Actuarial Notes

Life Contingencies: Study Note for CAS Exam S

By Tom Struppeck · Casualty Actuarial Society · 2014

Study guide: Exam MAS-I

A short CAS study note on life contingencies — survival models for human lives and the cash flows that start or stop contingent upon survival. Written by Tom Struppeck for CAS Exam S and marked "Revised 9/19/2015", it illustrates terminology and techniques in straightforward contexts for a P&C reader, for whom such calculations are typically intermediate results, and points to where further refinement is possible; exercises follow several chapters, with an appendix of selected solutions.

On the syllabus

  • Exam MAS-I — objectives A5–A6; the whole note (cited as October 2014, revised September 2015; an online publication).

Introduction

  • "Life contingencies" is a term for survival models for human lives and the resulting cash flows that start or stop contingent upon survival — a central topic for life insurance actuaries.
  • In P&C work, life contingency calculations are typically intermediate results used to estimate some other quantity of interest.

1 Life tables and an extended example

  • A life table presents a family of conditional survival random variables concisely; the chapter builds one from 300 vacuum cleaners that all fail within four years.
  • The mortality rate (or mortality) is the number expected to fail in the coming year divided by the number still functioning at its start; l_x, d_x and q_x denote the number functioning, the number failing and the mortality.
  • Actuarial present value (APV) is the present value of a stream of cash flows that depend on a random variable; assuming a single deterministic interest rate, the chapter prices a $100 warranty paid at the end of the year of failure (APV $90.09 at 5%) and the level annual subscription that funds it ($43.30 a year).

2 Comments on Life tables

  • (x) denotes a life of exact age x; d_x = l_x - l_{x+1}, q_x = d_x / l_x and p_x = l_{x+1}/l_x.
  • The radix l_0 can be any positive number, usually a round one such as 100,000, since only the mortalities matter.
  • {}_np_x = l_{x+n}/l_x = p_x\,p_{x+1}\cdots p_{x+n-1}, and {}_nq_x = 1 - {}_np_x.
  • Exercises use the exam's Illustrative Life Table.

3 Contingent Cash flows

  • Life insurance contracts pay $1 at the end of the year of death: term insurance for a fixed period, whole life with an unlimited term, endowment insurance (term plus $1 at maturity), and the pure endowment (the payment at maturity alone).
  • The whole life APV satisfies A_x = v\,q_x + v\,p_x A_{x+1}, so A_x = \sum_{n \ge 0} v^{n+1}\,{}_np_x\,q_{x+n}.
  • A life-annuity-due pays at the start of each year while (x) is alive, \ddot a_x = 1 + v\,p_x\,\ddot a_{x+1}; a life-annuity-immediate pays at the end of the year, a_x.
  • By an arbitrage argument 1 = d\,\ddot a_x + A_x, so \ddot a_x = (1 - A_x)/d, and the level annual premium P solves A_x = P\,\ddot a_x.
  • Annuities can be limited to n years or guaranteed for at least n years; the exercises derive the variance of an insurance (its second moment is the first moment at double the force of interest) and of an annuity.

4 An example from workers' compensation insurance

  • An injured worker's monthly stipend stops on return to work, so "mortality" here is returning to work; the examples assume annual payments and a level 3% discount rate.
  • An annuity-due on (20) with a 5-year certain period is a five-year annuity-certain plus a life annuity on (25) deliverable in five years: \ddot a_{\overline{5}|} + (1.03)^{-5}\,\frac{l_{25}}{l_{20}}\,\ddot a_{25}.
  • An annuity on (40) paying until return to work or age 65 is replicated by a life annuity on (40) less one on (65) taken back in 25 years: \ddot a_{40} - (1.03)^{-25}\,\frac{l_{65}}{l_{40}}\,\ddot a_{65}.
  • Under US regulatory accounting most reserves for future loss payments are not discounted, but certain workers' compensation reserves are.

5 Why P&C companies don't sell level premium products

  • A level premium buys the first year's coverage plus the option to buy future years at the same price; a purchaser who does not renew has lapsed.
  • With non-decreasing mortality the early excess premium (policy values, or premium reserves) is positive; with decreasing mortality the rational insured lapses after taking the subsidy and buys cheaper term cover.
  • P&C risks tend to have decreasing hazard rates over time, because the worst risks have their accidents earlier and only better risks remain in the cohort.

6 What happens mid-year?

  • Monthly timing needs mortalities between integer ages, typically interpolated from the annual ones: linearly, exponentially or hyperbolically.
  • In continuous time mortality is described by survival functions and hazard rates, covered elsewhere in the syllabus; a hazard rate describing mortality is called the force of mortality.
  • A workers' compensation claim in paying status faces competing forces — return to work, reaching 65, death — only one of which stops payment.
  • Increased mortality is applied, by custom, by multiplying the mortalities (a footnote notes that multiplying the hazard rate by k turns p_x into p_x^{\,k}).
  • An annuity can pay based on the survival of more than one person (joint lives): until one of the two dies, \ddot a_{xy}, or until both have died, \ddot a_{\overline{xy}}; an arbitrage argument relates them to annuities on single lives, \ddot a_x + \ddot a_y = \ddot a_{xy} + \ddot a_{\overline{xy}}.

7 Life expectancy

  • The curtate life expectancy, the expected number of full years, is \sum_{n \ge 1} {}_np_x from the life table; the complete expectation of life adds 0.5 when deaths are uniformly distributed over the year.
  • The same calculation applied to claim closure gives the expected number of full years a claim stays open (0.75 in the example; 1.25 complete).

Appendix: Solutions for selected problems from Life Contingencies Study Note for CAS Exam S

  • Worked solutions to exercises from Chapters 2, 3 and 6.

Related readings

  • Poisson Processes and Mixture Distributions (Daniel - 2008) — assigned with it for Domain A (Probability Models) of the MAS-I outline
  • Introduction to Probability Models (Ross - 2019) — assigned with it for Domain A of the MAS-I outline

Sources