Perspectives · September 2026

Almost every retirement plan now ends with one number: the probability of success. Ninety-five percent. It is presented as an answer, and most people take it as one.

It is not an answer. It is an average. And for someone who has just retired, it averages away the only risk that really matters.

What the number measures

A Monte Carlo simulation runs a thousand or more market histories, each with returns in a different random order, and counts the ones in which the money lasts. Some of those histories begin with several bad years. They are in the count. The risk that bad returns arrive at the wrong time is already inside the number.

So the problem is not that the simulation ignores that risk. The problem is that the number buries it.

What it buries

Look only at the failures and a pattern appears. Almost all of them begin the same way: a severe decline in the first few years of withdrawals. The success rate treats those failures like any others and blends them with hundreds of ordinary histories in which nothing goes wrong. The result looks reassuring precisely because bad starts are rare.

In a recent case, a newly retired couple's plan succeeded in about 95% of simulated futures. Among the futures that began with a severe five-year decline, it succeeded in about 55 to 65%. Same plan, same assumptions. The only difference was where we looked.

The familiar fan chart does not help much. Its lower band shows how low the portfolio can go in each year, but it stitches together different histories year by year. No one lives along that line. It tells you how bad, not when, and not whether the portfolio comes back.

The standard run also rests on two quiet assumptions. The first flatters the result: returns and inflation are usually drawn separately, so stretches in which markets fall while prices rise, as they did in the 1970s, show up less often than history suggests. Even the expected return itself is uncertain. Eugene Fama and Kenneth French, simulating U.S. stock returns over horizons as long as thirty years, found that this uncertainty has little effect over short periods but a substantial effect on long-horizon outcomes. The second hides the remedy: spending and the portfolio's mix are held fixed, as if nobody would do anything differently when markets turn. That is not how people live, and it is not how plans should be built. What a household does in the first bad years decides whether a bad start becomes a setback or a lasting problem, and the simulation, as usually run, cannot show it. We come back to those decisions below.

Why the early years are different

A decline early in retirement is not dangerous because of the decline. It is dangerous because of the withdrawals that continue through it.

You sell at low prices to pay the bills. The portfolio shrinks. The same spending becomes a larger share of what is left. In one scenario we tested, a couple drawing 3% of their portfolio before a five-year decline was drawing 5.5% after it, without spending a dollar more. At that rate, ordinary returns only hold the balance steady. The bad years end. The damage does not.

The same decline twenty years later barely registers. Timing, not size, is what makes it costly.

What the research says

This is not a new observation, and the response is not a house theory. The study of retirement withdrawals goes back more than thirty years, to William Bengen's work in the 1990s. The most useful evidence for this question comes from the research tradition our investment approach is built on.

Eugene Fama and Kenneth French have shown that even when stocks are expected to beat safer assets over time, their volatility makes a shortfall over three- and five-year periods a substantial possibility, and a meaningful one over ten and twenty years. A bad five years is not a prediction. It is a real possibility that a retirement plan has to be built to absorb.

Dimensional's retirement research measures what that means once withdrawals begin. In 100,000 simulated retirements, a stock-heavy allocation ran out of money by age 85 in 5.7% of cases. Among the retirements whose first five years brought the weakest stock returns, that figure rose to 33.7%, a one-in-three chance of running dry twenty years into a thirty-year plan. An allocation that held less in stocks and matched inflation-indexed bonds to planned spending failed in 0.1% of cases overall, and in only 1.2% of those same bad starts.

The lesson is consistent. The stock allocation, and the bonds that sit beside it, matter most at the start of retirement, when wealth is largest and withdrawals have just begun. The allocation that suits a thirty-year retirement is not necessarily the one to hold on its first day.

Same stocks, different timing

A conservative start does not mean owning fewer stocks over a retirement. It means owning them at a different time. A portfolio that begins at 50% and rises to 70% holds roughly as much in stocks, on average, as a steady 60/40 or 70/30. The difference is when the exposure is carried.

That is also why a well-run static 60/40 remains a sound choice, and why the research finds the improvement from a rising path to be real but modest. The two approaches converge in practice. If a 60/40 portfolio pays for spending from its bonds first, and is not mechanically rebalanced back into stocks in the middle of a decline, its stock share rises on its own as the bonds are drawn down. A deliberate glide path simply sets the starting point and the pace on purpose, instead of leaving them to the market.

The reasoning underneath is simple. Expected returns don't depend on the calendar, but the consequences of volatility do. A retiree's real risk is the spending the portfolio has to support, not volatility in the abstract. Bonds matched to near-term spending neutralize that risk; stocks serve the long-term obligations. Viewed that way, a rising path is simply what asset–liability matching looks like over time.

What we do about it

We still use Monte Carlo. It is the best available picture of the range of outcomes. We just do not stop at the headline.

For households entering retirement, we test the plan against the thing that can actually break it: several years of poor returns and rising prices, starting now. Then we ask what would make the plan hold.

The answer is rarely a product. It is almost always three decisions.

The first is spending, and it is by far the most powerful. A few thousand dollars a year changes the outcome of a bad start more than any reasonable change to the portfolio. The second is where the stock allocation begins. Starting lower and rising over about ten years protects the years that count most. The third is flexibility: holding the next several years of withdrawals in high-quality short-term bonds so a decline never forces a sale, and being willing to trim discretionary spending for a year or two if one arrives.

None of this is free. A more careful start gives up some growth if markets are kind. For most of the people we work with, that is a modest price for what it protects. But it is a choice, and it should be made deliberately, with the tradeoff on the table, not discovered in the middle of a downturn.

Ninety-five percent is good news. It is worth knowing what the other five look like, and building a plan that survives them.

The companions to this piece are The Planning Problem, which takes up how time horizon shapes the balance between stocks and bonds, and Asset–Liability Alignment, which describes how we match capital to the spending it has to support.

A note on method

Figures for the newly retired couple come from simulation models using their plan’s spending, Social Security and tax projections, with assumed returns, volatility and inflation. The bad-start results isolate simulated futures that begin with a severe five-year decline. Holdings and household details are described generically and figures are rounded. Different assumptions, periods and models will produce different results.

References

Bengen, William P. 1994. "Determining Withdrawal Rates Using Historical Data." Journal of Financial Planning 7 (4): 171–180.

De Santis, Massimiliano. 2023. "Optimal Spending and Portfolio Rules to Protect Desired Spending in Retirement." Retirement Management Journal 12 (1): 44–53.

Fama, Eugene F., and Kenneth R. French. 2018. "Long-Horizon Returns." The Review of Asset Pricing Studies 8 (2): 232–252.

Fama, Eugene F., and Kenneth R. French. 2018. "Volatility Lessons." Financial Analysts Journal 74 (3).

Pellerin, Mathieu. 2021. "Researching Retirement: The Impact of Inflation, Interest Rates, and Market Risks." Dimensional Fund Advisors.

Pfau, Wade D., and Michael E. Kitces. 2014. "Reducing Retirement Risk with a Rising Equity Glide Path." Journal of Financial Planning 27 (1): 38–45.

Learn More About Us

Cardiff Park Advisors works with clients throughout the United States. We welcome the opportunity to discuss your financial goals and how we can help you reach them. You may reach us by emailing our principal at jgorlow@cardiffpark.com or calling our office at 760-635-7526.

For more information about Cardiff Park Advisors please review our brochure at https://adviserinfo.sec.gov/firm/summary/126752 or visit www.cardiffpark.com.

This paper is for educational purposes and is not investment advice, nor a recommendation to buy or sell any security. Figures are hypothetical and for illustration only; they come from simulation models, do not reflect any actual client’s results, and are not guarantees of future results. Actual results will vary and may differ materially. Research findings cited are those of their authors. Past performance does not guarantee future results. Cardiff Park Advisors is a registered investment adviser. This material does not constitute tax or legal advice.