A DYNAMIC FACTOR APPROACH TO MORTALITY MODELING DECLAN FRENCH † , COLIN O’HARE † ABSTRACT. Longevity risk has become one of the major risks facing the insurance and pensions markets globally. The trade in longevity risk is underpinned by accurate forecasting of mortality rates. Using techniques from macroeconomic forecasting we propose a dynamic factor model of mortality that fits and forecasts mortality rates parsimoniously. We compare the forecasting quality of this model and of existing models and find that the dynamic factor model generally provides superior forecasts when applied to international mortality data. We also show that ex- isting multifactorial models have superior fit but their forecasting performance worsens as more factors are added. The dynamic factor approach used here can potentially be further improved upon by applying an appropriate stopping rule for the number of static and dynamic factors. JEL Classification: C51, C52, C53, G22, G23, J11 Keywords and Phrases: Mortality, dynamic factor models, forecasting. 1. I NTRODUCTION Defined benefit pension schemes and life insurance companies with books of annuity policies are in the business of managing risk, effectively providing an unknown stream of payments in return for a known stream of premium payment(s). They agree to provide a pension for the remainder of lives and in some cases for the remainder of spouses’ lives and so take on the uncertainties associated with that payment stream. Generally the annuity provider will receive payment for the annuity product in advance of the first payment. They invest this money to generate a return and draw from their total funds to make pension payments as and when they are necessary. When purchasing an annuity the price is agreed at the outset and no further payments are received by the annuity provider after the commencement of the contract. The provider quantifies the cost of providing an individual with an annuity (a pension for the remainder Date: Latest version: December 7, 2011. † School of Management, Riddel Hall, 185 Stranmillis Rd, Queen’s University of Belfast, BT9 5EE, Belfast, United Kingdom. Tel.: +44 28 9097 4671; Fax: +44 28 9097 4201. Emails: c.ohare@qub.ac.uk and de- clan.french@qub.ac.uk. Acknowledgements: This work was funded by a CARDI data mining grant. We are grateful for insightful com- ments received from participants at the Harvard School of Population and Development Studies Working Paper seminar series. We are also grateful for input from colleagues from the Queen’s University, Economics and Fi- nance Research Group, particularly Michael Moore, for his helpful feedback. 1