Optimal Retirement Withdrawal Strategy

Introduction

Most retirement calculators follow the conventional retirement savings withdrawal strategy of withdrawing all the funds from the After-tax Account first, followed by the Tax-deferred Account and finally the Roth IRA.

ORP withdraws money from all three accounts in parallel in order to reduce personal income taxes.

The question is: How much (in real numbers) does ORP improve over a conventional retirement calculator's withdrawal method?

There was no easy way to compare the results produced by an Internet accessible, conventional retirement calculator to those of ORP and vice versa.

To answer this question we implemented a conventional retirement calculator (named Naive) that produces results compatible with ORP. Naive follows the conventional withdrawal method of withdrawing all after-tax funds first and then withdrawing funds from the Tax-deferred Account. Naive does not model Roth IRA accounts.

Naive is a typical conventional retirement calculator except that instead of determining the year that the retirement funds are exhausted, it iterates to the stated estate amount by incrementally adjusting annual spending. Naive's after-tax, annual spending expressed in today's dollars is directly comparable to ORP's version of the same. Furthermore, the plan value for both calculators is the sum of all inflated, after-tax spending for the life of the plan.

Computational Results

Table 1 compares results between the two methods for a basic situation uncluttered by extenuating circumstances::

  • husband age 65, wife 65,
  • retirement savings of all types = $3M, proportioned between the after-tax and tax-deferred accounts,
  • retire at 65,
  • no social security,
  • life expectancy 85,
  • estate $10K,
  • no house,
  • no Roth IRA:
  • inflation 3.5%
  • investment returns 7%
  • Table 1: Computational Summary

    Test

     

    Spending

     

    Value

     

    % Improve

     

    After-Tax

    IRA

    Conv.

    ORP

    Conv.

    ORP

    Annual Sp

    Value

    3000

    0

    178

    179

    5,398

    5,399

    0.3%

    0.0%

    2700

    300

    177

    181

    5,311

    5,490

    2.3%

    3.4%

    2400

    600

    176

    182

    5,235

    5,510

    3.8%

    5.2%

    2100

    900

    174

    182

    5,168

    5,500

    4.8%

    6.4%

    1800

    1200

    172

    181

    5,106

    5,472

    5.3%

    7.2%

    1500

    1500

    170

    180

    5,046

    5,438

    6.0%

    7.8%

    1200

    1800

    168

    177

    4,986

    5,337

    5.2%

    7.0%

    900

    2100

    165

    172

    4,926

    5,221

    4.3%

    6.0%

    600

    2400

    162

    169

    4,867

    5,097

    3.9%

    4.7%

    300

    2700

    160

    164

    4,814

    4,981

    2.8%

    3.5%

    0

    3000

    157

    160

    4,749

    4,802

    1.9%

    1.1%

    Table 1 the compares the performance of Naive's conventional withdrawal method to ORPís optimized schedule for various sizes of the After-tax Account and the Tax-deferred Account but always totaling $3M. The two Spending columns are expressed in today's dollars. The two Value columns are inflated dollars. The first and last rows of Table 1 are trivial cases that serve to indicate the computational discrepancies between the two implementations of the same model.

    Ignore the fact that when reading down the columns the annual spending values get smaller as the After-tax Account gets smaller, giving the impression that the After-tax Account is outperforming the Tax-deferred Account. This illusion occurs because the After-tax Account pays its taxes as the income is incurred and pays no personal income taxes on withdrawals. Table 1 ignores the important fact that contributions to the Tax-deferred Account were not taxed as they were made and investment returns prior to withdrawal were not taxed.

    The important message of Table 1 is in the two rightmost columns; the percentage improvement of ORPís optimized withdrawal strategy over the conventional approach. Annual Spend shows ORP's advantage for each year of retirement. The Value improvement is over the life of the plan.

    Next in Table 2 Social Security benefits are added to the mix to give a more typical retirement situation. Table 2 assumes that both the retiree and spouse began receiving $24,000 annual Social Security benefits at age 65.

    Table 2: With Social Security Benefits

    Accounts

     

    Spending

     

    Value

     

    % Improve

     

    After-Tax

    IRA

    Conv.

    ORP

    Conv.

    ORP

    Annual Sp

    Value

    3,000

    0

    224

    223

    6,776

    6,763

    -0.2%

    -0.2%

    2,700

    300

    221

    224

    6,634

    6,792

    1.4%

    2.4%

    2,400

    600

    218

    224

    6,513

    6,790

    2.7%

    4.3%

    2,100

    900

    216

    223

    6,407

    6,762

    3.6%

    5.5%

    1,800

    1,200

    213

    221

    6,311

    6,692

    4.0%

    6.0%

    1,500

    1,500

    210

    219

    6,222

    6,596

    4.4%

    6.0%

    1,200

    1,800

    206

    214

    6,139

    6,490

    3.9%

    5.7%

    900

    2,100

    203

    210

    6,059

    6,370

    3.4%

    5.1%

    600

    2,400

    200

    207

    5,981

    6,229

    3.5%

    4.1%

    300

    2,700

    196

    201

    5,909

    6,072

    2.4%

    2.8%

    0

    3,000

    193

    196

    5,831

    5,910

    1.5%

    1.4%

    Table 2 shows that the largest percentage spending improvement (4.4%) for ORP over Naive comes for the 50/50 split between the two accounts. For a typical situation savings are heavily weighted toward the Tax-deferred Account, as shown in rows 8, 9 and 10 . We see that optimal withdrawal strategy will increase annual retirement spending from 2.4% to 3.5% per year. This turns out to be $163,000 to $320,000 over the span of the plan.

    Comparing rows 8, 9 and 10 of Tables 1 and 2 shows that Social Security benefits lessen the percentage improvement of ORP over Naive. Other quantitive studies on this topic such as Horan[1] and Spitzer and Singh [2] omit Social Security benefits from their computations and thus tend to overstate the advantage of their Informed withdrawal methods over the Naive methods.

    Table 3 extends the Table 2 situation by including a $300,000 illiquid asset, the couples home, to be liquidated in year age 80.

    Table 3: Illiquid Asset Included
    Scenarios   Spending   Value   % Improve  
    After-Tax IRA Conv. ORP Conv. ORP Annual Sp Value
    3,000 0 236 237 7,175 7,158 0.2% -0.2%
    2,700 300 231 238 6,887 7,187 3.1% 4.4%
    2,400 600 231 238 6,887 7,182 3.1% 4.3%
    2,100 900 228 237 6,781 7,151 4.0% 5.5%
    1,800 1,200 225 234 6,686 7,076 4.4% 5.8%
    1,500 1,500 221 230 6,595 6,977 4.0% 5.8%
    1,200 1,800 218 228 6,516 6,864 4.4% 7.1%
    900 2,100 215 223 6,437 6,747 4.0% 4.8%
    600 2,400 211 218 6,354 6,600 3.2% 3.9%
    300 2,700 208 213 6,282 6,442 2.6% 2.5%
    0 3,000 204 208 6,202 6,278 1.8% 1.2%

    Adding the home to the retirement mix increases the couple's assets to $3,300,000 and gives ORP more room to work in. Again rows 8, 9 and 10 are typical of most retirement plans. Comparing these rows between Tables 2 and 3 shows that ORP is able to take advantage of the additional asset.

    Table 4 shows ORP's Withdrawal Plan for the 50/50 split scenario that includes the both Social Security benefits and an illiquid asset.

    Table 4: ORP Withdrawal Plan
    Age TaxDef AfterTax RothIRA IRA2Roth SocSec Taxes Spending
    65 139 214 139 48 -31 230
    66 144 222 144 50 -32 238
    67 149 230 149 51 -33 247
    68 152 237 152 53 -34 255
    69 65 222 65 55 -12 264
    70 67 180 17 57 -12 274
    71 69 185 17 59 -13 283
    72 72 175 61 -13 293
    73 74 181 63 -14 303
    74 77 187 65 -14 314
    75 79 194 68 -15 325
    76 82 201 70 -16 336
    77 161 151 73 -35 348
    78 217 118 75 -49 360
    79 225 122 78 -51 373
    80 94 230 80 -18 386
    81 98 238 83 -19 399
    82 249 73 62 86 -56 413
    83 258 140 89 -58 428
    84 267 145 92 -60 443
    85 276 150 96 -63 458

    Table 4 shows that ORPís strategy is to do continuous withdrawals from the Tax-deferred Account in a manner that keeps taxes under control and takes maximum advantage of the Roth IRA. Early in retirement the After-tax Account (column 2) is being drawn down for spending without paying personal income taxes. Partial distributions are being annually rolled over from the Tax-deferred Account (column 1) to the Roth IRA (column 4) paying personal income taxes on each withdrawal.. When the After-tax Account is empty the Tax-deferred and Roth IRA accounts provide for spending.

    At age 80 the illiquid asset is sold and the proceeds fund spending, along with a low level of withdrawals from the Tax-deferred Account. Once the house proceeds are exhausted then withdrawals resume from the Roth IRA for the remainder of the plan. Age 82 is particularly interesting because it shows withdrawals from all three accounts in the same year.

    The Taxes column (column 7) tells the story. The highest taxes are paid early on during the IRA to Roth IRA rollover phase. The Roth IRA is stocked up early in retirement to provide tax free distributions to supplement the IRA distributions and keep them in a lower tax range, 15% in this case, for the rest of retirement.

    Conclusion

    : ORPís optimal retirement withdrawal strategy increases the amount of money available for spending during retirement. ORPís strategy is to take advantage of the Federal progressive income tax code. ORP makes partial rollovers to the Roth IRA when they are only lightly taxed and then withdraws the money from the Roth IRA tax free later on when Naive's strategy makes large, taxable withdrawals from the Tax-deferred Account.

    References

    1. Withdrawal Location with Progressive Tax Rates; Stephen M. Horan; Financial Analysts Journal; June 2006; Volume 62, No. 6; pages 77-87
    2. Extending Retirement Payouts by Optimizing the Sequence of Withdrawals John J Spitzer and Sandeep Singh; Journal of Financial Planning; Apr 2006; 19, 4; ABI/INFORM Global pg. 52
    Last Update: July 31, 2008

    © 1998-2008, James S. Welch, Jr
1