"Best" withdrawal decisions

Have a question about your personal investments? No matter how simple or complex, you can ask it here.
Post Reply
Topic Author
Posts: 61
Joined: Sun Mar 04, 2007 6:23 pm
Location: Portland, OR

"Best" withdrawal decisions

Post by baldeagle » Tue Mar 11, 2008 7:39 pm

We are a couple of 63-year-olds who retired about three years ago, and took SS at age 62. I have always felt that financial decisions made in the ten years between when we retired and age 70.5, when we cede withdrawal and tax control to MRDs, could be significant for the rest of our lives. In particular, the multi-variable decisions of account withdrawal sequencing, Roth conversion, and SS repay/restart seemed particularly important. But what decision is “best?”

I have read many books, articles, and posts that address these decisions individually, but saw nothing that I could map wholly and quantitatively to our situation. So, this post describes what we did to figure it out for us. Have any of you done something similar? Did you come to similar conclusions?


We have a 60/40 portfolio comprised of about 80% IRA, 15% taxable, and 5% Roth. Our living expenses require a withdrawal rate of about 3.2%. We are well within the 15% FIT bracket. Our current WD approach has been:
1. Living expenses from IRA
2. Taxes and extraordinary expenses from taxable account (for as long as it survives, then switch to Roth)
3. Convert IRA to Roth up to the top of the 15% bracket for as long as we can

The reason for this approach has been to get the IRA value reduced as much as possible as fast as possible, without taking “too big” a tax bite, so that RMD amounts are reduced when they hit us forever in seven years. Quantifying “too big” was part of the purpose of this exercise.


I build a model to see how different scenarios would play out over the next 40 years, accounting for inflation and taxes. The model consists of an Excel workbook of five spreadsheets, each of identical structure with about 50 rows and 35 columns. The differences between the spreadsheets implement the scenarios described below.

Evaluation criteria
1. Total portfolio value over time (most important)
2. Maximum Roth account value (for maximum control)
3. Minimum excess RMDs (i.e., minimum required WDs over what we needed)
4. Minimum incursion into 25% bracket over the years

Criteria 3 and 4 are viewed as hedges against expected future tax increases.

Scenarios - Five were chosen:
1. Current approach (described above)
2. Living expenses and taxes from taxable until exhausted, then revert to current approach, plus Roth-convert to top of 15% bracket
3. Current approach but repay SS and restart at age 66, plus Roth convert to top of 15% bracket
4. Current approach but repay SS and restart at age 70, plus Roth convert to top of 15% bracket
5. Current approach but abandon Roth conversion – just stay low in 15% bracket until age 70.5 when RMDs take control

Calculations – For every year from now to age 100, the model computes the following, adjusting for inflation and investment return every year:
1. Tax brackets, deductions, exemptions
2. SSI, pension, living expenses, extraordinary expenses (cars, etc.), fed/state taxes
3. IRA withdrawals, conversions, RMDs
4. Account values (taxable, IRA, Roth, total)

Economic sensitivity – To assess scenario sensitivity to economic conditions, I ran the model for three conditions:
1. Mid – 7% return, 3% inflation
2. Low – 6% return, 4% inflation
3. High – 8% return, 2% inflation


The model shows that for our specific situation:
1. Scenario number 2 is always better than our current approach.
2. It is also the best scenario of all in Mid and High economic conditions.
3. Only in Low economic conditions does SS repay/restart become best, but that takes until age 87 to happen.

We will change our approach to scenario 2 and forget about SS repay/restart.

For our particular situation, there seems to be a "sweet spot" in all this:
1. Spend taxable account first. Pretty common wisdom, but it only works in our situation if we also make Roth conversions. Otherwise we save taxes early, but end up with bigger IRAs at age 70.5 and bigger RMDs requiring more taxes.
2. Convert as much as possible as early as possible
3. But not over the 15% bracket
4. Because the additional taxes (nearly double the rate) take too big a bite too early
5. Making it take too long for the portfolio to recover

Again, this post describes what we did to figure it out for our situation. Have any of you done something similar? Did you come to similar conclusions?

User avatar
Taylor Larimore
Advisory Board
Posts: 28744
Joined: Tue Feb 27, 2007 8:09 pm
Location: Miami FL

Order of withdrawals

Post by Taylor Larimore » Tue Mar 11, 2008 8:41 pm

I retired over 20 years ago with about half our portfolio in IRAs and half in a taxable account.

We started withdrawing from our taxable account first for these reasons:

1. Only profits in taxable account withdrawals are taxed, and these are at low capital-gain tax rates (15% max.). In traditional IRAs, (All withdrawals are taxed at income tax rate (35% max.).

2. Our taxable account is nearly all stocks. Selling our taxable funds first, automatically reduces our stock exposure as we age.

3. Selling taxable funds first, eliminates capital-gain taxes that would otherwise be paid when rebalancing.

4. Selling taxable funds first, allows our IRA accounts to continue maximum tax-deferral.

5. Selling taxable funds first, maximizes our retirement accounts which provide better creditor protection.
At age 70 1/2 we had to begin taking our IRA RMDs (required minimum distributions) ahead of our taxable account withdrawals.

There are some cases when it is better to withdraw retirement funds first. This is particularly true for taxable securities with large capital-gains that are to be left to heirs.

I strongly endorse your plan to convert (up to the next tax-bracket) to your Roth IRA during periods of
low income.

Below is a link to an article written by the Financial Planning Association with a recommended order of portfolio withdrawals. I hope our experience and the article are helpful.


Best wishes.

Posts: 38142
Joined: Thu Jan 10, 2008 12:56 pm

Post by retiredjg » Wed Mar 12, 2008 12:08 pm


Thank you for posting this information. This is something I've been wondering about since I retired about 10 months ago. Obviously, you have gone to a lot of trouble and this information is helpful. Sorry, I don't have any info of my own to share back as you asked. So far, I'm just winging it, spending taxable first. But mostly by luck.

Thanks Taylor, for sharing your thoughts and the link. Both will be helpful. jg

User avatar
Posts: 974
Joined: Mon Feb 19, 2007 9:45 pm
Location: New Orleans

Post by Prokofiev » Wed Mar 12, 2008 12:24 pm

Your model can be no better than the inputs you used.

Tax rates for the next 40 years?

Inflation rates and related changes in tax brackets?

Changes in dividend / LTcap gain tax rates?

In particular, Roth conversions up to the 15% bracket max are clearly advantageous, but for many filling part or all of the 25% bracket will also make sense. Once you begin RMDs you may be surprised at the true marginal rate. My bet is that the vast majority of investors will be above a 15% marginal rate.
Everything should be made as simple as possible, but not simpler - Einstein

Posts: 3564
Joined: Tue Feb 20, 2007 4:58 pm

Post by heyyou » Wed Mar 12, 2008 12:30 pm

Kotlikoff's consumption smoothing program addresses some of those spending sequences. Perhaps someone who has purchased it, could chime in about what scenarios they plugged in. Better yet, what are the assumptions?

Posts: 2584
Joined: Wed Feb 21, 2007 4:57 am

Post by Chip » Wed Mar 12, 2008 2:01 pm

I'm also executing your scenario 2. I thought I did a good deal of analysis, but I think you've covered the angles in much more detail than I did.

I'm beginning to question whether the top of the 15% bracket is going to be enough to control the RMDs, given that I expect tax rates to rise.

Topic Author
Posts: 61
Joined: Sun Mar 04, 2007 6:23 pm
Location: Portland, OR

Post by baldeagle » Thu Mar 13, 2008 11:01 pm

Thanks all for your valuable comments. I finally figured out how to get a crude web site going at Geocities to post some of the Excel charts I made from my model. You can see them here: http://www.geocities.com/baldeagleNW

The first three charts show how portfolio value changes over time for the seven scenarios. My original post only described five scenarios but I've included two more here:

1. Min RMD -- Target Roth converting so as to keep MRDs from ever exceeding our personal IRA withdrawal needs. I.e., we wag the tail, the tail doesn't wag us.
2. Min 25% bracket -- Target Roth converting so as to keep us out of the 25% bracket for life after age 70.5.

These are both pretty heavy conversions, especially the second one. Neither fares well in poor economic conditions, presumably because the early tax hits to portfolio value preclude it from ever recovering during such conditions.

The fourth shows how the individual accounts fare over the years with scenario 2, the best choice for our situation.

This is, of course, a simple, linear model. Maybe I will figure out a way to add some randomization of returns and inflation. But for now, I think it still offers some insight in decision making. Dissenting opinions solicited.

User avatar
Posts: 4764
Joined: Sun Feb 25, 2007 8:55 pm
Location: Southern California

Post by CABob » Fri Mar 14, 2008 1:29 pm

Interesting data and analysis. On your scenarios 3 and 4, did you assume both of your SS benefits would be repayed or just one? And did you consider both ways?
I am considering repaying SS benefits, but, only for myself and not for my wife. I am a few years older than wife and receive higher benefits than she does.


Topic Author
Posts: 61
Joined: Sun Mar 04, 2007 6:23 pm
Location: Portland, OR

Post by baldeagle » Fri Mar 14, 2008 2:26 pm


Both -- I only did the SS analysis for both of us together.

I could run it the other ways (repay/restart her not me, me not her), but I am guessing that whatever I do with my benefit will dominate the results because my benefits are slightly more than 2x my wife's.

Post Reply