John R. Wilmoth

Overview

Demographer at the University of California, Berkeley and Director of the Population Division at the United Nations. Works on mortality measurement, forecasting, and the formal demography of life expectancy. A principal contributor to the Human Mortality Database (HMD).

Key Contributions / Features

• TP2003 methodology audit (Wilmoth 2005 GENUS): As a member of the 2003 Technical Panel on Assumptions and Methods, audited OACT's 70-parameter cause-of-death projection methodology. Showed that TR2003's intermediate scenario selects the longevity-pessimistic option at each of five sequential methodological decision points — a pattern inconsistent with a genuine central forecast. Documented that U.S. old-age mortality stagnation in the 1990s was an outlier among 15 high-income countries (11/15 showed accelerating decline), supporting a recovery projection rather than anchoring to the anomalous decade. Key benchmarks: TR2003 $e_0$ 2070 = 82.8 yr; TP2003 = 84.4 yr; TP1999 = 85.2 yr; LTAB (TR2003) = −1.92 pp. See SSA Mortality Forecasting and Long-Term Actuarial Balance.
• Period vs. cohort lifespan taxonomy (Wilmoth 2005): Formalized a six-measure taxonomy distinguishing period e₀, CAL, MAD, the Bongaarts-Feeney approximation, expected cohort LE, and achieved cohort LE. Showed that tempo-adjusted e₀ (Bongaarts-Feeney generalized) is mathematically identical to CAL — a population-dynamics measure — not a corrected period life expectancy. See Period vs. Cohort Life Expectancy and Cross-Sectional Average Length of Life (CAL).
• Rejection of tempo bias in mortality (Wilmoth 2005): Demonstrated that the quantum/tempo distortion argument from fertility (Bongaarts-Feeney 1998) does not apply to $e_0$, because $e_0$ is already a timing measure, not a count measure. The analogy to TFR tempo adjustment does not hold. See Tempo Effects in Mortality.
• Linear shift model (Wilmoth 2005): Introduced a class of "percentile models" for mortality trends specifying that each percentile of the age-at-death distribution shifts linearly over calendar time. Derived the exact period-to-cohort speed conversion ($s = r/(1-r)$) and connected period e₀ to cohort e₀ analytically. Validated on Sweden 1751–2003 (HMD data). See Linear Shift Model and Period Mortality.
• Human Mortality Database: Co-developer and principal steward of the HMD — the international repository of high-quality historical mortality data used across most formal demography research on longevity trends.

Related

• Human Mortality Database
• SSA Mortality Forecasting
• Long-Term Actuarial Balance
• Office of the Chief Actuary
• Period vs. Cohort Life Expectancy
• Cross-Sectional Average Length of Life (CAL)
• Tempo Effects in Mortality
• Linear Shift Model
• Period Mortality
• Ronald Lee

Sources

• Wilmoth 2005 — Some Methodological Issues in Mortality Projection
• Wilmoth 2005 — On the Relationship between Period and Cohort Mortality
• Bongaarts 2004 — Long-Range Trends in Adult Mortality Models and Projection Methods
• Carter and Prskawetz 2001 — Examining Structural Shifts in Mortality Using the Lee-Carter Method