Heligman-Pollard Model
Also known as: Heligman-Pollard Mortality Law, Eight-Parameter Mortality Model, HP Mortality Model, Heligman-Pollard Ölümlülük Modeli
The Heligman-Pollard model is an eight-parameter parametric law that describes the age pattern of mortality across the entire human lifespan in a single equation. Introduced by Larry Heligman and John Pollard in 1980, it represents the odds of dying at each age as the sum of three additive components — a rapidly declining childhood term, a young-adult accident hump, and an exponentially rising senescent term — capturing the full characteristic shape of the mortality curve from birth to old age.
Key highlights
- Describes the entire lifespan mortality curve — childhood, accident hump, and senescence — in one coherent equation.
- Each component and parameter carries a clear demographic interpretation, aiding comparison across populations.
- Produces a smooth, graduated life table from noisy or fragmentary age-specific data.
- Captures the young-adult accident hump that purely senescent laws such as Gompertz-Makeham omit.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use the Heligman-Pollard model when you need a single smooth parametric law that fits mortality across the whole lifespan, including infant and child mortality and the young-adult accident hump that simpler senescent laws miss. It is well suited to graduating complete national life tables, summarizing an age pattern with interpretable component parameters, interpolating and extrapolating fragmentary data, and comparing mortality structure across populations. The model assumes mortality decomposes cleanly into the three named components in their specified functional forms. It can be difficult to fit — the eight parameters are correlated and the optimization is sensitive to starting values — and it may overfit small populations or fail when an accident hump is absent or when very-old-age deceleration matters; in those cases relational or factor models like Brass or Lee-Carter, or oldest-old logistic adjustments, may be preferable.
Strengths & limitations
- Describes the entire lifespan mortality curve — childhood, accident hump, and senescence — in one coherent equation.
- Each component and parameter carries a clear demographic interpretation, aiding comparison across populations.
- Produces a smooth, graduated life table from noisy or fragmentary age-specific data.
- Captures the young-adult accident hump that purely senescent laws such as Gompertz-Makeham omit.
- Estimation is notoriously delicate: eight correlated parameters make the nonlinear fit sensitive to starting values and prone to non-convergence.
- Can overfit small populations, producing unstable or implausible component parameters.
- The fixed functional forms may fit poorly where an accident hump is weak or absent, or where oldest-old mortality decelerates.
- Lacks a time dimension, so it describes a cross-sectional age pattern rather than forecasting mortality change.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
How does the Heligman-Pollard model differ from Gompertz-Makeham?
Gompertz-Makeham describes only adult mortality as a constant background plus an exponential senescent rise, and it ignores infancy, childhood, and the young-adult accident hump. Heligman-Pollard keeps a Gompertz senescent term but adds two more components — a steeply declining childhood term and a lognormal accident-hump term — so it can fit the entire lifespan in one equation. The price is more parameters and a harder, less stable estimation problem.
Why are there eight parameters and why is fitting difficult?
Three for the childhood decline (A, B, C), three for the accident hump (D, E, F), and two for senescence (G, H) give eight in total. The terms overlap in age and their parameters are correlated, so the nonlinear optimization has a flat, multimodal objective surface that is highly sensitive to starting values. Good initial estimates, sometimes obtained by fitting components sequentially, are usually needed for convergence.
Does the model work at the oldest ages?
The senescent term is a pure Gompertz exponential, which keeps accelerating and tends to overpredict mortality at the oldest-old ages where empirical rates decelerate. Variants of the model replace or modify the G·Hˣ term with logistic or other decelerating forms to better fit centenarian mortality, and analysts focused on extreme ages should prefer those adjustments.
Sources
- 1.Heligman, L., & Pollard, J. H. (1980). The age pattern of mortality. Journal of the Institute of Actuaries, 107(1), 49–80.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 22). Heligman-Pollard Model. ScholarGate. https://scholargate.app/demography/heligman-pollard-model