Fractional Age Assumption
Study guides: Exam FAM, Exam ALTAM
A Fractional Age Assumption fixes how deaths, or other decrements, are spread within each year of age. With it, a life table that gives only \ell_x or q_x at integer ages also yields {}_sp_x, {}_sq_x and the force of mortality \mu_{x+s} for 0 < s < 1. The two in AMLCR Section 3.3 are uniform distribution of deaths (UDD) and constant force of mortality.
\text{UDD:}\quad {}_sq_x = s\,q_x
\text{UDD:}\quad \mu_{x+s} = \frac{q_x}{1 - s\,q_x}
\text{Constant force:}\quad {}_sp_x = \left(p_x\right)^s
\text{Constant force:}\quad \mu_{x+s} = -\ln p_x
- What UDD means.
\ell_{x+s}is linear ins, so the density{}_sp_x\,\mu_{x+s} = q_xis flat over the year. Equivalently,T_x = K_x + R, whereRis uniform on(0,1)and independent of the curtate lifetimeK_x. This gives\mathring{e}_x = e_x + \tfrac12. Under a constant force,\ell_{x+s}is exponential insand\muis flat within the year. - Within a year, not from age
x. For0 \le t < t+s \le 1, UDD gives{}_sq_{x+t} = s\,q_x/(1 - t\,q_x), while a constant force gives{}_sp_{x+t} = (p_x)^s. Across an integer age, split the period and multiply the pieces. - Insurance values (FAM). Under UDD,
\bar{A}_x = \tfrac{i}{\delta}A_xandA^{(m)}_x = \tfrac{i}{i^{(m)}}A_x. The claims acceleration approximation\bar{A}_x \approx (1+i)^{1/2}A_xassumes instead that deaths are paid, on average, half a year early. - Multiple decrements (ALTAM). UDD in the multiple decrement table, or constant forces, gives
p'^{(j)}_x = \left(p^{(\tau)}_x\right)^{q^{(j)}_x/q^{(\tau)}_x}. UDD in each single decrement table instead gives, for two decrements,q^{(1)}_x = q'^{(1)}_x\left(1 - \tfrac12 q'^{(2)}_x\right).
Survival Across a Birthday Given
q_{60} = 0.02andq_{61} = 0.025, find{}_{0.75}q_{60.5}under UDD and under a constant force of mortality.Answer Split at age 61:
{}_{0.75}p_{60.5} = {}_{0.5}p_{60.5}\cdot{}_{0.25}p_{61}.\begin{align*} \text{UDD:}\ {}_{0.75}p_{60.5} &= \frac{p_{60}}{1 - 0.5\,q_{60}}\left(1 - 0.25\,q_{61}\right) \\ &= \frac{0.98}{0.99}(0.99375) \\ &= 0.983712 \\ \text{Constant force:}\ {}_{0.75}p_{60.5} &= (0.98)^{0.5}(0.975)^{0.25} \\ &= 0.983703 \end{align*}So
{}_{0.75}q_{60.5} = 0.016288under UDD and0.016297under a constant force. Whenqis small the two assumptions agree to about five decimals.
From Annual to Continuous Insurance Values On the SULT at
i = 5\%,A_{50} = 0.18931. Find\bar{A}_{50}under UDD and by claims acceleration, and findA^{(12)}_{50}under UDD.Answer
\begin{align*} \bar{A}_{50} &= \frac{i}{\delta}A_{50} \\ &= \frac{0.05}{\ln 1.05}(0.18931) \\ &= 1.024797(0.18931) \\ &= 0.19400 \\ \bar{A}_{50} &\approx (1.05)^{1/2}(0.18931) \\ &= 0.19399 \\ A^{(12)}_{50} &= \frac{0.05}{0.048889}(0.18931) \\ &= 0.19361 \end{align*}Paying at the moment of death raises the value by about
2.5\%. Monthly payment raises it by a little less.
Deconstructing a Multiple Decrement Table A two-decrement model has
q^{(1)}_x = 0.03(death) andq^{(2)}_x = 0.12(withdrawal). Assuming UDD in the multiple decrement table, find the associated single decrement ratesq'^{(1)}_xandq'^{(2)}_x.Answer
q^{(\tau)}_x = 0.15, sop^{(\tau)}_x = 0.85.\begin{align*} p'^{(1)}_x &= (0.85)^{0.03/0.15} \\ &= 0.968019 \\ p'^{(2)}_x &= (0.85)^{0.12/0.15} \\ &= 0.878082 \end{align*}So
q'^{(1)}_x = 0.03198andq'^{(2)}_x = 0.12192, andp'^{(1)}_x\,p'^{(2)}_x = 0.85. Each independent rate is higher than its dependent rate. With withdrawal removed, more lives remain exposed to death, and the reverse holds for withdrawal.
How it appears on the exam
- FAM gives a life table at integer ages and asks for a probability,
\mu_{x+s}, or the complete expectation\mathring{e}_xor{}_n\mathring{e}_xat a fractional age or duration. The question states which assumption to use.- FAM converts annual insurance values to continuous or
m-thly ones withi/\deltaori/i^{(m)}under UDD, or with claims acceleration. These are then used in a premium or variance calculation.- ALTAM builds a multiple decrement table from single decrement rates, or recovers them from it, under UDD in the multiple decrement table or in the single decrement tables. The table then feeds a premium, policy value or pension valuation.
Exam traps
{}_sq_{x+t} \neq s\,q_x. Under UDD the denominator is1 - t\,q_x. Dropping it understates the probability slightly, which matters when the answer choices are close.- Mixing the assumptions.
(p_x)^sis the constant-force answer, while1 - s\,q_xis UDD. A question that says UDD and is worked with exponents lands on the neighbouring choice.- Annuities do not convert like insurance.
\bar{a}_x = (1 - \bar{A}_x)/\deltamust be built from the converted\bar{A}_x. Multiplying\ddot{a}_xbyi/\deltais wrong.- Two kinds of multiple decrement UDD. UDD in the multiple decrement table, or constant forces, uses the power formula. UDD in each single decrement table uses
q'^{(1)}\left(1 - \tfrac12 q'^{(2)}\right). They give different tables.