The Cross‐Section of Asia‐Pacific Mortality Dynamics: Implications for Longevity Risk Sharing

Abstract

We study the dynamics of longevity risk across a subset of countries in the Asia‐Pacific (APAC) region. We use hand‐collected and existing data on age‐specific mortality rates from emerging and developed economies to understand how secular changes in mortality vary within and across APAC countries. We use our results to identify cross‐hedging opportunities among longevity risk exposures in the APAC region. We also introduce k‐forward contracts, which offer natural risk‐sharing opportunities to hedgers in different countries. We consider the example of Korea and Japan as a case study.

The Cross-Section of Asia-Pacific Mortality Dynamics: Implications for Longevity Risk Sharing

“…Finally, section 5 concludes. An online appendix (see Biffis et al, 2016b) provides a number of additional results, including a horse race among different mortality models, including GDFMs, which provide support for the use of the LL model in our dataset, as well as for the factor structure on which the model relies. 18…”

“…23 This aspect has been discussed extensively by demographers (see Park, 1998, for example, and references therein). 24 The online appendix (Biffis et al, 2016b) compares the goodness-of-fit of the LL model with alternative models. The use of an explanation ratio as performance metric shows that the LL model fits the historical data better than competing models.…”

“…In contrast, the negative value ofr 0,KOR of the countryspecific risk factor k(t, KOR) is consistent with the estimated downward trend. To do mortality forecasting, we simulate the realizations of factors K(t) and k(t, i) beyond the end of the sampling period, as discussed in the online appendix (see Biffis et al, 2016b).…”

“…On the methodological side, we develop our analysis of APAC mortality from the perspective of the LL model, which studies multiple populations jointly, and ensures that mortality forecasts for individual populations do not diverge in the long run (e.g., Hyndman et al, 2013). The use of the LL model is supported by a horse race between several competing models reported in Biffis et al (2016b). We apply the model to all countries in our sample simultaneously, and disentangle common mortality improvement factors from country-specific risk factors that show considerable variation in terms of trend and volatility.…”

Changes of Relation in Multi-Population Mortality Dependence: An Application of Threshold VECM

“…Originally proposed by Li and Lee (2005), coherent mortality forecasting assumes that the relative mortality rates of any two related populations are mean-reverting; hence, coherent mortality forecasts do not diverge indefinitely over time. The model of Li and Lee (2005) has been adopted by several researchers, including Biffis et al (2017), Li and Hardy (2011), and Milidonis et al (2011). Other examples of coherent models include the two-population model proposed by Cairns et al (2011), the gravity model by Dowd et al (2011), and the application of vector error correction model (VECM) by Zhou et al (2014).…”