PENSION MATHEMATICS

for

ACTUARIES,

COMMENTARY AND SOLUTIONS

by

Keith P. Sharp, FSA, FCIA, Ph.D.

University of Waterloo

ACTEX Publications, Inc.

Winsted, CT
 PENSION MATHEMATICS

for

ACTUARIES,

COMMENTARY AND SOLUTIONS

by

Keith P. Sharp, FSA, FCIA, Ph.D.

University of Waterloo

ACTEX Publications, Inc.

Winsted, CT
Copyright © 2018, ACTEX Learning, a division of SRBooks Inc.

All rights reserved. No portion of this book may be reproduced in any form or by any means
without the prior written permission of the copyright owner.

Request for permission should be addressed to:

ACTEX Publications
PO Box 715
New Hartford CT 06057
support@actexmadriver.com

Manufactured in the United States of America

10 9 8 7 6 5 4 3 2 1

ISBN: 978-1-63588-651-1
 Ill

TABLE OF CONTENTS

Preface

Chapter 1: Introduction
Introductory Overview.....................................................................................................1

Chapter 2: Actuarial Cost Methods

Introductory Overview.................................................................................................... 2
Text Exercise Solutions................................................................................................. 12

Chapter 3: Contributory Plans

Introductory Overview.................................................................................................. 51
Text Exercise Solutions................................................................................................ 53

Chapter 4: Ancillary Benefits

Introductory Overview................ 76
Text Exercise Solutions................................................................................................ 77

Chapter 5: Assets
Introductory Overview.................................................................................................105
Text Exercise Solutions.............................................................................................. 107

Chapter 6: Assumptions
Introductory Overview.................................................................................................123
Text Exercise Solutions.............................................................................................. 127

Chapter 7: Cost Methods in Current Actuarial Practice

Introductory Overview................................................................................................ 147
Text Exercise Solutions.............................................................................................. 150

Appendix I 186
IV

Acknowledgements

This book briefly reviews pension funding, adds to motivation and intuition and gives solutions
to Pension Mathematics for Actuaries, Third Edition, by Arthur W. Anderson, ACTEX
Publications Inc.

I am grateful to the many people who have helped in the preparation o f this and previous
volumes. Special mention should be made o f the late Professor W.H. Aitken, and o f Shawn
McKenzie and Isa Rachmatarwata, all o f whom made significant contributions. All errors remain
the responsibility of the author, who would appreciate notification of them at
keith@actsharp.com.

Keith P. Sharp, FSA, PhD.

April, 2006
 1

CHAPTER 1: INTRODUCTION

It is clear that saving a portion o f income to provide a future pension is an important goal. One
can perform a simplistic calculation to determine a funding rate as a proportion o f salary. Use the
following “dream plan” :

Pension: 100% o f final salary

Termination benefit: Full accrued liability

Indexation: Full CPI after retirement.

Working life 40 years

Retired life 18 years

Assume equality o f future interest rate, salary increase rate and price inflation. Assume zero pre­
retirement deaths, or equivalently assume the value o f the pre-retirement death benefit to be the
full accrued liability. Entry age is 25. Then the funding Tate depends on the ratio o f the number
o f years retired to the number o f working years: 18/40 = 0.45. We can see this without using
funding formulas. Each of the 40 years o f working life needs to pay for 18/40 years o f pension.
If we include the contributions paid by or on behalf o f the individual to government and
employment pension plan we should, on this very rough basis, be saving 45 percent o f our
salaries.

Very few individuals are saving 45% o f their income. And most employment pension plans have
a total contribution rate much less one-third of the 45%. The above calculation has made several
excessively simplifying assumptions. The biggest simplification is the assumption that
investment returns will be no more than the rate o f salary rise. Another significant point is that
the after-retirement income should be compared with the pre-retirement income net of pension
contributions. Also, in the context of a corporate pension plan, there are usually substantial cost-
savings when employees resign or are terminated, since the termination benefit is typically much
less in value than an accrued liability calculated assuming future salary increases. It is clearly
important to make more accurate determinations o f the rate at which millions o f people should
be saving. Pensions are important. Let’s start looking at them more closely.
2

CHAPTER 2: ACTUARIAL COST METHODS

2.2 Unit Credit

First we will set up the notation. The arrow ‘PV’ implies that present values under the unit credit
method are calculated as at age x, the age at valuation. Strictly speaking we should use the
notation xJ for the age at valuation of person j. However, the j is usually omitted.

w u x
y
entry plan inception valuation retirement

T set of terminated pension plan members

R set of retired members
Af set of actively working plan members
w entry age from which pensionable service is calculated
u age at plan inception
X attained age at valuation
PVFB present value of future benefits
F fund actuarial value
NC normal cost
sx actual salary in the year from age x
salary scale
AL PVFB - PVFNC = accrued (actuarial) liability
= A L a + ALi where AL / is for inactives/pensioners.

The above gives the most complicated case. It allows for the possibility of generosity at
inception: the period w to plan inception is given ‘pensionable service’ status. For a person
hired after inception we can drop plan inception from the diagram.

Aff(x)
---------------- ►

w x+1
y
entry plan inception valuation retirement

| pv
 3

Pensions are assumed actually not paid until retirement age y. But we allocate a portion AB(x) to
the year following the valuation date. For a United Auto Workers - type plan for example,
pension = (y-w) x 40 x 12 per annum from age y. So we have AB(x) = 40 x 12 per annum
‘accrued’ in the 12 months from x

We turn now to the derivation o f accrued liability under unit credit:

B(x)
----------------- ----------- ------------------ ------------ ►

y
w ^ x~t 1 retirement
entry plan inception valuation

| pv

We allocate a portion B(x) o f the age y pension to the period w to x:

B(x) = AB(w) + AB (w+1) + ... + AB(x-1)

which, it should be noted, does not include AB(x)

For Traditional Unit Credit (TUC), which is usually used for non-final salary plans, AB and B(x)
come straight from plan provisions. A career pay plan might define the retirement benefit as, for
example, B(y) = 2% x total career earnings, with no adjustment for interest or inflation:

AB(y) = 0.02 x Sx

B(x) —0.02 x (Sw+ Sw+j + ......+ Sx.i)

Considering now Projected Unit Credit (PUC), usually used for final/final average salary/fmal
average earnings plans, AB(x) and B(x) use the salary projected using scale sz. For example,

B(y) = 2 % x final 12 months salary x (r-w) = 0.02x Sx x (sy. i / s x (y-w)

AB(x) = B(r)/(y-w)

B(x) = B(r) x (x-w) /(y-w)

The normal cost is the payment for the next 12 months accrual

NCJ = ABf(x) x y.4 jx x ^ x x Uy(12)

4

The accrued liability is the value o f the benefit ‘earned’ by age x:

AU = B J (x) Xy-xPx x vy'x x a y 12}

The normal cost is, under Unit Credit, the amount to pay for ‘this year’s benefit accruals,
summed over all active plan members

NC=£,eAt NCj (x)

Use o f the Accrued Liability

A major use of the accrued liability is to assess the safety o f the benefits. The accrued liability is
compared with the value o f the assets. The accrued liability should include all who are entitled
to benefits that will deplete the assets. Inactives should be included, e.g. those receiving pension
AL=B(y) ax(,2). The comparison o f the plan ALt with assets F, gives the unfunded accrued
liability:

UAL, = AL,-F,

f icactivesAL[ f k c mactives AL I - F,

UAL, indicates, roughly, the value o f promises not yet paid for, a matter o f particular interest to
regulators.

Bfx) ABf(x)

w x 1 y
entry plan inception valuation retirement

t pv

The Projected Unit Credit method apportions projected the benefit B1(y) proportional to service:

Aff(x)=B'(y)/(y-w)

0 (x) = B'(y)x (x-w)/(y-w)

and the normal cost and accrued liability are found from

NO = Aff(x) x y.xPx x x ay02)

ALJ = B J (x) x y.xpx x F'x x a J I2)

 5

2.3 Entry Age Normal, Level Dollar

Compared with benefit allocation methods such as Unit Credit, this is a whole new way of
thinking about pension cost. We know that we need a lump sum B(y) ay (12) at retirement. Now
we arrange to spread cost (not benefit) level over service. A defining feature of Entry Age
Normal is that we spread the pension cost from entry age w to retirement age y. Here it is
possible that entry age equals a hire age at a time before the plan had been created, with past
service to hire granted:

NC jean

w u x x+1 y
entry plan inception valuation retirement

| PV

As for a mortgage, we have that the initial PF(payments) equals the value o f debt when
payments start. Hence we have for one particular member j:

pvwJ =pvwf f

and more specifically:

NC xeanj x aa)w:y.wl= y .wp m w x /-* ’ x B(y) x ay (I2)

The total plan normal cost is given by:

JVC, Z j e actives N C f ( j , t)

= Zactives NCX (abbreviated notation)

and the accrued liability for member j is given by:

AIxean 1(Prospective) = pvxB r -pvx NCJ

= y-xp m x x J-Xx B(y) x ay ° 2) -NCJ ax:y.xl

If there are zero decrements we can prove that the same AL is given by

Alx (Retrospective) = Accumulated value o f past NCJ

= (l+i)NCJs x.wl
6

This is much the same as the equivalence of retrospective and prospective reserves in life
insurance, using the valuation net premium.

However, in pensions the ‘past NC ’ is a notional NC using the age x plan document and the
current age x valuation assumptions. Use of the prospective formula

AlxEANJ(Prospective) = pvx B y - pvx NCJ

for AL is safer and less confusing since it is clear that the most recent assumptions, data and
benefit definitions are being used throughout.

2.4 E ntry Age N orm al, Level %

By
Ux S
<
*►

w u x x+1
^ try plan inception valuation
y
retirement

PV
As at entry we set the payments so that they will fund the pension:

UJx (Sx x s j s x) x saw:y.wj= y.wp m wx vy'wx B(y)] x ay(,2>

There is a subtlety in getting NC as percentage U o f salary. We use Sx x s-Jsx , not the actual
historical Sweven if Swis known. This is consistent with using projected Sx to get B(y)

Then we can calculate the prospective accrued liability:

AIJ(Prospective) = pvx ffy -pvx NC1

= y-xPm x x J 'x x Bj y X ay (I2) - U JX Sx x % y.x7

Not-' that under EAN at plan inception AL J =? 0. If it is desired to avoid making supplemental
cost payments, and regulations permit, then ILP might be used instead o f EAN.
 7

Amortization o f UAL: e.g. Under EAN

SC-U ALiioi/iiwl

1.1.01 plan inception with 1.1.11

back service

We assume that at plan inception, service before 1.1.01 is made ‘pensionable’ Rarely, a pension
is awarded to those who retired before the plan existed, which also boosts the plan accrued
liability. No normal cost NC was actually paid before 1.1.01 since the plan didn’t exist.
Probably there is no fund Fat the 1.1.01 plan inception so F} I Oi = 0. Hence

UALU.o. = AL - Fu .oj = Z j (pvxBiy - pvx N G ) - 0 > 0

Because, recalling at entry age w (not x) we have:

pvw NO = pvwBJ

So at inception, we have UAL] A. .01 > 0. We pay off the UAL] ] 01, typically by making level
dollar supplemental cost payments. The maximum period for the amortization is set by
applicable regulation (e.g. ERISA). Assume here that 10 years is chosen. So the plan is to pay
for 10 years a total cost

NC + SC = [ZjeacuvesNC^.j] + UALu .o, /dial

2.5 Individual Level Premium

By
< N d LP ►
►

w u x x+1
plan inception valuation
y
entry retirement

Under ILP we adopt the philosophy that at retirement we need a lump sum B(y) x ay (I2>' If plan
inception is at age u, after pensionable service commences at w, then the obvious way o f
8

accumulating the lump sum is over the period u (not w) to y. Assume in this discussion u>w,
otherwise ILP is the same as EAN.

We set up the accumulation by taking an equation of value at u. We require /^ (p a y m e n ts) to

equal the present value o f the required lump sum. So PVs and annuities now start at inception u:

N C X1LPJX aU:y-ui= y-uPm ux y T x B(y) x ay (I2) (ILP.01)

For the whole plan we add the normal cost of each individual:

NC, = £ a NCx

The accrued liability found prospectively is:

Alx (Prospective) =pvxffy - pvx NCJ

= y-xPm Xx x BJy X a'y° 2) - NCxx ax:y.x]

Note that imder ILP at plan inception we have a zero accrued liability:

AL uJ = y.up m u X /■“ X BJy X ay (,2) - NCUX au:y.ui

=0 (using ILP.01)

ILP is virtually always used in its level dollar form. For both EA and ILP, the accrued liability
can be calculated prospectively (AL = PVFB - PVFNC) or retrospectively as an accumulation of
normal costs.

2.6-2.9 FIL, AAN and Aggregate Methods

One formula can be used for the Frozen Initial Liability, Attained age Normal and Aggregate
methods in their level percent form. We use the notation U for the normal cost as a fraction o f
payroll. All quantities are at time t, where ages are x'(t), abbreviated to x:

U(plan) = (EPianPvxB- UALt -F )/£A pVxS (1.01)

We multiply by Sx to get plan dollar TNC

The level dollar form for these Frozen Initial Liability, Attained age Normal and Aggregate
methods is, per person (‘Joe’)

NC(Joe) = (ZpianPvxB- UAF-FJ/Z a ^ ^ - u] (L02)

 9

We multiply by the number o f actives to get the plan total dollar normal cost.

We choose UALt appropriate to the cost method, through considering the gain formula (see for
example Anderson’s Equation (2.6.5)):

Gain, = (UAL,., + NC,.,) (1 + i) - C - I c- UAL, (11.01)

Under FIL and FIL(AAN) we define Gain,=0.

So we can now get UAL, from UAL,.i (see Anderson’s (2.6.11)):

UAL,= (UAL,.! + NC,.,) (1 + i)-C - lc (11.02)

Equation (11.02) says for FIL and AAN the UAL is frozen. We can use (11.02) to iterate UAL,
from year to year.

The starting values o f the iteration at plan inception 00 for the two methods are defined as:

UALoo(FIL) = UAL0o(Individual-EAN)

UALoo(AAN) = UALoo(Unit Credit, usually TUC)

The phrase ‘Frozen Initial Liability’ implies that the UAL decreases only by the effect o f the
amortization payments. Items such as good investment performance do not reduce the UAL for
these methods. Let’s examine this more closely in the light o f our UAL iteration formula:

UAL,= (UAL,., + NC,.,) (1 + i)-C - Ic (11.02)

Let’s say we are amortizing level dollar over n years, so SC is chosen as one would calculate a
periodic mortgage payment. The SC is the portion o f cost paid which does not represent normal
cost:

SC = (C + Ic) /(1+i) - NC,.,= UALoo/dnl

Hence the iteration is:

UAL, = (UAL,., -UALoo/dnj)(l +i)

This is a standard level dollar loan amortization with fully predictable (frozen) progression o f the
outstanding principal:

UAL, UALoo * &n-t 1(hnl

10

It may seem as if defining the gain to be zero is a dramatic step, which we consider here. Notice
that the pension methods take the PVFB and say that this is paid for in three ways:

1. the fund F
2. the present value o f future normal costs PVFNC
3. the present value o f future amortization payments (supplemental costs) UAL = PVFSC

So we have the very general pension formula (see e.g. Anderson’s equation (2.6.12))

PVFB = F + PVFNC + UAL =AL + PVFNC

This formula, suitably specialized, applies to all pension methods. There is an analogy to buying
a house with a combination o f down payment, mortgage and second mortgage: there are two
payment streams, one for each mortgage.

In pension funding, the two payment streams must sum to PVFB - F. But the choice o f method
is really the actuary’s decision about the load taken by each payment stream (normal cost or
supplemental cost for amortization). Our ‘gain’ is really ‘unfunded liability gain’. We could
(but usually we don’t) define also a normal cost gain, an unexpected decrease in the present
value o f necessary future normal cost payments. By saying ‘gain=0’ for the FIL and AAN
method we are really saying that any good or bad experience is all taken into the normal cost.
For example, (1.01) above leads to a decrease in U if the fund F is increased by good investment
performance.

For the aggregate method we define UAL, = 0 at all times. We throw away the option o f having
the second payment stream and just use one payment stream, usually calling it a normal cost.
For the level percent o f payroll version, which is usually used (1.01) becomes

U(plan) = (ZpianPVxB- UAL,-FJ/Za pVxS

= (ZpianPVxB -FJ/E a PVxS

Gains and Losses:

The above Gain, given by 11.01 is the total gain. The interest portion o f the total gain is found
from the excess o f interest earned over that which would have been earned at the assumed rate

G!t = I - (iF, + Ic - h)

= F,+I.- F, (l+i)— (C+Ic)+(B+Ib)

For decrements such as mortality and termination, the gain in general is given as the difference
between the actual and expected release of liabilities; if a lump sum benefit L x+i is given and
interest at the assumed rate to year-end is 1l then:

GDec= Z T(ALjx+I- F x+I- I 1) - Z Tqx(ALjx+I- L Jx+1- I l)

 SOLUTIONS

2.2.1
LH S = (px + <7x) d^ 7 = v*+i<x+1 +

= «*+*/« + = (1 + + Qx~
d^ I

2 .2 .2 (a) We recall th a t px = —D log

1+ i = es, log (1 + i) = 6, D = <f/cfa.

L i /S = Z> log NC*

= D log A Bx + D log ay + D log Dy - D log Dx

= D log A Bx + 0 + 0 — D log vx — D log l x

= D log A B x + Dx log (1 + i) + px

= D log A Bx + $ + px

(b)
A Bx = .015, = .015w(l + r)x~w
log A Bx = log .01 + log Sw + x log (1 + r) - w log (1 + r)

D log A Bx = log (1 + r)

••• D log N C X = px + 6 + log (1 + r)

(c )
D log Sx = D log 100ABX = D log 100 + D log A Bx

= D log A Bx

D log Ag* = D log N C X - D log Sx

= Px + 6 + log (1 + r) - log(l + r)

— px + £
(d)
NC
D log — = px + 8 — aebx + 6
^x
 13

aebx is exponential

aebx + 5 is “more than exponential”

Under the credit method, costs generally rise more rapidly than covered payroll.

2.2.3

ALi = B(x)-ay - ^ -
X~\

= accumulated past normal costs.

2.2.4

Dv
ALi = B(x)-ay -=f-
^X
1 -A ALi = A log ALJX
ALi
AALi ALi (A log B(x) + 0 + 0 - A lo g ( v ^ j )
1
= AL A8(X) + +8
B(x)

Vy(y dB(X)
= a + (mx+S)AL x
y vxi dr

:± T d NCi + (jux +5)ALi

ax

Equation (2.2.16) says that the accrued liability increases because of the increasing
accrued benefit, survivorship and interest. Equation (2.2.17) says that the normal cost
pays for the increase in the accrued benefit.
 14

2.2.5 Equation (2.2.3) is:

AL <+i ALt + E A £ i a<12>§“ ( l + o [E te rm is NC, (2.2.4)]

at x

5 B i ( x + l j a ^ ) ^ - Z q , 0 ( x + u a W jg j
Dx+1 [= 0 if assum ptions OK]
IT

- T
2g B
n xj +, ,a(12)
la y -^
£)x+1 [= (P + Ip) if correctly guess
am ounts for retirements]

(2.2.18)

ATC.+, =
«t+l

= o i E _ r ^ a < ” > [ g ( i + .) + « ,D,.,

£ ]

= EA B >a<12) ^ ( i + i) [= NCt( 1 + «)]

at

[= 0 if assum ptions OK]

- 5 a^ 12)t& [= 0 since A B J = 0 for rets]

= j v c f(i + *) (2.2.19)

2.3.1
a l p r o sp e c t iv e = J 2 B i ( y ) a W ^ - Z N C j Nx -N„
at x at Dx

= J2NCj ^ ^ - - Z N C j - *D-xn 9
at x at

= J2N C j ^ ^

R E TR O SP E C TIV E
 15

2.3.2
N w Nx+i
LH S D x+ i (Px “t“ 9ar)

__ Nut Nx+1 tx + 1 i Nw Nx+\

- «* + !* ,+ 1 t x ^ V* D x+ 1

= + + qx Nw-Nx^

= RHS

2 .3 .3 (a)
A LX N C XN*-Nx
Dx

B • ay _A ,_ N w- N x
N w- N y Dx

B ■ay Dy N,.,-N_r
Dx N w—N y ‘

0>)
A LX — N C + ( fix + 8)ALX (2.2.16)

= N C at age w because A LW — 0 .

(c) R ate of increase in A LEAN at age w is greater because N C EAN > N C UC.
(d) They intersect at retirem ent age. The AL is greater under E A N , so term inations
greater th an expected give a greater release under E A N , i.e. greater gain. Early
year N C is larger under E A N , hence a larger fund.

2.4.1
SNX = SDx + SDx+1 + . . .

= Sx ■ D x + SiT-fi • D x + 1 + . . .

= (1 T r ) x • Dx + (1 + r ) x+1 • Dx+\ + . . .

= N'x iff (1 + r)x ■Dx = D'x etc.

16

which implies th a t (1 + r)xvxl x = v'xt x

or v' = (1 + r)v = l±r

i+*

or 1 _ 1+r
1+«' ~ l+i

or t' — l±i _ i
l+r

2.4.2 B in formulas (2.3.1) and (2.4.1) are the same if B not dependent on salary.

(2.3.1) implies th a t N C J = Bay^ / ( Nw

d^ v)

Dv / / sz Dz
(2.4.1) implies NC w ' \ sw “ sw D u

— = i
g
— > 1 for z > 1 if salaries increase
$11!
V_1 s* Dz Dz Nw - N y
■ Eyj $w Dw >Ew Dw Du
N C under (2.4.1) is lowered .

2.4.3
B„ = c Sy , say

C Sw(l+r)»-"aW§%iNlD
JLNv
NC) C 4 12)^ 7 7 ^

— M I r \ v —,w 3Dw D w + D w + \+ :* + D y —\
' ' D w sD w + s D w+ \+ ...+ sD y - \

— (1 J_ r*^2/-u,n 4_ __________ ^w+I>w+i + —+-Py-1__________

" ^ (l+r)«-yDw+(l+r)^-»+iDw+1+ ...+(l+r)-iDy- 1

__ __________________ D w + D w + l __________________
” (H-r)^-J/Dty+(l+r)^-s/+1Dti;+1Dti;+i+...+(H-r)-1Dy_i

> 1
 17

because (1 + r)w y+l < 1 \/i st 0 < i < w — y

No. To end up w ith same fund, the “no salary increase” N C m ust overtake at some
time.

At age xl -£ § f =
c s * 4 12>

( \ | r'\v ~ x (^ 1 r \ w
" V1 + T) V1 ■+■ T) - D ^ + ' D ^ + i . ..+*£>„_ i

- (1 i r \ w - x _______________ D w + D w + \ + —+ D y - l _______________ 1 fQ _ \
\L ^ r) (l+r)' "-vDw+ ( l + r ) w- v + 1D w+1+ . . . + ( l + r ) - 1D y - i ^ 1 101 J' W)

- (\ 1 27 D w+ D w+ i + . . . + D y- i
VA ' r ' / D „ ,+ ( l + r ) D u,+ i + ...+ ( l + r ) s / - l- " '£ » y_ i

(big for small x, < 1 for x close to y (< 1 for x = y)

 18

2.4.4
NCi = B>

= c s* § tv5?tr, for some c

= c S „ ( l + r ) ' - * a ! l, s> if c j£ S v ;
So norm al cost increases by ratio (1 + r) each year
So constant as a fraction of salary

ALj
/\ l,x — c S a^12) ffy
ox ay Dx Nw_Ny

-— c S (1+- tr)x~w
) a(12l £}
ay N w _~n*
D xlnv Ny

So have the usual increase in AL3

X (proportional to Nv~N* = ^ 7 r^ |) accelerated by
the (1 + r)x~w factor.

2.4.5
£ * ± 1 sN x+1- ° N v _ ‘NX- ‘ NU-
sx SD X+1

= - l) (! +*)(‘+rffc)
= - l ) (1 + .•) + (2.4.9)
 19

2.4.6

Define AL t+i
7(^£35*0 (from 2.4.6)

ALU, (B i + (from (2.4.6))

• "(12)
Bl±!L
4 12 )gj> , »X £ 1 ( sN w- » N x+1\ , A D j z ( 1 2 ) _ D Il_ f ‘Nw- ‘Nx+1\
‘N, ‘N y b *
Nw- ( *£>*+i ) a y D * + i ^ >Nws Ny )

B i a W — 2 * ___ + l ) M -I- i ) -|- a 3N w - ’N x^ ]

Dtav >Nw->Ny IV °DX + l ) V1 + + " x sx *DX+i J
. a i- > W io \ D„ 1\
ay d x+1 ( >Nw- > N y J

= +o+ + o
+ 5Nu,

= (JVC/ + ALJ )(1 + i) + + A B iifW -fc

— A n in(12)D» ‘D'“ sx+l ,NX+l-"Ny (2.4.7)

‘~^a uy Dw *Nw- ° N y sx »DI+ 1

Recall — -Fi -b / C —P (2.2.5)

20

UALt+i ALt+1 - Ft+1 = E AL{+1 - E AL{+1 - Ft+1

T

(ALt + N C t)( 1 + t) + E + E A B* a(12)- g h

at at x‘tl

Y' A R j/j (12)^2- >Dw s*+! SNx+l-"Ny

2- uy D w >N w - * N y sw *DX+1
at

£ r (B ’ + A B > )a f - Ft+i
(N oteA B = 0 onT, i?, so :)
(AL, + JVCt)(l + i ) + E 9*aX h - E A ^ +1
at X+K

I A f>j *’ (12) ( Dy _ Dy 3Dw sDw sx+i sN x+ i - 3N y \

t- % \D Dw Dw *Nws Ny Sw sDx+1 ) - Ftt+i
at

ALt( 1 + 1 ) - Ft - I + NCt( 1 + i) - C + E qxAL3t+1 - E AL’t+i ~ E AL3t+1

at R

4- V A R i g ^ ( Dy Dy a*+1 SNx+i-’Ny\ i p
r LXD a v \ D X+1 D w >Nw - > N y sw 3Dx+i ) T
at+i

UALt{ 1 + i)

[C + / c —WCt(l + i)]

[/ ~ iFt —Ic + ip]

3Dw ^s+l 5-^Vz4-l ~ SNy \
£ A«S,!’f e D w 3N w —3Ny Sw SD X+1 /

E ALt+i E 9r AZ/^_j_i
.T at

ealU , - p - ip (2.4.8)
mR

If experience follows assumptions, the last five terms, which comprise the gain, are zero

Gain = (UALt + NCt)( 1 + i ) - { C + Ie) - UALt+1 (2.3.12)

2.4.7 No: A Lt = P V F B t - P V F N C t For UC & EAN and other cost methods:

Entry Age

= T .B ‘ ( y ) a , % ^ - T N 0 ^ - (2.3.3)
at x at 1

= P VFBt - P V F N C t

U n it C r e d it

ALt = E # ( x ) a j 12>g«- = E = E E N C i ^ - (2.2.13)

at x at at x = 0 x

= AVPNCt

= PVFBt ~ PVFNCt (Eqn of value a t t)

W h a t is D iffe re n c e UC vs. E A N ?

U nit Credit AL\ = E B i{x)a^2}jf- = P V Benefit accrued to date.

Normal cost is “cost of benefit accruing this year”
E ntry Age “Benefit accrued to d ate” not defined.
No simple interpretation of A Lt.
Cost is spread over pensionable service.
 22

2.4.8 Interpretation:

Increase in present value at x + 1 of retirem ent benefits because of change A B in

estim ated benefit

LESS

value at (x + 1) of increase in future norm al costs

y' A ( Dy________ iQw__**+i SNx+i~’Ny\

J- uy \ D X+1 D w sNw- ’ Ny sw °D X+1 )

= E, ^ (l - ^ 5 ^ )

__ pi "(12) / SNW—SNx+i \ p i “ (12) Dy ( sN w—sNx+i \

~ a5i +1 * D"+1 ' **«■"*** ' S i t y Dx+1 ' /

= E ( ^ 4 +i - < , )
at+i \ /
 23

{N x- N y Nx+1- N , )
2 .5 .1 A I4 » - B . O ' j r f r + A5 a .— ?
+1 Z>X + l DX + l

- 'W " .
' A* + l

„ •• A, 1>«
= B„a„—y *.. — ------+ 0
W .-N , DI+1

= . . D‘„ ? * - = NC£ D° x
Dx Nx- N , Dx+1 X+l

. Dx+ t
2 .5 .2 AL1+1 = £(AL/+NC/)
Dz+t-i

= £ (ALt+NC{)(+±L)(px+t+qx+t)
Px+f

= £ W + N C /K l+ O + £ 9x+((AL/+NC/)(-^-)
^ P*+‘

= £(ALf+/VC/)(l+t) + £?*+tAL/+1
24

2.5.3 L eft side of (2.5.17) ■AL{

n
.. Dy n
.. A, *DZ «x+1 Dv *DX+1 8NX+-*N9
■ (B-\-&B)av~~t ------- {Baym
rr------------------------------- -h ABay~p~ ) •
A+l Dx SNx-8Nv 8x A + l 8NX+,-«Nf »D,+l

_ ^
R- DD*
= BavTr~
y Dy e D* 8 *+ l 6 " , + r S * y
^ -----^ + 0
y Dx+1 'A «» «X>,+1
#x>.+i

Dy *DX A 8N ~ 8Ny 5*+l *^*+1 6^y Dy

Dx 6Nx—6Ny Dx+\ 8Dx 6* SDz+1 A
>
A 6*nx~ 6ny 6Nx+ ~ 8Ny Dx
= NC3
X
D x+1 8Dx 6d x Dx+i

= v c .,- A ^ ,- y K .r x )
A +1 SDX

A eDz . A
= iVCi = NCi-
• A + l e D. D* + i

2.5.4 Ul = norm al cost as a c o n stan t fractio n of salary

NCj
.*. FVFNC* = UilVFSi and U{ =
~ sf

FVFNC} . PVFBNAL! .
NC} = „ „ J Si = ---- - Si (2.5.18)
FVFSf FVFSj

E asier to use th a n 2.5.15 and 2.5.14 p a rtly because th e re ’s no need to go back and
exam ine last y e a r’s inform ation.
 25

2.6.1 G iven FVFB{ = Bt(y)av-^ - FVFBt = ^FVFB}

UX

B„
E w fbU = E A +,(y)«, D
^+1 *+l

= T ,B t+1{y)a - E = E^® /
At T+ R Af

D, D, D„
= + E ^ ' vT ^ - E Bt{y)av- ^ ~ AB = 0 for T+R
/ z+ i 1+ 1 T+H *+l

B„ B„ D„
= -^ (1 + 0 + qx
D t +i

Dv (i+t)
, Dv
= E Bt{y)ay — + qz — + T j ABav—^ - - E Bt ' a . j f —
u x u x+ \ At u x +1 T+R u x+ \

D„
= F V F B t { l + x ) + E f c ^ B / f i + E A -B“ v T r ! - “ E ™ J ' +i
'x + \ t +R

D„
= /V F B t ( l + 0 - ( E ^ / +1 - Efl«™ /+i) - E ^ 5 y+ '£ ^ B a v—^~
T At R At U *+ l

~ implies values computed as though the increase AB in the projected benefit

between times t and t+ i = 0.

^ X+1 Ny N x+ \ Ny t N*, x+l

■ x+1- N’yy ‘g-t-l
lx+l _ N
■ x+1- N* 'yv
‘ ■x+l
2.6.2 (P*+ 9x) x^ + qz - D
'x+l 7x+l -" 'u , X+l

N z Dx A/’,, -^*+1 . Nx Ny A j.+1—A y

---------------
Dx (1+0 + 9*— K-------= (— k -------- 0(1+0 + qz-
Dr D.
/*+i X+l

••• E % - ^ = E - E - E
^t+i ^x+l \ T R

N x-N y N X+1- N y N X+i~Ny N X+l~Ny

= E ( —TT-1 - - 0 (1 + 0 -E B x+i
T R

JV
D
2.6.3 D + A D VN+AN
N
A N - — AD
7V+A7V = N_ D
N D+AD D + A+AB
A N —- ^ A D