Publications and Q and A's


"In Search of the 'Best' Retirement Strategy", Journal of Financial Service Professionals, January 2011.
This paper examines eight retirement strategies proposed by researchers and pundits by simulating and evaluating the strategies according to a common set of criteria.  An approach is presented to help guide a retiree or planner in selecting the "best" strategy for the retiree.

"Creating Safe, Aggressive Retirement Income Profiles", Journal of Financial Planning, May 2010.
"Much research on retirement income focuses on real income profiles that are flat or increasing over time. This paper defines safe profiles for clients who plan to spend relatively more early in retirement and less later."

"Using Decision Rules to Create Retirement Withdrawal Profiles", Journal of Financial Planning, August 2007
"
Instead of applying one-size-fits-all retirement withdrawal rules first, this author proposes that retirees pick their own retirement profile and success rate for the profile, then apply the appropriate “decision” rules to maximize withdrawals for the selected profile."

"Decision Rules and Maximum Initial Withdrawal Rates", Journal of Financial Planning, March 2006.

Topics

Withdrawal Rates
Asset Allocation

Asset Returns and Expenses
Decision Rules
Retirement Profiles
Other Models
 

Withdrawal Rates

Q:
Your withdrawal rate is significantly higher than Mr. Bengen's. Can you elaborate?
A:  This is because Mr. Bengen did not use decision rules when doing his research.  Decision rules are designed to rescue poor performing portfolios.  This allows one to start with a higher initial withdrawal rate and plan to cut back if the financials require it. 


Q: You mention using the "Withdrawal Rule" in year's following a market downturn. Can you briefly explain that rule and an application/example?

A: The original Withdrawal Rule proposed by Guyton said that in years where your portfolio had a negative return, you would freeze your income level at the prior year’s amount.  You would never recover freeze but continue from that point going forward.  When we modeled that rule, it became clear that that rule was overly harsh.  If your portfolio had a negative return but your withdrawal rate was, say only 1%, it would seem silly to deprive yourself of a cost of living adjustment.  The rule was modified to take effect only if your portfolio had a negative return and your initial withdrawal rate was higher than your initial rate.

 

Q: We have dealt with 30 and 40 years, and seen that lowering from 40 to 30 years does not make any significant difference in raising the IWR. My question is, would there be even less of a difference if we tried to stretch this to 50 years?

A: One can try it easily enough using Retirement Quant.  Simply change the expected lifetime in the assumptions.


Q: What is the impact on the withdrawal rates if an advisor charges 1% of the assets?

A: It is the same as if the median return of all assets was reduced by 1%.  You can model any basis point change in returns using Retirement Quant’s Return Sensitivity function and see the effect on the success rate, withdrawals, etc.

 

Q:  Did you do the withdrawals in a lump sum once a year? When did you rebalance?

A: Withdrawals were made once at the beginning of a simulated year.  Rebalancing was done then also.


Q: When dealing with clients who have most of the assets in qualified plans, how do you handle the increasing RMD withdrawals late in retirement when they begin to exceed the IWR? Can this have an impact?

A: Tax implications and strategies were not modeled.  It is assumed that taxes are paid out of withdrawals and so the model deals with establishing a retirement income level.  Taxes certainly do have an impact that is complex and probably unique to each situation.

 

Asset Allocation

Q: At what point in this process does one select specific funds to achieve the stated allocations and how is it done?
A: The simplest and arguably the best approach to try to reproduce the results of the model is to buy index funds or ETFs that match the asset class.  For example, there are many mutual funds that track the various large and small cap, growth and value equity indices as well as REIT and international indices. 

Q: Are the test portfolios rebalanced periodically throughout the period?

 A: In the earlier paper, portfolio balancing is determined by the Portfolio Management Decision Rule.  In my subsequent paper, I modeled portfolios with rebalancing every year.  Retirement Quant allows you to choose either one of those strategies, a rule called the “Fastest Horse” strategy, or a rule which rebalances the stock/bond allocation each year as a function of your age (more bonds and less stocks as you get older).


Q: In looking at the asset allocation variable in determining retirement withdrawals, what are your thoughts on incorporating some exposure to alternative asset classes such as commodities, private equity, equipment leasing, etc.? Do you think incorporating these alternatives could increase the IWR or improve the POS?

A: Adding uncorrelated asset classes should improve the success rate and allow the initial withdrawal rate to increase.  One simple way to test this is to change the market assumptions in Retirement Quant to use uncorrelated asset returns.  If you know the historical mean and standard deviation of an asset class, you can override the value for an existing class and then model the returns.
 

Q: How do you handle large cash inflows during the plan? For instance an Inheritance or Property Sale during the plan?

A: These are not handled in the paper and should be simulated to see their effects.  Retirement Quant has the ability to simulate future future cash inflows.
 

Q: The withdrawal rate should decrease over time if your investments are the top ones so you limit/minimize the concerns of the Fall rule.  So, what funds will get us there?  Could this be that point in the model where one's planner assesses the funds and lists those that are say in the top 10 to then select from?
A: Yes, the better your portfolio performs, the better your chance of success.  Trying to beat the market, however, is a tough business.  There are a lot of very bright people out there working very long hours trying to do that.  And there is a reason that those bright people call it "placing a bet" when one picks something other than the market.  Those top 10 funds are yesterday's news.  I highly recommend the book "Fooled by Randomness."

Asset Returns and Expenses

Q:
Given the fact that nearly everyone agrees that investors get investor returns and not the theoretical investment returns, even with an advisor's help, don't you realistically have to reduce the withdrawal rate? If so, by how much?"
A: Use Retirement Quant to model what you think are “investor returns”, evaluate the risk and choose the appropriate strategy.  Again, it is very easy to see the effect of different returns, there is a dedicated function to just that, and select the level of risk you are comfortable with.


Q: What expenses did you assign to the various asset classes, and did you assign any management expenses?

A: No expenses were assigned.  They are assumed to be included in the return of the asset class.  There are two ways that one can explicitly add expenses to the model.  One way is to modify the returns for the asset classes and subtract some expense from the median returns.  The other is use the Return Sensitivity function described above.


Q: If asset management fees were not factored into the analysis, is it fair to say that the chosen withdrawal rate simply be reduced by that % each year?

A: No.  The relationship between withdrawal rates, market returns, and decision rules is not that simple and one should model the scenario to see the impact.  If the management fee is a percent of assets, then you could model the effect by reducing the mean returns by that percent.  If the management fee is a fixed dollar amount, then model that by increasing the retirement income need by that amount.


Q: Does the modeling methodology make any tax assumptions, or is it assumed that the client has to pay taxes out of the withdrawals? Thank you.

A: Tax implications and strategies were not modeled.  It is assumed that taxes are paid out of withdrawals and so the model deals with establishing a retirement income level.  Taxes certainly do have an impact that is probably unique to each situation.


Q: Do you believe that the PE rate inflation that has occurred over the last 80 years inflates equity returns and means we should count on lower rates in the future?

A: It is probably unwise to “count on” anything.  The admonishment “past returns are not a guarantee of future returns” is not just there for legal protection.  You might enjoy reading “Fooled by Randomness” and “The Black Swan” by Nassim Taleb.
 

Decision Rules

Q: What if I use other values for FALL, RAISE, EXCEEDS, and CUT?
A: You will get a different profile.  The paper has a table which shows how a change in each of these variables will affect a retirement profile.  In addition the graphs in the paper show you the effect on a basic retirement portfolio.  Making this perhaps more complicated is that real-life situations, such as a pension or social security, also impact the profile.  You can use the Retirement Quant to simulate individual profiles for yourself.  Retirement Quant implements the Capital Preservation and Prosperity Rules slightly differently from the paper.  Fall and Exceeds are implemented as absolute withdrawal percentages rather than as percent differences from the initial withdrawal rate.  For example, if the initial withdrawal rate was 5% and the value of Fall is 20%, then a raise would be given if the current withdrawal rate hits 4% (twenty percent less than the original five percent).  In Retirement Quant, you will specify the absolute number, i.e. 4%, rather than the 20%.  This change was made to make the rule simpler and perhaps easier to use.

 

Retirement Profiles

Q: Are there other retirement strategies that would give a Progressive or Aggressive Withdrawal Profile?
A: Absolutely.  Use the Retirement Quant program to see what others have suggested and play with it and see what you can create on your own.
 


Q: It seems that your WD rates assume the portfolio value maintains full value (inflated) throughout lifetime, leaving a large asset to heirs.  Wouldn't many clients want to use it up entirely over the 40 year period?
A: Every person is different.  The projected legacy (final estate) is one of the outputs from Retirement Quant.  It is easy to see the effect of different withdrawal rates on the legacy using the Retirement Income function.  Not only that but you can also see the effect of different asset allocations, retirement strategies, market returns, and retirement ages on the legacy.

 

Other Models

Q: Did you ever do these exact same scenarios only changing the distribution period to 30 years rather than 40 in order to compare more closely to the Bengen and Trinity studies? 

A: Simulations were run under conditions similar to Bengen’s and the Trinity study and produced results similar to what they obtained. 

 In the News
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

© 2016, All Rights Reserved

 Retirement Quant