What Moves the Answer — and What Doesn't

Below is the sensitivity analysis behind the headline results: what actually moved the July 2026 model's DC-vs-SS ratio, and what turned out to be second-order. All figures are from the corrected engine (build August 2026).

Correction (July 2026): Near-retirement cohorts (workers past their earnings peak) previously had their Social Security benefits overstated by roughly 2×. The career-scaling anchor inverted the age-earnings profile at the worker's current age, where the profile's late-career floor — an artifact of Census averaging in retired workers — deflated the anchor and inflated the whole career. The anchor is now floored so a still-employed worker's income is treated as approximately peak earning power; the born-1960/$48K case moves from ~$3,100/mo to ~$1,730/mo, matching the SSA Quick Calculator ($1,558) within ±15%. DC/SS ratios rise for older cohorts under this correction (the SS leg falls while the DC leg is unchanged) — that is the fix working, not a new bug. See Corrections & Retractions.

Retirement age — later was better, in every tier

Working longer widened the DC advantage in every tier tested. (An earlier version of this study reported a tier-dependent reversal here; on the corrected engine it did not reproduce, and that claim was withdrawn.)

Life expectancy — the long-life assumption favored Social Security

A natural objection to any DC comparison is that a long horizon flatters compounding. The opposite was true here. Social Security pays for as long as you live; the portfolio's bequest is fixed at death. Extending life expectancy therefore helped Social Security. The base case uses LE 95 — well above actuarial average, and the most Social-Security-favorable choice available. The headline numbers are conservative on this axis by construction.

Payroll basis — the disability carve-out

The base case credits the DC side with only the retirement portion of the payroll tax, excluding the 1.8% that funds Disability Insurance — because the DC model provides no disability benefit. Including the full 12.4% would raise every DC result by about 17%. We do not do this; the conservative choice is the honest one.

Wage growth — a material assumption

The DC/SS ratio is sensitive to the wage growth assumption. Higher wage growth increases Social Security benefits (through faster bend-point indexing) while reducing the DC portfolio's compounding advantage (later contributions have less time to grow). The two legs move in the same direction, not opposite, so the ratio shifts materially with the assumption. The headline ratio uses the SSA intermediate NAWI growth rate (3.6%); you can adjust this on the Assumptions tab to see the sensitivity.

Correction (July 2026): an earlier version of this section reported that wage growth "cancels out" of the ratio, citing identical 5.42× results at 3.6% and 4.5% NAWI. That was a defect in the sensitivity run, not a property of the model. See Corrections & Retractions.

Demographics — second-order

Holding peak income fixed at $50K and varying one factor at a time, the entire demographic space spanned 7.52× to 9.15× — roughly ±10%. Gender, education, and marital status substantially change benefit levels (a college earnings curve nearly doubles the SS benefit at the same peak income), but they move both sides of the comparison together and therefore largely cancel in the ratio. Against an investment-tier lever spanning 0.33× to 184×, demographics were second-order by two orders of magnitude.

Birth cohort — the formula is fair; the market history is not

These differ — but not because Social Security treats generations differently. On the corrected engine, replacement rates varied by ≤ 2.4 percentage points across birth years 1960–2000 at every income level: the benefit formula is cohort-invariant. What differs is the slice of actual market history each cohort's career happens to span. The 1980 cohort's accumulation window captured a historic equity run; the 2000 cohort's is mostly projected.

This is a fact about markets, not about the policy — and it is precisely why the Monte Carlo measure, which resamples that history, is the more reliable guide across cohorts.

A note on our earlier generational study. An earlier standalone analysis reported that "the advantage generally rises for younger generations" and showed 100% Monte Carlo wins across generations, using a 4.5% wage-growth assumption. Those figures were produced on the pre-audit engine and were an artifact of two of the defects we later fixed. That claim is formally retracted. The cohort-invariant result above is what the corrected engine supports as of July 2026.

About these figures — please read. Every number here is an output of the Is Social Security Worth It? simulator, read from the August 2026 build. These are model estimates, not predictions, advice, or guarantees, describing the model on that date, not a promise about future versions. See Corrections & Retractions; figures may change in later builds. Past performance does not guarantee future results. Your actual benefit is set by the SSA at ssa.gov/myaccount. Per-cell data: DCvsSSv3substudies.csv.