16

| M c K ins ey on F inanc e | S u m m e r 2 0 0 4

Internal rate of return:

A c autionary tale

Temp ted b y a p ro ject w ith a h ig h in tern al rate

o f return ? B etter ch eck th o se in terim cash
fl o w s ag ain .

John C. Kelleher
a nd Ju s tin J.
M a c Corm a c k

M ay b e fi nanc e m anag ers just en jo y liv in g

o n th e ed g e. W h at else w o uld ex p lain th eir

w eak n ess fo r usin g th e in tern al rate o f
return (IR R ) to assess cap ital p ro jects? F o r
d ecad es, fin an ce tex tb o o k s an d acad emics
h av e w arn ed th at ty p ical IR R calculatio n s
b uild in rein v estmen t assump tio n s th at
mak e b ad p ro jects lo o k b etter an d g o o d
o n es lo o k g reat. Y et as recen tly as 19 9 9 ,
acad emic research fo un d th at th ree-q uarters
o f C F O s alw ay s o r almo st alw ay s use IR R
w h en ev aluatin g cap ital p ro jects.1
O ur o w n research un d erlin ed th is p ro cliv ity
to risk y b eh av io r. In an in fo rmal surv ey o f
3 0 ex ecutiv es at co rp o ratio n s, h ed g e fun d s,
an d v en ture cap ital firms, w e fo un d o n ly 6
w h o w ere fully aw are o f IR R s mo st critical
d eficien cies. O ur n ex t surp rise came w h en
w e rean aly z ed so me tw o d o z en actual
in v estmen ts th at o n e co mp an y mad e o n th e
b asis o f attractiv e in tern al rates o f return .
If th e IR R calculated to justify th ese
in v estmen t d ecisio n s h ad b een co rrected fo r
th e measures n atural flaw s, man ag emen ts
p rio ritiz atio n o f its p ro jects, as w ell as its
v iew o f th eir o v erall attractiv en ess, w o uld
h av e ch an g ed co n sid erab ly .

S o w h y d o fin an ce p ro s co n tin ue to d o
w h at th ey k n o w th ey sh o uld n t? IR R d o es
h av e its allure, o fferin g w h at seems to b e a
straig h tfo rw ard co mp ariso n o f, say , th e
3 0 p ercen t an n ual return o f a sp ecific
p ro ject w ith th e 8 o r 18 p ercen t rate th at
mo st p eo p le p ay o n th eir car lo an s o r
cred it card s. Th at ease o f co mp ariso n
seems to o utw eig h w h at mo st man ag ers
v iew as larg ely tech n ical d eficien cies th at
create immaterial d isto rtio n s in relativ ely
iso lated circumstan ces.
Ad mitted ly , so me o f th e measures
d eficien cies are tech n ical, ev en arcan e,2 b ut
th e mo st d an g ero us p ro b lems w ith IR R are
n eith er iso lated n o r immaterial, an d th ey can
h av e serio us imp licatio n s fo r cap ital b ud g et
man ag ers. W h en man ag ers d ecid e to fin an ce
o n ly th e p ro jects w ith th e h ig h est IR R s, th ey
may b e lo o k in g at th e mo st d isto rted
calculatio n s an d th ereb y d estro y in g
sh areh o ld er v alue b y selectin g th e w ro n g
p ro jects alto g eth er. C o mp an ies also risk
creatin g un realistic ex p ectatio n s fo r
th emselv es an d fo r sh areh o ld ers, p o ten tially
co n fusin g in v esto r co mmun icatio n s an d
in flatin g man ag erial rew ard s.
W e b eliev e th at man ag ers must eith er av o id
usin g IR R en tirely o r at least mak e
ad justmen ts fo r th e measures mo st
d an g ero us assump tio n : th at in terim cash
flo w s w ill b e rein v ested at th e same h ig h
rates o f return .

T h e troub le w ith IR R
P ractitio n ers o ften in terp ret in tern al rate o f
return as th e an n ual eq uiv alen t return o n a
g iv en in v estmen t; th is easy an alo g y is th e
so urce o f its in tuitiv e ap p eal. B ut in fact,
IR R is a true in d icatio n o f a p ro jects
an n ual return o n in v estmen t o n ly w h en th e
p ro ject g en erates n o in terim cash flo w s o r

Internal rate of return: A cautionary tale | 17

exhibit 1

Identical IRRs, but very different annual returns

Internal- rate- of- return (IRR) values are id entical for 2 projects . . .
P roject B

IRR

P roject A
Year
C as h flo w s , $ m illio n

10

IRR

Year
C as h flo w s , $ m illio n

41%

10

41%

. . . h ow ever, interim cash flow s are reinvested at d ifferent rates

Key assumption: reinvestment rate = IRR

Key assumption: reinvestment rate = cost of capital

CAG

P roject A
Year
V alu e o f c as h flo w s
at y ear 5 if
rein v es ted at 41%

5
5
5
5

R1

Year

20

41%

14

41%

10

41%

41%

C A G R1

P roject B

V alu e o f c as h flo w s
at y ear 5 if
rein v es ted at 8 %

5
5
5

5
Y ear 5 value of $ 1 0 million investment =

$56
million

5
7

8%

8%

8%

8%

5
41%
C A G R1

Y ear 5 value of $ 1 0 million investment =

$29
million

24%
C A G R1

True return is nearly 5 0 % less

b ecause of low er reinvestment rate
1 Compound annual growth rate.

when those interim cash flows really can be

invested at the actual IRR.
When the calculated IRR is higher than the
true reinvestment rate for interim cash flows,
the measure will overestimatesometimes
very significantlythe annual equivalent
return from the project. The formula
assumes that the company has additional
projects, with equally attractive prospects, in
which to invest the interim cash flows. In this
case, the calculation implicitly takes credit
for these additional projects. Calculations of
net present value (N PV ), by contrast,
generally assume only that a company can
earn its cost of capital on interim cash flows,
leaving any future incremental project value
with those future projects.

Consider a hypothetical assessment of two

different, mutually exclusive projects, A and
B, with identical cash flows, risk levels, and
durationsas well as identical IRR values
of 4 1 percent. U sing IRR as the decision
yardstick, an executive would feel
confidence in being indifferent toward
choosing between the two projects.
H owever, it would be a mistake to select
either project without examining the
relevant reinvestment rate for interim cash
flows. Suppose that Project Bs interim cash
flows could be redeployed only at a typical
8 percent cost of capital, while Project As
cash flows could be invested in an attractive
follow-on project expected to generate a
4 1 percent annual return. In that case,
Project A is unambiguously preferable.

IRRs assumptions about reinvestment can

lead to major capital budget distortions.

E ven if the interim cash flows really could

be reinvested at the IRR, very few

18 | McKinsey on Finance | Summer 2004

practitioners would argue that the value of

future investments should be commingled
with the value of the project being
evaluated. M ost practitioners would agree
that a companys cost of capitalby
definition, the return available elsewhere to
its shareholders on a similarly risky
investmentis a clearer and more logical
rate to assume for reinvestments of interim
project cash flows (Exhibit 1).
When the cost of capital is used, a projects
true annual equivalent yield can fall
significantlyagain, especially so with
projects that posted high initial IRRs. Of
course, when executives review projects
with IRRs that are close to a companys
cost of capital, the IRR is less distorted by
the reinvestment-rate assumption. But when
they evaluate projects that claim IRRs of
10 percent or more above their companys
cost of capital, these may well be
significantly distorted. Ironically, unadjusted
IRRs are particularly treacherous because
the reinvestment-rate distortion is most
egregious precisely when managers tend to
think their projects are most attractive. And
since this amplification is not felt evenly
across all projects,3 managers cant simply
correct for it by adjusting every IRR by a
standard amount.
How large is the potential impact of a
flawed reinvestment-rate assumption?
M anagers at one large industrial company
approved 23 major capital projects over five
years on the basis of IRRs that averaged
7 7 percent. Recently, however, when we
conducted an analysis with the reinvestment
rate adjusted to the companys cost of
capital, the true average return fell to just
16 percent. The order of the most attractive
projects also changed considerably. The topranked project based on IRR dropped to

the tenth-most-attractive project. M ost

striking, the companys highest-rated
projectsshowing IRRs of 800, 15 0, and
130 percentdropped to just 15 , 23, and
22 percent, respectively, once a realistic
reinvestment rate was considered (Exhibit 2).
Unfortunately, these investment decisions
had already been made. Of course, IRRs this
extreme are somewhat unusual. Yet even if a
projects IRR drops from 25 percent to
15 percent, the impact is considerable.

W hat to d o?
The most straightforward way to avoid
problems with IRR is to avoid it altogether.
Yet given its widespread use, it is unlikely to
be replaced easily. Executives should at the
very least use a modified internal rate of
return. While not perfect, M IRR at least
allows users to set more realistic interim
reinvestment rates and therefore to calculate
a true annual equivalent yield. Even then,
we recommend that all executives who
review projects claiming an attractive IRR
should ask the following two questions.
1. What are the assumed interimreinvestment rates? In the vast majority of
cases, an assumption that interim flows can
be reinvested at high rates is at best
overoptimistic and at worst flat wrong.
Particularly when sponsors sell their projects
as unique or the opportunity of a
lifetime, another opportunity of similar
attractiveness probably does not exist; thus
interim flows wont be reinvested at
sufficiently high rates. For this reason, the
best assumptionand one used by a proper
discounted cash-flow analysisis that
interim flows can be reinvested at the
companys cost of capital.
2 . A re interim c ash f lo w s b iased to w ard
the start o r the end o f the p ro jec t? Unless

Internal rate of return: A cautionary tale | 19

exhibit 2

A rude surprise
%
850

800

200

150

100

50

0
Project1 1

10

11

12

13

14

15

S tandard IR R calculation

16

17

18

19

20

21

22

23

M odified IR R with reinvestment at cost of cap ital

1 Disguised example of large industrial company.

the interim reinvestment rate is correct (in

other words, a true reinvestment rate rather
than the calculated IRR), the IRR distortion
will be greater when interim cash flows
occur sooner. This concept may seem
counterintuitive, since typically we would
prefer to have cash sooner rather than later.
The simple reason for the problem is that
the gap between the actual reinvestment
rate and the assumed IRR exists for a longer
period of time, so the impact of the
distortion accumulates.4

Despite flaws that can lead to poor

investment decisions, IRR will likely
continue to be used widely during capitalbudgeting discussions because of its strong
intuitive appeal. Executives should at least
cast a skeptical eye at IRR measures before
making investment decisions. MoF
T h e a uth o rs w is h to th a n k R o b M c N is h fo r h is
a s s is ta n c e in d ev elo p in g th is a rtic le.

John Kelleher (J o h n _ K elleh er@ M c K in s ey .c o m)

and Ju s tin M a c C orm a c k (J us tin _ M a c C o rma c k @
M c K in s ey .c o m) are consultants in McKinseys
Toronto offi ce.

J o h n R o b ert G ra h a m a n d C a mp b ell R . H a rv ey , T h e th eo ry
a n d p ra c tic e o f c o rp o ra te fi n a n c e: E v id en c e fro m th e fi eld ,
D uk e U n iv ers ity w o rk in g p a p er p res en ted a t th e 2001
a n n ua l meetin g o f th e A meric a n F in a n c e A s s o c ia tio n , N ew
O rlea n s (a v a ila b le a t h ttp ://s s rn .c o m/a b s tra c t= 22025 1).

A s a res ult o f a n a rc a n e ma th ema tic a l p ro b lem, IR R c a n

g en era te tw o v ery d ifferen t v a lues fo r th e s a me p ro jec t
w h en future c a s h fl o w s s w itc h fro m n eg a tiv e to p o s itiv e (o r
p o s itiv e to n eg a tiv e). A ls o , s in c e IR R is ex p res s ed a s a p erc en ta g e, it c a n ma k e s ma ll p ro jec ts a p p ea r mo re a ttra c tiv e
th a n la rg e o n es , ev en th o ug h la rg e p ro jec ts w ith lo w er IR R s
c a n b e mo re a ttra c tiv e o n a n N P V b a s is th a n s ma ller p ro jec ts w ith h ig h er IR R s .

T h e a mp lifi c a tio n effec t g ro w s a s a p ro jec ts fun d a men ta l

h ea lth imp ro v es , a s mea s ured b y N P V , a n d it v a ries
d ep en d in g o n th e un iq ue timin g o f a p ro jec ts c a s h fl o w s .

In teres tin g ly , g iv en tw o p ro jec ts w ith id en tic a l IR R s , a

p ro jec t w ith a s in g le b ullet c a s h fl o w a t th e en d o f th e
in v es tmen t p erio d w o uld b e p refera b le to a p ro jec t w ith
in terim c a s h fl o w s . T h e rea s o n : a la c k o f in terim c a s h
fl o w s c o mp letely immun iz es a p ro jec t fro m th e
rein v es tmen t- ra te ris k .