Arriaga Decomposition
- Introduced by
- Eduardo E. Arriaga
- Year
- 1984
- Type
- Age-specific decomposition of a difference in life expectancy
- Family
- Standardization & decomposition
What does it do?
Arriaga decomposition is a demographic technique that breaks down the difference in life expectancy between two life tables — two countries, two time points, or two groups — into the contributions of mortality change at each age.
Introduced by Eduardo Arriaga in 1984, it tells the analyst not just that life expectancy rose or fell, but exactly which ages drove the change, separating the direct effect of mortality change within an age interval from the indirect effect of the extra survivors that change passes on to older ages.
The intuition behind the method
Life expectancy is a single summary built from survival at every age, so when it changes it is natural to ask, 'Where did the years come from?' Saving a child from dying does two things: it gives that child the years lived during childhood (the direct effect), and it produces a survivor who then goes on to live through all later ages (the indirect effect). Saving an 80-year-old, by contrast, has a large direct effect but almost no indirect effect because there are few later years to pass on. Arriaga decomposition makes this bookkeeping exact, attributing every fraction of a year of the life-expectancy change to the age at which the underlying mortality change occurred.
History
Eduardo E. Arriaga, a demographer at the U.S. Census Bureau, published 'Measuring and explaining the change in life expectancies' in Demography in 1984, giving demographers a practical, exactly additive way to attribute changes in life expectancy to mortality change at specific ages. The method built directly on the decomposition tradition begun by Kitagawa (1955) and complemented later continuous-time approaches such as Pollard's and Vaupel's. It became, and remains, the most widely used technique for the age and cause decomposition of life-expectancy differences, and is documented in standard references including Preston, Heuveline, and Guillot (2001).
When to use it
Use Arriaga decomposition when you want to explain a change or difference in life expectancy by identifying which ages contributed, for two life tables — across time, between countries, or between subgroups such as sexes or social classes. It is the standard tool for the age decomposition of life-expectancy gaps and is frequently combined with cause-of-death partitioning to attribute change to age-by-cause cells. Assumptions: two comparable, complete life tables exist with consistent age intervals, and the life-table relationships (the linear-in-the-interval survivorship convention) hold. Do NOT use it as a substitute for a causal model — the contributions are an exact arithmetic attribution, not an estimate of what intervention would change. For more than two populations or several interacting non-age factors, use the Das Gupta generalization instead.
Limitations and considerations
- Defined for comparing two life tables at a time; multi-population or multi-factor problems require the Das Gupta generalization.
- Sensitive to the life-table assumptions used to build Lₓ, especially the linearity convention within wide age intervals at very young or very old ages.
- Folding the interaction into the indirect term makes the split slightly asymmetric, so swapping the order of populations changes the within-component allocation.
- Like all decompositions it is descriptive accounting, not a causal estimate of the effect of changing mortality at a given age.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Provides an exact, age-specific decomposition whose contributions sum precisely to the total life-expectancy difference with no residual.
- Distinguishes the direct effect of mortality change within an age from its indirect downstream effect on later survival, a uniquely demographic insight.
- Requires only standard life-table outputs, which are routinely produced for most national populations.
- Extends naturally to age-by-cause decomposition, allowing attribution of life-expectancy change to specific causes at specific ages.
Applications
- Identifying which age groups drove a country's gain in life expectancy over a decade, such as the shift from infant-mortality gains to old-age gains.
- Decomposing the sex gap in life expectancy into the ages at which male and female mortality differ most.
- Attributing life-expectancy losses during crises (e.g., epidemics or economic shocks) to specific ages and, with cause data, specific causes.
- Comparing the age profile of mortality improvement across countries to study mortality convergence.
- Quantifying socioeconomic inequalities in survival by decomposing life-expectancy gaps between deprivation groups by age and cause.
How it works
How it works
The method begins from two complete life tables, one for each population or time point. The required columns are the survivorship function lₓ (the number surviving to exact age x out of the radix), the person-years lived in each age interval Lₓ, and the cumulative person-years Tₓ from which life expectancy eₓ = Tₓ / lₓ is obtained. The difference in life expectancy at birth, e₀² − e₀¹, is the quantity to be decomposed across ages.
Formula
l_x,\; L_x,\; T_x,\; e_x = \frac{T_x}{l_x}The direct effect for an age interval measures the change in person-years lived within that interval, attributable to the change in mortality inside the interval itself, scaled by the probability of surviving to the start of the interval in population 1. Intuitively, it captures the years added or lost directly because death rates within that specific age band changed, holding the chance of reaching the interval at the first population's level.
Formula
{}_{n}\!DE_x = \frac{l_x^{1}}{l_0^{1}}\left(\frac{{}_nL_x^{2}}{l_x^{2}} - \frac{{}_nL_x^{1}}{l_x^{1}}\right)The indirect (and interaction) effect captures the downstream consequence: a mortality change at a given age alters the number of people who survive to all later ages, and those additional or fewer survivors then contribute the person-years lived beyond the interval. Arriaga's formulation folds the pure indirect effect and the direct–indirect interaction into one term, weighted by the cumulative person-years remaining beyond the interval in population 2, so the accounting stays exact.
Formula
{}_{n}\!I_x = \frac{T_{x+n}^{2}}{l_0^{1}}\left(\frac{l_x^{1}}{l_x^{2}} - \frac{l_{x+n}^{1}}{l_{x+n}^{2}}\right)The total contribution of each age interval is the sum of its direct and indirect/interaction effects. By construction these age-specific contributions sum exactly to the overall difference in life expectancy at birth, with no residual. Reading the contributions across ages reveals the age profile of the change — for instance, whether rising life expectancy was driven by falling infant mortality or by improvements at older ages.
Formula
{}_{n}\Delta_x = {}_{n}\!DE_x + {}_{n}\!I_x,\qquad e_0^{2} - e_0^{1} = \sum_x {}_{n}\Delta_xCommon pitfalls
- Mismatched age groupings between the two life tables, which break the alignment of the direct and indirect terms.
- Interpreting a large contribution at old ages as a target for intervention, when it reflects arithmetic attribution rather than causal leverage.
- Forgetting that reversing which life table is labelled 1 versus 2 flips the sign of every contribution; the direction must be stated.
- Decomposing life-expectancy change at an open-ended top age interval without care, since the contribution there depends heavily on assumptions about mortality above the interval.
Frequently asked questions
What is the difference between the direct and indirect effects?
The direct effect is the change in years lived within an age interval caused by the mortality change inside that interval itself. The indirect effect is the downstream consequence: changing mortality at one age alters how many people survive to all later ages, and those survivors contribute the years lived beyond the interval. Saving a young life has a large indirect effect (many later years), while saving an old life has almost none.
How does Arriaga decomposition relate to Kitagawa decomposition?
Both belong to the demographic decomposition family that began with Kitagawa (1955). Kitagawa splits a difference between two crude rates into rate and composition components. Arriaga adapts the same exact-additivity philosophy to a more complex, nonlinear summary — life expectancy — by exploiting life-table structure to allocate the change across ages, which a simple two-component rate decomposition cannot do.
Can it decompose life-expectancy change by cause of death as well as by age?
Yes. A standard extension distributes each age interval's contribution across causes of death in proportion to the change in age- and cause-specific mortality rates. This produces an age-by-cause matrix of contributions that still sums exactly to the total life-expectancy change, and it is one of the most common applied uses of the method in epidemiology.
Related methods
Sources
Arriaga, E. E. (1984). Measuring and explaining the change in life expectancies. Demography, 21(1), 83–96.
Open DOI referenceCopied.Preston, S. H., Heuveline, P., & Guillot, M. (2001). Demography: Measuring and Modeling Population Processes. Blackwell.
ISBN: 9781557864512Copied.