Skip to main content

Inference for the Lee-Carter Model With An AR(2) Process


Researchers in studying longevity risk often employ the Lee-Carter model with a unit root AR(1) process for unobserved mortality indexes. When one models the mortality index by a stationary AR(1) process, the widely used two-step inference in Lee and Carter (1992) is inconsistent. Some mortality datasets reject the unit root hypothesis. This paper develops consistent statistical inferences for a modified Lee-Carter model using an AR(2) process to model unobserved mortality indexes. It also provides a simulation study to examine their finite sample performance before applying them to the US mortality rates.

This is a preview of subscription content, access via your institution.


  1. 1.


  1. Biffis E, Lin Y, Milidonis A (2017) The cross-section of Asia-Pacific mortality dynamics: implications for longevity risk sharing. J Risk Insur 84:515–532

    Article  Google Scholar 

  2. Enchev V, Kleinow T, Cairns AJG (2017) Multi-population mortality models: fitting, forecasting and comparisons. Scand Actuar J 4:319–342

    MathSciNet  Article  Google Scholar 

  3. Girosi F, King G (2006) Demographic Forecasting. Cambridge University Press, Cambridge

    Google Scholar 

  4. Hall P, Heyde C (1980) Martingale Limit Theory and Its Applications. Academic Press

  5. Kwok KY, Chiu MC, Wong HY (2016) Demand for longevity securities under relative performance concerns: stochastic differential games with cointegration. Insurance Mathematics and Economics 71:353–366

    MathSciNet  Article  Google Scholar 

  6. Lee RD, Carter LR (1992) Modeling and forecasting U.S. mortality. J Am Stat Assoc 87:659–671

    MATH  Google Scholar 

  7. Leng X, Peng Liang (2016) Inference pitfalls in Lee-Carter model for forecasting mortality. Insurance Mathematics and Economics 70:58–65

    MathSciNet  Article  Google Scholar 

  8. Leng X, Peng L (2017) Testing for unit root in Lee-Carter mortality model. ASTIN Bulletin 47:715–735

    MathSciNet  Article  Google Scholar 

  9. Liu Q, Ling C, Li D, Peng L (2019) Bias Corrected Inference For Lee-Carter Mortality Model. ASTIN Bulletin 49:433–455

    MathSciNet  Article  Google Scholar 

  10. Liu Q, Ling C, Peng L (2019) Statistical inference for Lee-Carter mortality model and corresponding forecasts. North American Actuarial Journal 23(3):335–363

    MathSciNet  Article  Google Scholar 

  11. Phillips PCB (1987) Towards a unified asymptotic theory for autoregression. Biometrika 74:535–547

    MathSciNet  Article  Google Scholar 

  12. Staudenmayer J, Buonaccorsi JP (2005) Measurement error in linear autoregressive models. J Am Stat Assoc 100:841–852

    MathSciNet  Article  Google Scholar 

  13. Staudenmayer J, Buonaccorsi JP (2006) Measurement error in a random walk model with applications to population dynamics. Biometrics 62(4):1178–1189

    MathSciNet  Article  Google Scholar 

  14. Wong TW, Chiu MC, Wong HY (2017) Managing mortality risk with longevity bonds when mortality rates are cointegrated. J Risk Insur 84:987–1023

    Article  Google Scholar 

Download references


We thank two reviewers for their helpful comments, which improve the presentation. Li's research was partially supported by the National Nature Science Foundations of China grant 11971115. Liu's research was partly supported by NSF of Jiangxi Province (No. 20192BAB201005), NSF project of Jiangxi Provincial Education Department (No. GJJ190261), and China Post doctoral Science Foundation (No. 2020M671961). Peng's research was partly supported by the Simons Foundation and the NSF grant, DMS-2012448.

Author information



Corresponding author

Correspondence to Liang Peng.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Li, D., Ling, C., Liu, Q. et al. Inference for the Lee-Carter Model With An AR(2) Process. Methodol Comput Appl Probab (2021).

Download citation


  • AR(2) model
  • Lee-Carter model
  • Mortality rates
  • Unit root