Hi Guys,
A major uncertainty when making a retirement decision is the expected future market returns. A standard analytical tool that is frequently exercised when attacking this quagmire is Monte Carlo calculations.
When using a Monte Carlo approach, a constant issue is what to input as a likely returns profile as a function of the asset allocation and time. Many simulations deploy the historical statistical market segment data sets. Others use perturbed versions of these same data sets that reflect future expectations.
These inputs are the fundamental drivers that determine an allowable annual drawdown rate. Of course, the target goal is a high likelihood of portfolio survival for the specified retirement period. That target goal is typically a 90 % to 95 % portfolio survival probability.
A recent paper addresses this essential, but cloudy, input issue of candidate future market rewards. Rather than using a random draw to initiate the overall process, this refined Monte Carlo approach starts with current bond returns and the present equity P/E ratio to guide the future returns profile.
This 17 page report by Blanchett, Finke, and Pfau is titled “Asset Valuations and Safe Portfolio Withdrawal Rates”. Here is the Link to the paper:
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2286146If you are not particularly inspired to examine the paper itself, here is a Link to a detailed opinion of that work:
http://www.kitces.com/blog/archives/480-Safe-Withdrawal-Rates-In-Todays-Low-Yield-Environment-Walking-On-The-Edge-Of-A-Cliff.htmlThis Nerd’s Eye View of the academic study fairly examines both the strong points and the shortcomings of the research work.
In its abstract, the research paper properly identifies the motivation for the study as follows: “Portfolio returns in the first decade of retirement have an outsize impact on retirement income strategies. Traditional Monte Carlo simulation approaches generally do not incorporate market valuations into their analysis.”
The referenced report employs a regression curve fitting approach to current bond yields as a point of departure for the fixed income modeling, and Robert Shiller’s cyclically adjusted price-to-earnings (CAPE) ratio model as an estimate of future equity market rewards. Inflation is incorporated into the formulation and is closely tied to the mid-term bond yield. As is usually the case, the overarching modeling is not completely accepted by market wizards without some controversy.
Historically, most earlier Monte Carlo studies endorsed a roughly 4 % annual drawdown rate to sort of guarantee a high portfolio survival likelihood. The referenced work challenges this assertion given today’s investment environment. That’s not unexpected since the input anticipated return schedules are muted relative to historical averages. The referenced study challenges the safety of the 4 % withdrawal chimera.
My takeaway from all these eloquent simulations is that any investor preparing for a retirement should assemble a sufficiently robust portfolio such that its planned drawdown rate should be approximately 3 % less than the projected annual returns for the portfolio. For example, if your portfolio is designed to generate a 6.5 % annual return over the long haul, than a 3 % withdrawal schedule, adjusted for inflation, should deliver a comfortable 90 % to 95 % probability of success.
Also, an alert retiree will adjust that drawdown schedule if the portfolio suffers some shortfalls. Monitoring and flexibility are always vital elements in any planning and execution exercise.
All this need not be complex rocket science. As Albert Einstein said: “ Out of clutter, find simplicity. From discord, find harmony. In the middle of difficulty, lies opportunity.”
Monte Carlo tools provide such an opportunity. I really like Monte Carlo simulations, but I also recognize their limitations. They serve best in providing guidelines. Here is a Link to a very simple and useful Monte Carlo tool that is easily used:
http://www.moneychimp.com/articles/volatility/montecarlo.htmThe code provides an estimate of portfolio survival probability. The ease of input and speed of execution allows the user to do many probing parametric cases quickly. Enjoy.
Monte Carlo analysis is applied only when definitive, precise assessments can not be made. That’s exactly a characteristic of the marketplace. That’s particularly true when planning for retirement, and forced to project highly uncertain market returns. I do recommend you use available Monte Carlo analysis when doing your retirement planning. Many websites offer that service.
Good luck to all you folks who are now evaluating the retirement option. Given the results of the referenced study, that luck is most productive to your survival prospects when gifted early in the retirement years.
Best Regards.
Comments
However, if my math with calculators is correct; one finds that at least for traditional IRA accts. and MRD/RMD (minimum required distributions, per IRS), the following applies for a variety of inputs I placed into MRD calculators:
Using various dollar values for an IRA, different age ranges and several different rates of return for a portfolio, ,the first year mandatory drawdown from an IRA account starts at 3.6% of the total value, and the percentage increases about .1% each year going forward.
The IRS has already established a drawdown for many millions of baby boomers going forward.
'Course this example only reflects towards IRA monies, with no other consideration given to any other cash flows into a household during retirement.
Regards,
Catch
Another issue I contemplate - Hypothetical situation: Needing to take a distribution after age 55.5, but prior to RMD age. Better to take it from Traditional IRA or Roth? Conventional wisdom says take from Traditional. I'm not so sure. From the Roth to obtain $1000 you withdraw $1000. From the Traditional, however, you need to withdraw approximately $1200-$1300. That's $1000 to meet your spending needs and roughly $200-$300 to cover state and federal taxes. By electing to withdraw from the Roth, you have allowed that extra $200+ to compound tax-free inside your Traditional IRA. This is an over-simplification. But, I do think there's good arguments on both sides, and the decision's not as cut-and-dry as some would suggest.
Hi Hank,
Thank you for contributing to this discussion.
But I’m just a little confused. Actually, that’s an improvement given that I am frequently very confused. I am puzzled by the apparent ease by which I resolved your proposed scenario with regard to the IRA or Roth drawdown decision.
I quickly concluded that the conventional wisdom was correct. When necessary, take the needed funds from the standard IRA. Perhaps I’m missing some nuance.
I did one simple calculation to reach my conclusion. Assume you had both a 10,000 dollar IRA and Roth. Further assume both are expected to generate a 6 % annual return, and you plan to withdraw the end total 15 years from now. Postulate a 1000 dollar emergency.
Let’s evaluate the two options. First assume the needed funds are withdrawn from the IRA. Then,
The expected end value for the Roth is 10000 X (1.06)^15 = 23,966 dollars tax free.
The expected end value for the IRA is 8800 X (1.06)^15 =21,090 dollars assuming 200 dollars are needed for taxes. The tax is a current 16.7 % tax bite. If the tax rate remains the same, that leaves an after tax net of 17,568 dollars.
The end wealth for the IRA option is 23,966 plus 17,568 dollars after taxes. The total is 41,534 dollars.
Next, I do the same calculation assuming a Roth initial withdrawal.
The expected end value for the Roth is 9000 X (1.06)^15 = 21,569 dollars tax free.
The expected end value for the IRA is 10000 X (1.06)^15 =23,966 dollars. If the tax rate remains the same 16.7 %, that leaves an after tax net of 19,964 dollars.
The end wealth for the Roth option is 21,569 plus 19964 dollars after taxes. The total is 41433 dollars.
Given this very simple, single case analysis, the slight edge favors the conventional wisdom of taking this emergency withdrawal from the IRA sleeve.
Timeframe, projected market returns, and the tax schedule now and in the future all enter the calculation. If the conditions change, the optimum option selection could also change. A single strategy doesn’t fit everyone for all times.
Complexity contributes to confusion. Maybe that partially explains my normal state.
Best Wishes.
Hi Hank,
Indeed, the anticipated returns profile influences final results and the IRA/Roth decision.
As I mentioned in my initial post, projected returns as well as the timeframe under consideration and the tax structure obligations at withdrawal are key elements in the assessment. Circumstances dependent, I believe (but surely have not proven) that decision crossover points exist. An extensive study is needed to identify those crossover intersections.
One additional point is easily calculated with the simple method that I used. Best intentions aside, it is easy to assess the worthiness of either option if the annual estimated return is Zero percent.
Given that disastrous scenario, the end wealth after taxes for both cases is exactly equal at 17,330. dollars. It is a push.
In all scenarios, a decision depends on a projected annual rate of return schedule. Monte Carlo analyses designed for this specific decision would yield guideline probability estimates.
Best Wishes.
Secondly, I have an issue with your calculation. It is too simplistic even if we have constant growth of portfolio. You have calculated as if you are not taking distributions during those 15 years. In the case of IRA you will be paying taxes each year based on the distribution you got for that year. Assuming you do not have another source to pay your taxes to have the net income the same between Roth and IRA, you will have to withdraw more from IRA. Higher withdrawal will taper the tax over time.
Hi Investor,
Good to hear from you again. I always look forward to your informed postings.
Yes, I elected to do a highly simplified calculation. It was necessary to capture the essential tradeoffs between the two options being examined. To fully explore the real world scenario of variable annual market returns requires access to a more robust analytic tool like a Monte Carlo simulator. I do not presently have such access.
Even with the simplifications, I believe the manageable analysis that I did correctly identifies and illustrates the complexity of the IRA/Roth decision.
I currently own IRA-like and Roth products. I have been in the mandatory drawdown phase of the IRA holdings for many years now. I managed to avoid tapping into those IRA funds until the mandatory start date. I suspect that is a constant goal in the planning for most investors when making their IRA contributions. Hence, the calculations were executed with that perceived norm.
I am definitely not a tax expert. So I hire a professional to do my annual Federal and State taxes. From memory, since I never withdrew any IRA monies earlier than the mandatory distribution date, I never worried or suffered tax consequences from accumulated earnings in those holdings until I initiated the mandatory drawdown. So, I do not fully understand your logic in the closing paragraph. Please verify the taxation laws applicable to IRAs.
The model calculation postulates no distributions for the 15 year period. I think that is a reasonable assumption and a very likely real-life scenario. It duplicates my situation.
Thanks again for your interest in this submittal, and your numerous excellent MFO references and commentaries.
Best Wishes.
You noted: "I think you are beginning with a wrong assumption. If you invest in Roth instead of IRA, you are likely to have contributed less since you had to pay taxes on that amount so you had less money to invest (assuming that all your other spending/saving remains the same)."
The amount of complexity and/or what an individual may hold in various retirement accounts and where the money lands after retirement can have many paths.
Example: Mid-1970's, company "A" has a defined benefit pension plan. Any individual investing at this time was likely in taxable, personal accts.
---late 1970's, finds that an individual may now invest in an "IRA". $1,500 maximum and deductible against one's gross income (if not covered by a company pension plan). In 1981, this provision was removed and opened IRA's to most folks.
--- mid-1980's, now finds something named a 401K being offered by some companies.
1990 example of possible choices for an individual:
--- trad. IRA, $2,000 maximum input and perhaps deductible against one's income, dependent upon gross income upper limits.
--- 401k/957,403b's, etc. The amount invested may be used to reduce one's taxable income.
1997 now finds that addition of the Roth IRA.
At this house (beginning 1981), it was invest the max. per year into the trad. IRA. 401k's were not yet available. Beginning 1990 finds 401k's available to our house investments.
At some point in time around 1992, no more was added to the trad. IRA; with all monies going into the 401k's for the maximum allowed. The 401k's and then the addition of Roth IRA's allowed us maximum input amounts into both areas.
Well, anyway; the summary could be that at an age 65 retirement (at least for boomer group folks) may find not much money in a trad. IRA, more in a Roth IRA and a larger amount in a 401k, etc.
Now, if the retiree rolls the 401k plan; the amount that is now in traditional IRA accts. could be a very large amount of money versus the limited amount that was available to place into the Roth IRA.
Sadly, I know too many folks who didn't do much or any of the above choices. Time (compounding) would have been their dear money friend.
I am not trying to make any particular point with this. Only that some folks will have various money values in different acct. types, depending upon where they placed monies during their working years. We always maximized the limits for our trad. IRA's, 401k's and Roth's.
The real serious challenges will be for those who now only have a 401k and related; and/or the Roth path to save/invest, as their pension plans.
Take care,
Catch
My comment in amount of taxes payed tapering was based on some fixed percentage of the remaining amount in the portfolio. As the portfolio shrinks so does the withdrawals and the required taxes. If you use this type of distribution the earlier years of the distribution years have more effect on total taxes paid over time.