Monte Carlo Simulation in Retirement Planning

A Monte Carlo simulation runs thousands of randomized scenarios against your financial plan to estimate the probability that your money lasts through retirement. Each scenario draws random returns, inflation, and spending shocks, then reports what percentage of paths end with a positive balance. A result of 90% means your plan succeeded in 900 out of 1,000 simulated futures.

The math is useful. But the most common failure is not in the simulation itself — it is in what gets left out. Taxes are almost always the largest category of missing information.

Why Taxes Belong Inside the Simulation

Most Monte Carlo tools operate on pre-tax dollars and apply a single blended return assumption across all accounts. Real retirement portfolios are split across taxable brokerage accounts, traditional 401(k)/IRA accounts, and Roth accounts — each with fundamentally different tax treatment at withdrawal. A simulation that ignores this distinction will overstate how much spendable income you actually have.

The gap is not small. A portfolio generating $150,000 per year in gross withdrawals might produce $105,000 in spendable income after federal and state income taxes, depending on your bracket, state of residence, and account mix. Running the simulation on $150,000 instead of $105,000 produces an artificially high success rate.

Mistake 1: Treating All Accounts as Pre-Tax

When a simulation models your $2 million portfolio without distinguishing between $1.2 million in a traditional IRA and $800,000 in a Roth IRA, it treats every withdrawal identically. In reality, traditional IRA withdrawals are taxed as ordinary income; Roth withdrawals are tax-free. The order and proportion in which you draw from each account each year changes your effective tax rate — and therefore your actual purchasing power — dramatically.

Mistake 2: Ignoring Required Minimum Distributions

Once you reach age 73, the IRS requires you to withdraw a minimum amount from traditional IRAs and 401(k)s each year, regardless of whether you need the money. These Required Minimum Distributions (RMDs) are calculated as a percentage of your account balance and increase as a share of the account over time. If your simulation does not model RMDs, it will underestimate your taxable income in your 70s and 80s — potentially pushing you into a higher bracket, triggering higher Medicare premiums (IRMAA surcharges), and increasing the taxable portion of Social Security benefits.

Mistake 3: Using a Fixed Tax Rate

Many simplified simulations apply a single assumed tax rate — say, 22% — to all withdrawals in all years. This is almost never accurate. Your effective tax rate in early retirement, when you may have no earned income, is often very low. It rises when Social Security begins, rises again when RMDs start, and fluctuates depending on realized capital gains, Roth conversions, and large one-time expenses. A flat rate assumption can easily be off by 5–10 percentage points in either direction in a given year, which compounds into large errors over a 30-year retirement horizon.

Mistake 4: Omitting Capital Gains Tax on Taxable Accounts

Withdrawals from taxable brokerage accounts are not simply income — they depend on your cost basis. If you sell an ETF held for more than one year, only the gain above your purchase price is taxed, and at the preferential long-term capital gains rate (0%, 15%, or 20% depending on income). A simulation that models taxable account withdrawals as ordinary income will overstate your tax burden. Conversely, one that ignores capital gains entirely will understate it. The correct approach requires tracking cost basis and realized gains separately by year — something few generic tools do.

Mistake 5: Not Modeling Tax on Social Security

Up to 85% of Social Security benefits are subject to federal income tax, depending on your combined income (adjusted gross income plus non-taxable interest plus half of Social Security). For most retirees who draw from both taxable accounts and traditional IRAs while receiving Social Security, a substantial portion of their benefits will be taxed. Simulations that treat Social Security as fully tax-free are overstating real income in the years benefits are received.

Mistake 6: Skipping State Taxes

State income tax treatment of retirement income varies widely. California taxes all ordinary income including IRA withdrawals at rates up to 13.3%. Florida and Texas have no income tax at all. Some states exempt pension income but tax IRA distributions. If your simulation does not account for your state's tax rules, the success probability it reports may be materially too optimistic for high-tax-state residents and somewhat too pessimistic for those in no-tax states.

What a Tax-Aware Simulation Should Do

A properly constructed Monte Carlo model tracks each account type separately, simulates withdrawals from each bucket in an optimal sequence, calculates federal and state income taxes year by year based on the actual income stack (ordinary income, capital gains, Social Security), applies RMD schedules to traditional accounts, and models the impact of Roth conversions when marginal rates are low. This is more complex than a single-rate model, but it is the only way to get a success probability that reflects what you will actually be able to spend.