfolio. One in cash and one in shares from large dividend paying companies that traditionally have had lower volatility. The cash bucket gets replenished with dividends and opportunistic sales of shares when the share market is performing well. The cash is used to pay for life expenses.
This has worked out well for her and she has passed the danger zone of a share plunge early in retirement. If she did have the misfortune of retiring into a bear market it would have been more challenging. But given that I took a conservative approach of holding 5 years of living expenses in cash she would have had ample time for a market recovery.
How this would work for future retirees
We can explore how this approach would fit into the withdrawal rate models. As previously stated, the returns from a diversified portfolio of shares are forecast at 9.41% over the next 30 years. Cash returns over the same period are projected at 3.31%. That gives a portfolio of 80% shares and 20% cash a projected return of 8.19%.
That return is meaningfully higher than the return from a 30% allocation to shares. The question is how this approach will respond to bear markets. That is the mechanism to judge how a withdrawal rate is impacted. Remember that the withdrawal rate is supposed to be safe under a variety of situations. The market plunge is just the situation we want to guard against.
I’ve explored the following scenario. First, I’ve assumed that if no shares are sold the 20% allocation can support 4 years of living expenses. There are several adjustments to withdrawals I could make but I’ve decided on a 4.5% year one withdrawal rate and 7% annual increases in the dollar amount of withdrawals. I think this is rather conservative.
The next step is to look at historic bear markets. Much of the commentary we read is about the length of bear markets. But this is a bit misleading when thinking about retirement outcomes and sequencing risk. What is more useful when considering a bucket portfolio or cash as a buffer is to include how long the recovery to break even takes.
Since the end of World War II the US has experienced 13 bear markets. Since the US market is a good proxy for how global markets behave, I think this is a reasonable substitute for a globally diversified share portfolio. The average length of time from a market peak to a market bottom is approximately 12 months. To get back to peak – the breakeven point – took an average of approximately 21 months.
That gives us an approximate historic breakeven time of 33 months. That is less than 3 years which means a 4 year supply of cash provides a more than adequate buffer. But remember that while that is the average length we need to worry about the extremes. We can look at the 5 longest bear markets since World War II.
Source: A Wealth of Common Sense
As you can see three of the longest bear markets exceeded the 4 years of cash in my scenario. This includes 6.58 years for the dot com sell-off, 5.58 years for the 70s bear market and 4.5 years for the GFC bear market. While the bear market was over and the recovery starting at the end of 4 years the hypothetical retiree would still need to sell shares to fund life prior to break even.
At the end of 4 years the S&P 500 was still below breakeven even by the following percentages:
- January 1973 – (18.19%)
- March 2000 – (22.65%)
- October 2007 – (21.59%)
Each of the return scenarios does not include the impact of dividends. Dividends are an important component of total returns but there is a reason that I excluded them. In the bucket approach to retirement asset allocation the cash bucket is replenished with proceeds from the longer-term buckets. One opportunity is using dividends.
I’ve modelled out the dividend yield on the portfolio that would provide 2.5 additonal years of living expenses. This covers the time of the dot com breakeven. A 4.31% yield would do the trick. That assumes that dividends stayed constant during the 4-year period of replenishing cash. It assumes no additional cash collected in the 2.5 years of the dot com sell-off. This extra time could be used to build up extra cash.
There are a couple ways to think about this. A 4.31% yield on a diversified portfolio is difficult in the US where yields are much lower. With Australia’s higher yields it is feasible. Especially when franking credits are taken into accouont. But during a brutal bear market dividends may be cut.
Final thoughts
This isn’t a bullet proof approach. Unfortunately, the perfect approach doesn’t exist. We can’t account completely for the bad luck of retiring into one of the worse bear markets since World War II. But remember that the 4% rule and modelling around withdrawal rates is supposed to protect a retiree in 90% of modelled scenarios. Even the ‘safe’ withdrawal rate isn’t completely safe.
Ultimately in real life there were other levers retirees could pull in this situation. The 7% annual growth in cash taken out of a portfolio in my scenario could be cut. Small parts of the share portfolio could be sold off after the market started to recover. At the very least this approach protects a retiree from the largest impacts of sequencing risk in the most extreme scenarios.
More importantly it protects retirees from the other risk in retirement. Longevity risk. This refers to living a very long time and running out of money before death. The higher expected returns from a higher allocation to shares will take care of that. Depending upon personal circumstances and how risk adverse you are you can increase or decrease the years of spending needs held in cash.
I will be going through this approach and how dividends can be used in retirement in the last two sessions of my Retirement Bootcamp. You can sign up here or send any questions to mark.lamonica1@morningstar.com