Follow-Up to JP Morgan Post

Several individuals took me to task for criticizing JP Morgan's conclusion that "greater lifetime income through... pensions and/or lifetime annuities allows individuals to increase both their withdrawal rates and equity allocations."  While this may appear to be a "logical" conclusion, particularly for investment allocations, the math just doesn't support this conclusion as it applies to withdrawals rates.  Retirees who desire reasonably constant spendable income in retirement, should decrease, not increase, withdrawal rates from accumulated assets as the amount of their fixed immediate life annuity/pension income increases, all things being equal.

As an example, Let's go to the "Excluding Social Security 2.0" spreadsheet on this site.  If we enter $1,000,000 in accumulated savings, $0 immediate life annuity, 5% annual investment return, 3% per annum desired annual increases, 30 year payout period and $0 bequest, we get an initial withdrawal rate of 4.34%.  If we assume 3% inflation, the inflation-adjusted run out tab shows that annual withdrawals are expected to remain constant over the expected 30-year payout period.

However, if we input a fixed immediate payment of $20,000 per year, the initial annual withdrawal rate drops from 4.34% to 3.75% to keep total annual spendable income (withdrawals from accumulated savings plus annual annuity payment) constant in real dollar terms over the expected 30-year payout period.

Finally, if instead of $20,000 per year fixed annuity payment, we input $40,000, the initial annual withdrawal rate drops from 3.75% to 3.15%.

JP Morgan's Dynamic Withdrawal Strategy

JP Morgan has recently released its research on dynamic retirement income withdrawal strategies entitled, "Breaking the 4%Rule". 

As someone who has frequently ranted on this website against the 4% Rule and encouraged the use of a dynamic retirement income withdrawal strategy, I recommend this paper to readers of this blog.  In many ways, the expressed goals of the JP Morgan withdrawal strategy are similar to the goals of the withdrawal strategy suggested in this website.  The JP Morgan paper also combines its withdrawal strategy with an investment allocation strategy, which I do not address in this website as I have no investment expertise.

In addition to attempting to carefully balance lifestyle risk and longevity risk, JP Morgan also attempts to "maximize how much utility value investors receive from their withdrawals."  This utility value maximization is also not something that I address in my recommended withdrawal strategy.

Some concerns I have with the JP Morgan Strategy:

The withdrawal schedule is significantly more aggressive than withdrawal rates recommended in this website.  For example, JP Morgan specifies an initial 5.9% withdrawal rate for a 65 year old with $1,000,000 in accumulated savings and $50,000 in "lifetime income".  Based on the assumptions recommended in thiswebsite and zero bequest, I get a withdrawal rate of 3.45% using the Excluding Social Security 2.0 spreadsheet if I assume $30,000 of the $50,000 of lifetime income is in the form of a fixed immediate annuity (with the remaining $20,000 payable from Social Security).  The main reasons the JP Morgan withdrawal rate is so much higher is that their model assumes higher future investment returns, lower future inflation, a shorter payout period and does not anticipate using accumulated assets to provide for future inflation adjustments to fixed payment "lifetime income."  If comparable assumptions are used for both models, I would anticipate results to be very similar for the initial year's withdrawal.

Additionally, the JP Morgan strategy does not appear to have relatively constant inflation adjusted retirement income as a goal.  Therefore, all things being equal, their withdrawal strategy would be expected to be more volatile from year to year than the approach recommended in this website when measured in inflation adjusted dollars.

Because the JP Morgan paper was critical of the performance of the 4% Rule in volatile markets, "especially when a portfolio loses significant value during the early years of retirement" I decided to calculate spending budgets and remaining assets under the JP Morgan strategy and the Steiner Actuarial Approach (using recommended assumptions and smoothing methodology) for a hypothetical retiree where asset returns are somewhat unfavorable.  The two graphs below  compare retirement spending budgets (withdrawals from accumulated savings + Social Security + fixed pension) and remaining assets under the JP Morgan Dynamic Strategy with the Steiner Actuarial Approach for someone retiring at age 65 with $1,000,000 in assets, $20,000 in annual Social Security and $30,000 in fixed pension/life annuity income (for a total of $50K) of "lifetime income". 

I assumed about a 0% average annual rate of return for this hypothetical retiree's first five years of retirement with the following randomly chosen rates of investment return:  year 1: -15%,  year 2: 2%, year 3: 5%, year 4: -5% and year 5: 15%.  I also assumed 3% inflation each year.  I used the withdrawal rate table included in the JP Morgan article to determine withdrawal rates and I interpolated between relevant wealth and age factors.  I ignored the fact that inflation increases in the retirees Social Security benefit would increase the retirees "lifetime income" and thus perhaps slightly increase the retiree's withdrawal rate under the JP Morgan approach.  I also assumed that the JP Morgan withdrawal rate tables (and the Steiner recommended assumptions) would remain unchanged for the entire 5 year period.  For each approach, I assumed that the annual budgeted amount determined under the relevant approach would be spent during the year.

Comparison of Budget Amounts In Inflation-Adjusted Dollars

Comparison of Remaining Asset Amounts

The graphs show that based on this assumed investment experience,  the JP Morgan strategy produces a spending budget that is somewhat more volatile (when measured in inflation adjusted dollars) than the Steiner Actuarial Approach.  Because it is more aggressive than the approach in this website (based on recommended assumptions), it produces higher spending budgets each year and therefore lower remaining assets at the end of the five year period.  Is the JP Morgan strategy better than the Steiner Actuarial Approach?  I don't believe its use of Monte Carlo simulations or utility value maximization necessarily make the JPMorgan strategy superior.  If comparable assumptions are used, results under the two methods can be comparable, and the smoothing algorithm in the Steiner Actuarial Approach results in more real dollar stability in the retiree's spending budget from year to year.   The Steiner approach is also readily available on this website.  

The Actuarial Approach and Withdrawal Policy Statements--Its in There!

In his February 26 post, Michael Kitces encourages the use of a written Withdrawal Policy Statement to ensure that retirees have a plan for dealing with market declines.

The Actuarial Approach described in this website automatically adjusts spending not only for market declines (or spending more than the budget) but also for favorable experience (or spending less than the budget).  So like the slogan in the old Prego commercial goes--"Its in There!"

Anticipating "Lumpy" Expenditure Needs

As indicated in previous posts, I believe it is not unreasonable to manage risks in retirement by diversifying sources of retirement income.  This article discusses research that explains why individuals are "more likely to select an annuity option when a 'partial' option is offered instead of an 'all or nothing' option."  I especially liked the comment from one of the researchers who said, “If you anticipate lumpy expenditure needs in retirement (e.g., out-of-pocket medical expenses), you want some liquid wealth to cover those expenses."

Comparison of Four Withdrawal Strategies Based on Recent Experience

In my previous post of January 16, I stated my belief that each of the three systematic withdrawal options examined in the Stanford/SoA study was inferior to the actuarial approach advocated in this website.  While I was critical of the three approaches for not even attempting to focus on total retirement income by coordinating with annuity income that the retiree may currently have or expect to have in the future, I think that each of the three approaches have shortcomings even if the retiree has no other sources of income.  In this post, I will illustrate those shortcomings with an example that uses historical investment and inflation experience from 1998 to the present. 

Let's assume that Rachel retired on her 65 birthday on January 1,1998.  At that point, she had $500,000 in accumulated savings in addition to her Social Security benefit.  Since she was single with no children, she had no desire to leave money to heirs.  She wanted to maximize her income in retirement, particularly in her younger retirement years (when she wanted to travel more).  On the other hand, she did not want to outlive her savings.  She decided to invest the assets not budgeted for spending each year 25% in large cap equities, 25% in mid-cap equities, 25% in mid-term bonds, and 25% in short-term bonds (which she rebalanced at the end of each year).  Unfortunately, Rachel died in a car accident on January 1, 2014.  She had just turned 81. 

The graph shows withdrawals under the three different strategies discussed in the previous post compared with withdrawals under the actuarial approach advocated in this website.  All amounts are shown in 1998 dollars.  In using the actuarial approach, Rachel looked at annuity purchase rates in 1998 and decided to determine her first year's withdrawal assuming 7% investment return, 4% inflation and death at age 95.  In 2004, when interest rates had decreased somewhat, she changed the assumptions to 6% investment return, 4% inflation and in 2009 she changed to the now recommended assumptions of 5% investment return, 3% inflation.  She also used the recommended smoothing algorithm. 

None of the four strategies would have been successful in meeting Rachel's objective to die with only a small amount of assets remaining.  She expected to live well past age 81 and her investments did much better than she assumed (on average).  Following the actuarial strategy, she would have died with $626,635 remaining.  But this approach was better than the other three approaches in meeting Rachel's objective, as she would have $786,436 remaining under the IRS Required Minimum Distribution approach, $804,358 under the Constant 4% approach and $827,504 under the 4% Rule.

And while the 4% Rule produces a ruler-flat inflation adjusted withdrawal pattern, it failed to maximize Rachel's desire to maximize spending.  The other two approaches also failed to maximize spending and their withdrawal patterns were much less stable from year to year than under the actuarial approach.  Based on experience from 1998 to 2014, the clear winner of the four approaches in terms of meeting Rachel's objectives is the actuarial approach.

Systematic Withdrawal Strategies Examined in Recent Stanford/SoA Study Leave Much to be Desired

Last September, the Stanford Center on Longevity, in collaboration with the Society of Actuaries Committee on Post-Retirement Needs and Risks, released "The Next Evolution in Defined Contribution Plan Design".

The principal author of this work was Steve Vernon, who is now a Consulting Research Scholar at the Stanford Center on Longevity in addition to his many other activities, including blogging on retirement issues for CBS MoneyWatch, authoring books on retirement and running his business, Rest of Life Communications.  I have mentioned Steve and the good work he is doing many times in my blog.

The stated primary goal of this paper is "to help retirement plan sponsors, fiduciaries and managers make informed decisions about implementing income solutions [sometimes referred to in the paper as "Retirement Income Generators] that will improve the financial security of their plan participants."  Stated somewhat differently, the paper encourages defined contribution plan sponsors to take steps to make annuity and systematic withdrawal options available to their plan participants.  This is a well written paper that makes some excellent points and suggestions.  While I think that the arguments set forth for including annuity options in DC plans are somewhat more compelling than including specific systematic withdrawal strategies, the only real bone I have to pick with the paper is the choice of systematic withdrawal strategies it choses to discuss and examine.

The paper looks at three systematic withdrawal options (and three annuity options).  As indicated in Section 10, the main criteria for selection of these specific options appears to be that they are "readily available to retirement plan sponsors in today's marketplace."  In my opinion, each of the three selected systematic withdrawal options is inferior to the actuarial approach I advocate in this website.  Readers of my blog know that I have railed against the shortcomings of the 4% rule, which is the first alternative examined, so in order to keep my blood pressure down, I won't go into them again here.  The second alternative (referred to as the constant 4% strategy) is probably worse than the first.  It is similar to the strategy of spending interest on your accumulated assets each year if you expect to earn 4% per annum.  This strategy does not coordinate with any annuity income you might currently have or expect in the future, it does not consider the retiree's desire to have relatively constant inflation-adjusted income in retirement and you should expect to leave a pile of money to your heirs upon death as the annual income produced by this approach is very conservatively determined.  The third approach (which is referred to as the Life Expectancy Based Percentage Strategy [IRS Required Minimum Distribution] has a little more appeal than the first two approaches (and is practically recommended in the paper), but it does not coordinate with annuity income you might currently have or expect to have in the future, it does not consider the retiree's desire to have relatively constant inflation adjusted income, and while not as conservative as the constant 4% strategy, it still produces a lower expected pattern of withdrawals and higher probability of having significant amounts of assets remaining at death.

The paper does point out (as I have many times in this blog) that retirees looking to manage risks in retirement probably should consider combining annuity products with systematic withdrawal strategies, which is why I was somewhat disappointed to see comparisons in this paper (for the most part) between the six individual retirement income generators rather than between various combinations of retirement income generators.  I was also disappointed that my actuarial approach (with recommended assumptions and 10% corridor smoothing algorithm) was not one of the systematic withdrawal options examined in this paper.  After all, my approach is also readily available to retirement plan sponsors (Heck, it's free!), and was touted as having some "advantages" and "nifty features"  in Mr. Vernon's 2012 book, "Money for Life." 

Maximizing "Expected Utility" to Develop An Optimal Decumulation Strategy

Mark J. Warshawsky has formed a new company, ReLIAS llc, which is looking to partner with financial and insurance organizations interested in providing decumulation strategies and products to retired individuals. Mark advocates combining a systematic withdrawal approach from a portfolio of diversified assets with a laddered series of relatively small and regular purchases of single premium life annuities.  His general approach anticipates that the portfolio of diversified assets will eventually be replaced over time by the purchased single premium life annuities.  Mark has developed an algorithm which maximizes a mathematical function (the "Expected Utility) which considers (among other things) the preferences and goals of the retired household for higher income in retirement and accumulation of greater wealth, as well as the retired household's concern for risk.

You can read more about Mark's approach and his thoughts about optimal decumulation strategies in his new website.

http://reliasllc.com/

For those who wish to "drill-down" more into his research, I recommend that you look at his recent (January, 2014) Power Point Presentation in the section which describes his approach.

I worked with Mark when he was the Director of Retirement Research at Towers Watson.  He is one bright guy.  We wish Mark well in his new endeavor and look forward to more educational material on his website on the subject of optimal decumulation strategies.

End of Year Reminder--Time to Determine Spending Budget for Next Year

It's that time of the year for many of us retirees to determine our spending budget for next year.  You may also wish to take this opportunity to revisit your investment strategy.  I will illustrate how easy this process is with an example retiree, Richard.

Richard retired last year at this time at age 65.  At that time, he used about 20% of his accumulated savings to buy an immediate life annuity that pays him $15,000 per year.  At the beginning of 2013, he had $800,000 left after his annuity purchase.  He inputted the assumptions recommended in our October 11, 2013 post (5% interest, 3% inflation, 30 years expected payout period (95-65) and $10,000 as the desired amount of assets at death) into the spreadsheet in this website, to determine a total spendable amount (excluding Social Security) for 2013 of $45,179 ($30,179 from accumulated savings and $15,000 from the annuity).  He deposited $30,179 in his non-interest bearing spending account and decided to invest half of the remaining assets ($769,821) in equities and the other half in a variety of fixed income investments.  During 2013, Richard spent exactly the amount in his spendable account plus the $15,000 from the annuity. 

Easy Steps to Determine Richard's Spending Budget for 2014

The first step in the process is gather asset data as of the end of 2013.  Richard's equity investments yielded almost 29% during 2013 and his fixed income investments yielded about 1%, so his end-of-year assets are $884,909 (compared with expected end-of-year assets from the previous year's calculation of $808,312, or an asset gain for 2013 of $76,597). 

The second step in the process is to determine a preliminary spending value for 2014 by inputting new amounts into the spreadsheet on this website.  If the same assumptions and amounts are used as last year except using $884,909 for accumulated savings and 29 years for expected payout period, Richard's preliminary 2014 spendable amount is $49,947 ($34,947 + $15,000).

The third step in the process is to apply the smoothing algorithm discussed in our October 11, 2013 post to the preliminary spending value.  Richard determines that the Consumer Price Index has increased by about 1.3% during 2013.  Therefore, he determines his 2014 spendable amount as last year's total spendable amount ($45,179) increased by 1.3% ($45,766), but not less than 90% of the preliminary 2014 total spendable amount of $49,947 (.9 X $49,947 = $44,952).  Since the corridor value is lower than last year's value increased with inflation for the year, it does not apply and Richard's total spendable amount for 2014 is $45,766 ($30,766 from accumulated savings and $15,000 from the annuity).

Richard plans to transfer $30,766 of his accumulated savings in his spending account and rebalance the remainder ($854,143) so that he has 50% in equities and 50% in fixed income investments.  He recognizes that because he has the annuity and Social Security, his actual investment mix is weighted more heavily in fixed income than equities, but he is comfortable with that result. 

New Research On Variable Spending Strategies (Like the One Recommended in This Blog)

In this December 10 article, Dr. Pfau compares the spending paths created by two variable withdrawal strategies: The Guyton Decision Rules and the Blanchett actuarial approach discussed in our November 20 post.  Unfortunately, Dr. Pfau's compares what is essentially a spending smoothing algorithm (Guyton) with year-by-year application of Blanchett's spreadsheet calculator without application of any smoothing of the results. However, as Dr. Pfau revealed in his article, Mr. Blanchett, "would almost certainly incorporate a moving average approach to smooth out the cash flows."
 
As I said in my original 2010 article (available in the articles section), the most important step in the five step general actuarial process to developing an estimate of how much you can spend each year involves periodic calculation of the theoretically correct spendable amount (using the simple spreadsheets found in this website, or Mr. Blanchett's spreadsheet or some other more "robust" calculator) and application of an algorithm to smooth actual experience as it occurs. See our post of October 11, 2003 for our recommended smoothing algorithm.
 
Dr. Pfau concludes that, "More research about variable withdrawal rates should look to build in a smoother spending path with changes only made when thresholds are crossed, and to more carefully calibrate the relationship between withdrawal rates and age." I agree and encourage Dr. Pfau to look at the approach recommended in this website.

Follow Up To July 23, 2013 Post--Delaying Commencement of Social Security

The consulting firm October Three has written a nice article about the potential financial advantages of delaying commencement of Social Security benefits until age 70. Readers of this blog will remember that we discussed this strategy and pointed readers to a spreadsheet on our site that would enable retirees to use their accumulated savings to "bridge" the period from age of retirement until age 70 (or some other age) in our post of July 23rd of this year. 
 
Using that spreadsheet, information for the example retiree in the October Three article, accumulated savings of $500,000 and the recommended assumptions described in this website (5% investment return, 3% inflation and survival until age 95), readers can confirm that if the example retiree retires at age 62 uses his accumulated savings as a Social Security bridge and defers commencement of his Social Security benefit until age 70, he can expect (under the recommended assumptions) to have total lifetime real retirement income of $39,130 per year starting at age 62 using the delay strategy vs. $35,655 per year if he commences Social Security at age 62. As can be seen in the spreadsheet runout tab, at age 70 he will be expected to have $305,906 of accumulated assets at age 70 under the delay strategy as compared with $514,533 under the non-delay strategy (commencing Social Security and level withdrawals from accumulated savings at age 62). The example retiree has essentially used a total of $240,804 of his accumulated savings to purchase a higher Social Security benefit commencing at age 70. To see the calculations using the delay strategy follow this link. Note that the estimated Social Security benefit commencing at age 70 has been increased by 3% per year for eight years of assumed CPI increases.
  
October Three argues that, "Rather than annuitizing retirement wealth, participants can get a much better deal by spending down retirement assets and deferring Social Security." While I like the article, I will have to reserve the right to pick a small bone with October Three over their use of "much better" here, as our post of September 22, 2013 shows comparable increases in total retirement income through combinations of self-insuring and purchase of deferred annuities (immediate, delayed or deferred). 
   
When considering the delay strategy, readers will also want to factor in other considerations, such as comfort in spending a significant amount of accumulated savings in the early years of retirement, taxation of Social Security benefits, possible future changes in Social Security law and possible changes in general interest rates/investment returns.