Munnell and Soto 2005 — Sorting Out Social Security Replacement Rates

Summary

This Center for Retirement Research (CRR) policy brief resolves an apparent puzzle: the Social Security Administration's (SSA) policy model replacement rate for the median earner (42%) and the actual average replacement rate for new retirees (also ~42%) produce the same number despite using entirely different methods. Munnell and Soto show this agreement is pure coincidence — three methodological differences happen to offset each other. The brief clarifies the distinction between career-average earnings and Average Indexed Monthly Earnings (AIME) as replacement rate denominators, and documents the impact of earnings gaps and early claiming on actual versus hypothetical replacement rates.

Key Claims

• The policy model 42% and the actual 42% are produced by three offsetting errors that happen to cancel: (1) AIME is lower than career-average earnings, raising the rate; (2) actual workers have earnings gaps that further lower their AIMEs, raising the rate; (3) most workers claim before the Normal Retirement Age (NRA) with actuarial reductions, lowering the rate.
• If policy model benefits were expressed as a percent of AIME (rather than career-average earnings), the medium earner's replacement rate would be 48%, not 42%.
• Accounting for actual zero-earnings years (men: 6 on average; women: 13), the average worker's AIME in 2003 was $25,000 rather than the $28,950 the policy model assumes, raising the expected replacement rate to ~50%.
• 74% of men and 78% of women claim Social Security before age 65, triggering actuarial reductions that push actual rates back down toward 42%.
• Low earners and women have systematically higher actual replacement rates than the policy model implies, because the progressive formula amplifies the effect of lower AIMEs.

Concepts Introduced or Extended

• DI Replacement Rate

Entities Mentioned

• Alicia H. Munnell
• Mauricio Soto
• Social Security Administration

Quotes

"The two approaches happen to produce the same number for the median earner. However, this result is strictly a coincidence as the two methods involve different measures of pre-retirement earnings, different employment patterns, and different retirement ages."

"That the two numbers turn out to be the same is enough to drive a policy wonk crazy!"

My Take

A concise methodological note that every researcher using Social Security replacement rates should read. The key insight — that AIME and career-average earnings are different denominators producing systematically different replacement rate measures — is easy to miss and consequential for cross-study comparisons. The brief is Old-Age and Survivors Insurance (OASI)-focused (retirement, not disability insurance [DI]), but the AIME machinery is identical for DI replacement rate calculations, so the measurement clarity applies directly. The three-factor coincidence is a cautionary tale about using any single replacement rate number at face value without understanding what denominator it uses.