The Problem with Average Returns
Imagine a simple retirement projection: you have $1,000,000, plan to withdraw $50,000 per year, and your advisor assumes a 7% annual return. The math says your money lasts forever — the 7% gain exceeds your 5% withdrawal rate, so the portfolio grows indefinitely. Problem solved.
Except markets don't return 7% every year. In 2008, a diversified portfolio might have lost 38%. In 2019, it might have gained 31%. The sequence in which those gains and losses occur — not the average — determines whether you run out of money. If you hit a devastating loss in year two of retirement while also withdrawing funds, you permanently deplete the asset base available to recover. This is sequence of returns risk, and a straight-line projection completely ignores it.
How Monte Carlo Simulation Works
Monte Carlo simulation addresses this problem by modeling uncertainty explicitly. Instead of one projected return, it generates thousands of possible return sequences, each built by randomly drawing from a distribution of historical annual returns. Each sequence is used to run your complete retirement plan — inputs, withdrawals, inflation, Social Security, all of it — through to the end of the projected retirement period.
The result is not one answer but a distribution of outcomes. If 820 of 1,000 simulation trials end with money still in the portfolio, your success probability is 82%. That 82% represents genuine statistical uncertainty about what markets will do — not a single estimate dressed up as a guarantee.
The Inputs That Drive the Model
A Monte Carlo model for retirement planning typically incorporates:
- Mean return: The average expected annual return for your asset allocation, based on historical data (e.g., ~7% real return for equities, ~1–2% real return for bonds).
- Standard deviation: The volatility of those returns — how widely the actual returns vary from the mean from year to year.
- Inflation rate: Applied to withdrawals annually to maintain purchasing power.
- Withdrawal schedule: Annual spending, including any reductions in later years.
- Other income sources: Social Security, pensions, or rental income that reduce portfolio withdrawal needs.
- Time horizon: How many years the portfolio needs to last.
What "Probability of Success" Really Means
A 90% success probability does not mean there is a 10% chance you go broke. It means that in 10% of the simulated scenarios, the portfolio ran out of money before the end of the projected period. Some of those failures might be mild (the portfolio lasted 28 years instead of 30) while others might be severe (it ran out at year 15). The probability doesn't tell you about the severity of failures — only how often they occurred.
A 75% success probability is not a failing grade. It means your plan is quite good — three-quarters of simulated market environments would sustain your retirement. It also means there's meaningful room to make behavioral adjustments (spending less, working part-time, downsizing) if the real world starts looking like one of the unfavorable scenarios.
What Success Thresholds Mean in Practice
- 90%+: Very conservative plan — you may be significantly over-saving or under-spending. This is appropriate if you have no flexibility to cut expenses or no ability to return to work.
- 80–90%: The sweet spot for most retirees. Resilient across most market environments with reasonable room for adjustment.
- 70–80%: Viable but warrants attention. Consider increasing savings, working one more year, or planning for modest spending reductions in bad markets.
- Below 70%: Meaningful risk that should be addressed before retiring. Significant changes to the plan are needed.
Limitations of Monte Carlo
Monte Carlo simulation is the most rigorous retirement planning method available to individuals — but it is not perfect.
- Historical data dependency: Most models draw from historical market returns. If future returns are structurally lower — due to demographics, debt, or climate — the model may be optimistic.
- Normal distribution assumption: Many models assume returns are normally distributed. In reality, markets have "fat tails" — extreme crashes and booms occur more frequently than a normal distribution predicts.
- Behavioral factors: The model assumes you follow the withdrawal plan exactly. Most real retirees spend more in good times and less in bad ones — which generally improves outcomes relative to the model.
- Correlation during crises: In severe market stress, asset class correlations that normally provide diversification can break down — stocks and bonds both falling at once.
How to Use Monte Carlo Results
The most valuable use of Monte Carlo simulation is as a sensitivity tool — testing how changes to your inputs affect your probability of success. What happens if you retire one year later? If you reduce spending by $5,000? If you shift from 60% equities to 70%? Running these scenarios reveals which levers have the most impact on your plan and helps you make the right trade-offs.
Treat the output as a dashboard to revisit annually, not a one-time answer. As markets move, as your spending changes, as Social Security age approaches, your probability will shift. A plan that showed 85% success at age 60 may show 91% at age 65 after five more years of savings — or it may show 72% after a bad market sequence. Staying informed allows you to adjust before problems become crises.