What your retirement calculator is actually measuring

There is a number most people carry in their heads: the return their savings will earn. Seven percent. Ten percent. Sometimes more. They got it from a calculator, a brochure, or a conversation with an advisor. The calculator ran the math and showed them a reassuring pile at the end of 30 years.

That number is nominal. It counts units of currency. It does not count what those units will buy.

The gap between the two is not a rounding error. It is the entire argument.

Since 1928, U.S. stocks have averaged roughly 9.9% nominal annual return. After adjusting for inflation, that figure falls to 6–7% in real terms.1 Looking at a longer modern window — the S&P 500 from 1957 through 2026 — the nominal annualized return with dividends reinvested is 10.51%, while the inflation-adjusted real return is 6.64%.2

Those 3–4 percentage points are not a footnote. Compounded over 30 years, a $100,000 portfolio at 10% nominal grows to roughly $1.74 million. At 6.6% real, the same portfolio in purchasing-power terms reaches about $670,000. The gap between them is not market variance; it is inflation, silently converting nominal gains into a smaller pile of actual spending power.

Most retirement calculators default to nominal assumptions — or apply a generic inflation adjustment only at the very end, reducing the final number by a flat percentage. That late-stage discount misses the compounding structure of the problem: inflation does not wait until year 30. It takes its share every single year, on every reinvested dividend, on every added contribution, on every month's balance.

A common mental model treats inflation as a one-time haircut: earn 10%, subtract 3% inflation, net 7%. The math is more corrosive than that.

The correct formula for a single year is not 10% − 3% = 7%. It is (1.10 ÷ 1.03) − 1 ≈ 6.8%. That 0.2 percentage point difference per year compounds across decades. Over 30 years, the divergence between the arithmetic and geometric treatments can run to tens of thousands of dollars on a mid-size retirement account.

The deeper problem is what happens during high-inflation periods. Between 1965 and 1982, U.S. inflation averaged above 6% annually. The nominal S&P 500 did rise during parts of that era, but in real purchasing-power terms the index spent roughly three decades from 1965 to 1995 approximately flat — a period the humbledollar.com analysis calls "three lost decades" in real buying power, a figure rarely cited because it only appears when you run the inflation-adjusted chart.1 If your retirement savings window happened to run through any subset of that span, the nominal chart was a fiction.

The IMF's 2024 paper on household responses to inflation is direct on the redistribution mechanism: unexpected inflation erodes the real value of savings, transferring wealth from net savers (who hold fixed-nominal assets) to net borrowers (whose liabilities shrink in real terms).3 Wage earners who are also savers — the group retirement calculators are primarily aimed at — sit on the losing side of that transfer.

Even if you accept a 6–7% real return assumption, the calculator has a second structural flaw. It assumes you will earn the average every year, in steady installments. You will not. Markets deliver lumpy, volatile, non-sequential returns — and the order those returns arrive in, particularly relative to when you start withdrawing, has a catastrophic effect on outcomes.

Charles Schwab's analysis illustrates it with a clean example: two investors each start with a $1 million portfolio and withdraw $50,000 per year, inflation-adjusted. Both experience a 15% annual loss for two consecutive years. The only difference is timing. Investor 1 takes the loss in years 1 and 2 of retirement. Their portfolio runs out in roughly 18 years. Investor 2 takes the same-magnitude loss in years 10 and 11. After 18 years, they still have nearly $400,000 left.4

Same average return. Same withdrawal rate. A $400,000 swing.

The academic literature on this point is substantial. The CFA Institute's Financial Analysts Journal confirmed in 2017 that the sequence of returns can "dramatically affect retirement income, for better or for worse," independent of the average return achieved.5 The mechanism is straightforward: when you sell assets to fund withdrawals during a downturn, you lock in losses at the worst possible price. Those shares are no longer available to recover when the market rebounds.

A retirement calculator using a smooth average return does not model this. It cannot, without additional inputs — Monte Carlo simulation, historical sequence testing, or at minimum a sensitivity analysis across different entry-year scenarios. Most consumer-facing tools offer none of these. They take your target return, apply it uniformly, and produce a number that would only be correct if markets moved in straight lines.

William Bernstein's work at Efficient Frontier applied Monte Carlo simulation to the 1966–1995 historical period for a 100% stock portfolio at a 7% real return assumption. On a 5% withdrawal rate, the model showed a portfolio failure rate of about 6.6%. But using actual historical sequence data from that specific 30-year window, even a 5% withdrawal rate would have depleted the portfolio before the period ended — because the early years of that window happened to be the brutal inflation decade of the late 1960s and 1970s.6 The average was survivable. The sequence was not.

The core problem is not that markets perform badly on average. Over long horizons, equity returns have been real and substantial. The problem is that retirement calculators present average returns as if they were a contract — a fixed rate your savings will earn regardless of when you need to access them, regardless of what inflation does in the intervening decades, regardless of the sequencing of the market's actual path.

They measure the wrong thing. They measure nominal returns (or nominal-ish returns with superficial inflation adjustments). They measure expected averages rather than probable ranges. They treat future purchasing power as a mathematical consequence of past averages.

Real purchasing power is something else. It is the product of: the actual path of inflation across your accumulation years, the actual sequence of market returns in your withdrawal phase, the interaction between those two — since high inflation often coincides with poor real equity returns — and your ability to survive a bad early sequence without permanently depleting capital.

None of those appear on the calculator's output screen. The screen shows a number. The number has a compelling precision. The precision is an artifact of the math, not a description of reality.

If you are building a retirement model that will govern 30 years of decisions, the least honest thing you can do is run it on a smoothed nominal return assumption and treat the output as a plan. The most honest thing you can do is run it at 4–5% real (which is what 6–7% real looks like after you account for the gap between arithmetic and geometric compounding, fees, and tax drag), test it against adverse sequences, and then ask yourself whether the range of outcomes on the pessimistic end is something you can absorb.

That answer will be different from what the calculator told you.