# Understanding Long-term Mortality Dynamics and Improving Old-age Mortality Forecasts

> **NIH NIH R21** · STANFORD UNIVERSITY · 2020 · $195,625

## Abstract

Project Summary
The broad goals of this proposal are to increase understanding of the dynamics of human mortality
change and to develop improved methods to forecast national and regional mortality. Better forecasts
are central to the management and pricing of health and pension programs (including Social Security
and Medicare in the US and elsewhere) and to the insurance industry (for life insurance, annuities, and
longevity bonds). Most current stochastic mortality forecasts are based on work by Lee and Carter (LC)
and technical and actuarial modifications of the LC. It is well known that mortality change follows a
time-varying dynamic: in the long run (multi-decadal periods) there has been a shift from younger to
older ages in the rate and significance of change; in the medium term (decade based) the LC method
based on a dominant time signal and correlated age-change captures dynamics well; and finally in the
short term (year based) particular age-related forces (disease epidemics, the opioid crisis) can cause
notable year-on-year variation. In the LC methods, and here, these short term changes are treated as
stochastic fluctuations and are analyzed in terms of their statistical properties.
This project aims to improve analytical understanding of mortality dynamics in the medium and long
terms. Preliminary work (by the proposers) has focused on the shape of old-age mortality, accurately
described by the dynamics of percentiles of the human death distribution. Past the 25th, these
percentiles have advanced at a nearly constant speed over the last five decades. The first specific aim is
to extend this work to analyze trends in the shape and level of old-age mortality using percentiles. The
preliminary results on old-age mortality suggest the analyses and extensions of LC: improvements to
the forecast trend and error; analyses of the sensitivity of LC to noise at old ages and to the choice of
base period. This project aims to analyze mortality change over long (many decades) time periods via
multiple-timescale methods. In order to successfully apply LC methods more widely, this project will
develop and test relational methods to produce life tables for countries with limited data in some years.
Finally, this project aims to examine the nature of period relationships between income and mortality
that are consistent with aggregate change.

## Key facts

- **NIH application ID:** 9843626
- **Project number:** 5R21AG061639-02
- **Recipient organization:** STANFORD UNIVERSITY
- **Principal Investigator:** SHRIPAD D. TULJAPURKAR
- **Activity code:** R21 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2020
- **Award amount:** $195,625
- **Award type:** 5
- **Project period:** 2019-01-01 → 2021-11-30

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/9843626

## Citation

> US National Institutes of Health, RePORTER application 9843626, Understanding Long-term Mortality Dynamics and Improving Old-age Mortality Forecasts (5R21AG061639-02). Retrieved via AI Analytics 2026-05-23 from https://api.ai-analytics.org/grant/nih/9843626. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*