Roughly 40-50% of Americans nearing retirement are on track to fall short of what living costs will require — tens of millions of people

Summary

Depending on how conservatively "falling short" is defined, somewhere between roughly 39% and 70%+ of American households approaching retirement are on a trajectory that will not let them maintain their pre-retirement standard of living once they stop working — among people currently aged 55-64 (about 41 million Americans), that is somewhere on the order of 16 to 29 million people. Among households that do fall short, the average lifetime savings gap for those on the verge of retirement runs from about $19,000 per person for a married couple up to roughly $63,000 for a single woman in present-value terms — and climbs to $71,000-$105,000 per person once you look only at the households actually falling short rather than averaging in everyone who is fine. AskAmerica's own tables carry no retirement-savings, account-balance, or income-replacement data, so every number below is drawn directly from the Federal Reserve's Survey of Consumer Finances (SCF) as modeled by the Center for Retirement Research at Boston College (CRR), the Employee Benefit Research Institute (EBRI), and Vanguard.

How far short: two different, both defensible, measurement approaches

The National Retirement Risk Index (NRRI), built by CRR from the triennial Fed Survey of Consumer Finances, projects each household's retirement income (Social Security, pensions, and financial/housing assets converted to an annuity) as a share of pre-retirement earnings, then compares it to the replacement rate a household needs to keep its living standard. A household whose projected rate falls more than 10 percentage points short of its target is counted "at risk." Using the 2022 SCF (the most recent available, published by CRR in February 2024), 39% of working-age households (ages 30-59) are at risk — an improvement from 47% in the 2019 survey, driven almost entirely by a 22% real run-up in home prices between 2019 and 2022 plus pandemic-era savings and stock-market gains, not by any underlying change in retirement behavior.

That 39% figure has one major caveat that matters directly for "how far short": the NRRI assumes every household converts its home equity into an inflation-indexed annuity via a reverse mortgage at retirement. Very few actually do. Vanguard's 2023 Retirement Outlook study found that once home equity is excluded, at least 70% of households would fall short of maintaining their pre-retirement standard of living — CRR itself cites this as the more realistic floor. That is the range this report leads with: roughly 2 in 5 (with unrealistic reverse-mortgaging assumed) to at least 7 in 10 (without it) of the households approaching retirement.

How far short in dollars

EBRI's Retirement Security Projection Model (RSPM) answers the dollar-amount version of the question directly: for each household it simulates a distribution of retirement lifepaths against basic living expenses plus uninsured health and long-term-care costs, and reports the present-value shortfall. For households "on the verge of retirement" (EBRI's Early Baby Boomer cohort, i.e., those in their early-to-mid 60s), the average per-individual shortfall is:

• $19,304 for married households
• $33,778 for single men
• $62,734 for single women (women's greater longevity and lower average lifetime earnings both raise the figure)

Those averages blend in the households that are NOT projected to run short at all. Restricting the calculation to only the households actually simulated to fall short — the more useful figure for "how far short are the ones who fall short" — the average present-value deficit rises to $71,299 (married), $93,576 (single men), and $104,821 (single women) per individual. In EBRI's earlier Retirement Readiness Rating, roughly 47% of this Early Boomer cohort was classified as at risk in the first place — broadly consistent with the CRR figures above, though that particular age-cohort split is an older (2010-era) EBRI estimate and has not been re-published at that granularity since.

For scale, EBRI's RSPM has separately put the aggregate national retirement savings deficit (summed present value across all U.S. households) at roughly $3.5-3.8 trillion in recent estimates (varying by year and dollar vintage), down modestly from a 2019 estimate of $3.83 trillion for the same reasons the NRRI improved — higher home and stock values, not more savings behavior.

For how many people

Americans aged 55-64 — the age band most naturally described as "approaching retirement" — number about 41.2 million as of 2024 (Census/OECD working-age population data). Applying the 39% NRRI at-risk rate to that population implies roughly 16 million people in households on track to fall short even under the NRRI's home-equity-annuitization assumption; applying Vanguard's 70%+ estimate (excluding home equity) implies close to 29 million people. Both are approximations — the NRRI's working sample runs ages 30-59, not 55-64 specifically, and the exact at-risk share for the 55-64 band alone was not available in a recent published breakdown; but every source consulted agrees the true number is closer to "most" than to "a minority" once you stop crediting the home-equity annuity almost nobody actually buys.

What the AskAmerica data corpus could and could not answer

A catalog search of every AskAmerica schema (census, econ, fiscal, sec, health, housing, and the rest) returned zero tables matching "retirement," "401(k)," or household net worth by age — the corpus has no Survey of Consumer Finances ingestion and no retirement-account, pension, or replacement-rate data at all. This is a genuine coverage gap, not a query-construction failure: it was confirmed via search_catalog before any external research began. A find_recipe check for this question's method (applying an at-risk rate to a population count) returned no matching recipe in the catalog, consistent with this being a literature-driven answer rather than a corpus computation. Every figure in this report therefore comes from primary sources fetched directly — CRR's own NRRI issue briefs (crr.bc.edu) and EBRI's own issue briefs (ebri.org) — rather than from the connector. The only AskAmerica-adjacent figure used is the population scale (Americans aged 55-64), and even that was sourced externally (Census/OECD via FRED) after AskAmerica's census.acs_age table proved to only carry broad 18-64/65+ bands, not the specific 55-64 cut this question needed.