Friday, December 14, 2012
Jonathan Guyton on Combining Dynamic Withdrawals and Dynamic Asset Allocation
Two brief notes before starting:
-My December column at Advisor Perspectives is now available. It is on the role of income guarantee riders to support retirement income. This finishes my three-part series on income guarantee riders.
-My second column as a MarketWatch RetireMentor is out, called "Depleting Assets May be a Good Strategy." Subsequently, a very interesting discussion about the column took place at the Bogleheads Forum.
At the FPA Experience 2012 conference, I enjoyed an opportunity to hear Jonathan Guyton speak on the topic, “Volatility and Volatility-in-Advice: Applying Behavioral and Analytical Stress Tests to Safe Withdrawal Policies and Practices.” His talk was quite detailed, and I would like to focus on part of it. This part introduces a new avenue for research. Let me back up a minute to explain what that is.
The classic retirement income strategy put under the test for the 4% rule is to withdraw constant inflation-adjusted amounts for as long as possible until wealth is depleted. It isn’t a particularly realistic strategy, though it probably never was meant to be either. Real people will make adjustments throughout retirement. After big market changes, there are two key ways in which one could respond:
-Change the spending amount (up when markets do well, down when they drop)
-Change the asset allocation (have different asset allocations in mind for different market valuation levels)
Both of these types of strategies have been tested quite a bit by researchers. Jonathan Guyton, in particular, made his name by developing a series of decision rules which would support a higher initial withdrawal rate with the understanding that one will make cutbacks to spending when certain events take place.
What Jonathan Guyton is the first to do in this presentation (as far as I know) is to combine these two dynamic possibilities: adjust both spending and asset allocation at the same time. Past researchers have treated each separately.
He finds some intriguing results when applying this to the case of someone retiring at the start of 2000.
The decision rules for dynamic withdrawal amounts which he uses include:
-if the current withdrawal rate (in a given year) is within 20% of its initial withdrawal rate, then increase the prior year withdrawal amount by inflation
-if the prior year’s portfolio return is negative, do not make any inflation adjustment
-if the current withdrawal rate exceeds the initial withdrawal rate by 20%, reduce the spending amount by 10% (Capital Preservation Rule)
-if the current withdrawal rate falls below the initial withdrawal rate by 20%, increase the spending amount by 10% (Prosperity Rule)
And for the dynamic asset allocation, the total stock allocation will be either 50%, 65%, or 80 depending on the market valuation level. He didn’t explain the specific rule for this, but he noted that asset allocation changes happened in 2000, 2009, and 2011. I’d be willing to guess that the allocation is 50% for 2000-2008, 65% for 2009-2010, and 50% since 2011. This is compared to a static stock allocation case of 60%.
Consider a retiree on January 1, 2000, who has saved $1.2 million. This is what Jonathan found:
January 1, 2000
Initial Portfolio Value $1.2 million $1.2 million
Initial Withdrawal Rate 4% 5%
Initial Withdrawal Amount $48,000 $60,000
January 1, 2012
Portfolio Value $1.041 million $1.078 million
Current Withdrawal Rate 6.2% 5.8%
Current Withdrawal Amount $64,400 $62,300
This is interesting. In the dynamic case, withdrawals could not keep up with inflation, but it would be hard to argue that the dynamic person is worse off. Their initial spending started 25% higher, and after 12 years it is roughly the same as the static case. The dynamic case has been able to enjoy more total spending thus far in their retirement. As well, the current value of remaining assets is still slightly larger for the dynamic case as well.
I’d say that this is definitely an area deserving of more research focus.