pano_cae_006

CAE Research on Longevity Risk

Overview

The CAE grant “New Trends in Longevity” started in 2015 and is currently ongoing.

The problems of “mortality” and “longevity risk,” i.e. the risk that realized future aggregate mortality trends deviate from current assumptions, emerged as a core area in actuarial science in the mid 2000s. Since then, a notable amount of research has been produced. Notwithstanding these developments, conventional modeling approaches are shaped by exploiting structural similarities between longevity and financial risks. This reliance on financial models may not come as a surprise given the researchers’ expertise. Actuarial and risk management scholars interested in the new field of longevity risk were typically trained in dealing with financial risks, and thus are poised to make use of familiar ideas and their conventional toolbox.

The basis of our research is that there are fundamental differences between modeling and managing financial and mortality risk. While for financial risk the focus is on modeling short-term volatility, for mortality risk the most relevant risk appears to be a change in the long-term development—or trend—of mortality rates, which could have disastrous consequences in view of the sufficiency of reserves. Our objective is to develop a research program that focuses on modeling mortality trends rather than the variability in mortality rates—and the inference for and the evaluation of resulting models. While this may sound rather straightforward at first glance, we believe that this area is largely unexplored and that this simple change in perspective could yield significant advances in understanding, analyzing, and mitigating mortality/longevity risk—although it is also associated with considerable technical challenges requiring advanced statistical and probabilistic methods.


Specific Projects

  • Improving and Extending Lee-Carter: We illustrate some basic problems with the usual inference approach within the Lee Carter model, and propose improved approaches.
  • Forecasting Mortality Forecasts: Instead of building models by analyzing mortality rates, we perform analyses on time series of mortality forecasts. Resulting models are better suited to appraise the risk in mortality projections.
  • Stochastic Trends: We consider various stochastic trend models applied to stochastic mortality and evaluate them in view of forecasting.
  • Cointegration and Longevity Hedging: We revisit the question of cointegration between the mortality of multiple countries as well as cointegration with mortality and economic series. We explore whether and how the results can be used for hedging longevity risk.

Primarily Involved Researchers

  • Daniel Bauer (Principal Investigator) is New York Life Associate Professor of Risk Management and Insurance at Georgia State University (CAE)
  • Liang Peng is the Thomas P. Bowles Chair of Actuarial Science at Georgia State University (CAE)

Publications

  • Leng, Xuan and Liang Peng (2015). “Inference pitfalls in Lee-Carter model for forecasting mortality”, submitted
  • Leng, Xuan and Liang Peng (2015). “Test for unit root in Lee-Carter mortality model”, submitted
  • Zhu, Nan and Daniel Bauer (2015). “Modeling and Forecasting Mortality Projections”, working paper

Conference and Seminar Presentations

  • July 2015: “Mortality and Longevity Risk: Methods, Models and Management”, presentation by Daniel Bauer at the 2015 CEAR-Huebner Summer Risk Institute