Burch Conference, June 2009 Day 2: Behavioral Evidence Investment Location Profit Shifting Tax Payments and Location of Expenses Debt-Equity Decisions Repatriations
Bur
ch C
onfe
renc
e, J
une
2009
Day
2: B
ehav
iora
l Evi
denc
e
Inves
tmen
t Loc
ation
Profi
t Shif
ting
Tax P
ayme
nts an
d Loc
ation
of E
xpen
ses
Debt-
Equit
y Dec
ision
sRe
patria
tions
The
impa
ct o
f tax
atio
n.
!U
ntil 20 y
ears
ago,
our
under
stan
din
g o
f th
e im
pac
t of in
tern
atio
nal
tax
atio
n w
as r
ather
a
muddle
.!
In p
articu
lar,
ser
ious
peo
ple
voic
ed s
erio
us
ques
tions
about
whet
her
inte
rnat
ional
tax
atio
n
had
any
impac
t on inve
stm
ent
and o
ther
asp
ects
of firm
beh
avio
r.!
What
chan
ged
? N
ew d
ata
bec
ame
avai
lable
, but
more
im
port
antly,
peo
ple
beg
an looki
ng a
t th
e dat
a m
ore
rig
oro
usl
y.!
Nobody
now
ask
s w
het
her
tax
atio
n m
atte
rs;
we
now
ask
how
much
tax
atio
n m
atte
rs,
and h
ow
the
effe
cts
vary
with c
ircu
mst
ance
s.
Wha
t we
know
abo
ut th
e im
pact
of
taxa
tion
on F
DI.
!The
bott
om
lin
e: v
irtu
ally
all
the
relia
ble
ev
iden
ce p
oin
ts t
o lar
ge
effe
cts
of
taxa
tion o
n t
he
loca
tion,
volu
me,
and
char
acte
r of fo
reig
n d
irec
t in
vest
men
t (F
DI)
.!
Eve
n g
ranting t
he
pre
mis
e th
at t
axat
ion
has
im
port
ant
effe
cts,
it
is s
till
import
ant
to k
now
the
asso
ciat
ed m
agnitudes
of
beh
avio
ral re
sponse
s, s
ince
thes
e ar
e cr
itic
al t
o a
sses
sing t
he
impac
t of ta
x polic
ies
in p
ract
ice,
and t
o o
ptim
al t
ax
des
ign.
Old
er e
vide
nce.
!Ver
y old
evi
den
ce (
e.g.,
1950s)
com
es fro
m
surv
eys,
in w
hic
h m
anag
ers
indic
ate
that
they
m
ost
ly ignore
tax
atio
n iss
ues
."
Consi
sten
t w
ith F
DI
beh
avio
r in
the
1950s,
when
FD
I w
as c
once
ntr
ated
in h
igh-t
ax c
ountr
ies
in
Euro
pe,
USA,
Can
ada,
etc
."
Moder
n s
urv
eys
show
just
the
opposi
te,
with
man
ager
s co
nsi
sten
tly
ranki
ng t
ax c
onsi
der
atio
ns
as
one
of th
e to
p t
wo o
r th
ree
fact
ors
(al
ong w
ith
cost
s) d
eter
min
ing inve
stm
ent
loca
tion.
!O
lder
quan
tita
tive
evi
den
ce c
om
es f
rom
tim
e se
ries
, sh
ow
s th
at F
DI
is p
osi
tive
ly a
ffec
ted b
y af
ter-
tax
rate
s of
retu
rn.
This
evi
den
ce is,
how
ever
, open
to m
ultip
le inte
rpre
tations.
!Ther
e is
an im
port
ant
issu
e of how
one
thin
ks
about,
and m
easu
res,
the
burd
en o
f ta
xation.
The
early
mod
ern
stud
ies.
!Cro
ss-s
ectional
evi
den
ce (
e.g.,
G
ruber
tan
d M
utt
i, R
ESta
t1991 o
r H
ines
and R
ice,
QJE
1994):
U.S
. firm
s te
nd t
o inve
st in low
tax
fo
reig
n loca
tions.
They
do s
o f
or
two r
easo
ns:
"Kee
p a
hig
her
fra
ctio
n o
f pro
fits
ear
ned
th
ere.
"Can
rea
lloca
te t
axab
le inco
me
among
loca
tions.
!Evi
den
ce f
or
fore
ign t
ax h
aven
s.
Who
and
wha
t are
tax
have
ns?
!Tax
hav
ens
are
low
-tax
loca
tions
that
att
empt
to
attr
act
inve
stm
ent.
!N
ot
a new
phen
om
enon:
the
sam
e co
untr
ies
that
w
ere
maj
or
tax
hav
ens
in 1
980 a
re m
ajor
tax
hav
ens
today
.!
The
maj
or
tax
hav
ens
incl
ude
most
ly s
mal
l co
untr
ies;
only
sev
en h
ad p
opula
tions
exce
edin
g o
ne
mill
ion in 1
982.
!Am
eric
an inve
stm
ent
in 1
982 �
and in 1
999 �
was
co
nce
ntr
ated
in t
ax h
aven
s out
of pro
port
ion t
o t
he
econom
ic s
izes
of th
ese
countr
ies.
(S
epar
atel
y,
Am
eric
an firm
s re
port
ed v
ery
hig
h inco
me
leve
ls in
thes
e pla
ces.
)
Who
are
the
maj
or ta
x ha
vens
?A
ndor
raG
rena
daN
auru
!A
ngui
llaH
ong
Kon
g *
Net
herla
nds A
ntill
esA
ntig
ua a
nd B
arbu
daIr
elan
d *
Niu
e !
Aru
ba !
Isle
of M
anPa
nam
aB
aham
asJo
rdan
*Sa
int K
itts a
nd N
evis
Bah
rain
Leba
non
*Sa
int L
ucia
Bar
bado
sLi
beria
Sain
t Vin
cent
and
the
Gre
nadi
nes
Bel
ize
Liec
hten
stei
nSa
moa
!B
erm
uda
Luxe
mbo
urg
*Sa
n M
arin
o !
Brit
ish
Virg
in Is
land
sM
acao
*Se
yche
lles !
Cay
man
Isla
nds
Mal
dive
sSi
ngap
ore
*C
hann
el Is
land
sM
alta
Switz
erla
nd *
Coo
k Is
land
sM
arsh
all I
slan
dsTo
nga
!C
ypru
sM
aurit
ius !
Turk
s and
Cai
cos I
slan
dsD
omin
ica
Mon
aco
Van
uatu
Gib
ralta
rM
onts
erra
tV
irgin
Isla
nds (
U.S
.) !
! Not
incl
uded
in H
ines
-Ric
e ta
x ha
ven
list.
* N
ot in
clud
ed in
OEC
D ta
x ha
ven
list.
Inve
stm
ent l
evel
s as
func
tions
of
loca
l tax
es.
!To w
hat
ext
ent
do c
ountr
ies
with low
er t
ax r
ates
re
ceiv
e gre
ater
am
ounts
of
inbound f
ore
ign
inve
stm
ent
and a
ssoci
ated
em
plo
ymen
t?!
Hin
es a
nd R
ice
(1994)
anal
yze
U.S
. dat
a, f
indin
g
signific
ant
effe
cts:
1%
low
er t
ax r
ates
are
as
soci
ated
with 3
% g
reat
er loca
l use
of
pro
per
ty
pla
nt
and e
quip
men
t an
d 3
% g
reat
er e
mplo
ymen
t (a
s m
easu
red b
y w
ages
).!
Oth
er p
aper
s find s
imila
r pat
tern
s; A
ltsh
ule
ran
d
Gru
ber
tre
port
that
effec
ts incr
ease
ove
r tim
e.!
This
corr
elat
ion d
oes
not
PRO
VE t
hat
tax
es
influen
ce inve
stm
ent
to t
his
ext
ent,
sin
ce o
ther
th
ings
diffe
r bet
wee
n c
ountr
ies,
but
it is
hig
hly
su
gges
tive
, par
ticu
larly
giv
en o
ther
evi
den
ce.
Som
e re
leva
nt m
ultin
atio
nal d
ecis
ions
!D
iscr
ete
choic
es:
"To s
erve
fore
ign m
arke
t th
rough e
xport
or
pro
duct
ion a
bro
ad"
To s
hift
pro
duct
ion a
bro
ad for
cost
rea
sons
"W
hic
h �
and h
ow
man
y -
fore
ign loca
tions
to
choose
"To e
xpan
d b
y ac
quis
itio
n o
r gre
enfiel
din
vest
men
t!
Continuous
choic
es:
"Conditio
nal
on loca
tion,
how
much
to inve
st"
Conditio
nal
on �
real
� ac
tivi
ty,
how
to s
hift
pro
fit
thro
ugh u
se o
f deb
t, t
ransf
er p
rici
ng,
etc
"Choic
e of div
iden
d r
epat
riat
ion
Loca
tion
Cho
ice
!2 s
tatist
ical
appro
aches
use
d s
o f
ar1.
Choose
exa
ctly
one
of a
num
ber
of poss
ible
lo
cations
(Dev
ereu
x &
Griffith,
1998)
!This
does
n�t a
llow
for
com
pan
ies
to b
e pre
sent
in s
ever
al c
ountr
ies
sim
ultan
eousl
y2.
Choose
to loca
te o
r not
loca
te in e
ach p
lace
(B
uet
tner
& R
uf,
2006,
Bar
rios
et a
l, 2
008)
!This
ass
um
es e
ach c
hoic
e in
dep
enden
t of
the
oth
ers
!N
eed t
o m
odel
num
ber
of
loca
tions
chose
n,
and c
onditio
n o
n t
hat
, w
hic
h loca
tions
Dat
a
!Aggre
gat
e ca
pital
sto
ck!
Flow
s of
Fore
ign D
irec
t In
vest
men
t"
Aggre
gat
e an
d b
ilate
ral
!Cap
ital
ow
ned
abro
ad b
y U
S c
om
pan
ies
"BEA,
split
by
countr
y an
d indust
ry!
Firm
lev
el d
ata
"Confiden
tial
dat
a on fore
ign a
ffili
ates
and p
aren
ts
(BEA,
US T
reas
ury
, Bundes
ban
k, O
NS)
"Conso
lidat
ed a
ccounting d
ata
(Com
pust
atan
d
Dat
astr
eam
)"
Unco
nso
lidat
ed a
ccounting d
ata,
with o
wner
ship
in
form
atio
n (
Am
adeu
s an
d O
rbis
)
Agg
rega
te v
Mic
ro D
ata
!M
ost
ly inte
rest
ed in a
ggre
gat
e ef
fect
s"
How
much
cap
ital
is
ther
e is
in c
ountr
y i, a
nd w
ho
ow
ns
it?
"Aggre
gat
e dat
a ca
n e
nco
mpas
s eq
uili
brium
re
sponse
s eg
. H
ow
tax
rat
es a
ffec
t ag
gre
gat
e ca
pital
sto
ck!
Mic
ro d
ata
"Can
addre
ss t
he
separ
ate
dec
isio
ns
that
add u
p t
o
aggre
gat
e ef
fect
s"
In p
rinci
ple
, al
low
s us
to iden
tify
more
pre
cise
ly
role
of diffe
rent
aspec
ts o
f ta
x sy
stem
"
Is m
ore
det
aile
d,
so w
e ca
n c
ontr
ol fo
r m
ore
fa
ctors
"Allo
ws
us
to iden
tify
diffe
rence
s in
beh
avio
ur
bet
wee
n t
ypes
of co
mpan
ies
Agg
rega
te v
Mic
ro D
ata
!M
icro
dat
a"
Is m
uch
more
effec
tive
if w
e ca
n iden
tify
var
iation
in e
ffec
tive
tax
rat
es a
cross
com
pan
ies
!But
usu
ally
inte
rest
ed in n
atio
nal
tax
sys
tem
"Var
iation c
lear
ly d
oes
occ
ur
thro
ugh c
hoic
e of
asse
ts,
deb
t, e
tc �
and r
ate
of pro
fit,
but
thes
e m
ay
all be
corr
elat
ed w
ith t
he
indep
enden
t va
riab
le,
so
endogen
eity
pro
ble
m"
Exa
mple
is
com
monly
use
d �
effe
ctiv
e ta
x ra
te�
ie.
tax
char
ge
as p
roport
ion o
f pre
-tax
pro
fit
Mea
surin
g ta
x (1
)
!D
iscr
ete
choic
es s
hould
dep
end o
n c
om
par
ison
of
post
-tax
pro
fit
of
each
option -
affe
cted
by
an
aver
age
tax
rate
!2 a
ppro
aches
1.
Bac
kwar
d-l
ooki
ng:
ratio o
f ta
x ch
arge
to p
re-t
ax
pro
fit
"D
epen
ds
on p
ast
and c
urr
ent
inve
stm
ent,
pro
fit,
deb
t, e
tc2.
Forw
ard-l
ooki
ng:
for
hyp
oth
etic
al inve
stm
ent
pro
ject
, ra
tio o
f N
PV o
f ta
x to
NPV
of pre
-tax
pro
fit
"N
eces
sarily
a s
implif
icat
ion o
f ac
tual
tax
posi
tion
Mea
surin
g ta
x (2
)
!Continuous
choic
es a
ffec
ted b
y m
argin
al
consi
der
atio
ns
"Affec
ted b
y a
mar
gin
al t
ax r
ate
"In
vest
men
t af
fect
ed b
y an
effec
tive
mar
gin
al t
ax
rate
s, t
akin
g into
acc
ount
allo
wan
ce r
ates
as
wel
l as
st
atuto
ry r
ates
!Can
com
pute
this
for
hyp
oth
etic
al inve
stm
ent
pro
ject
"Lo
cation o
f pro
fit
dep
ends
at t
he
mar
gin
on
diffe
rence
s in
sta
tuto
ry r
ates
Impa
ct o
n in
vest
men
t in
the
U.S
.
!U
.S.
stat
es t
ax c
orp
ora
te inco
me
at r
ates
that
va
ry fro
m 0
% t
o 1
2%
.!
We
can c
om
par
e in
vest
men
ts in U
.S.
stat
es f
rom
co
untr
ies
that
exe
mpt
fore
ign inco
me
from
ta
xation w
ith inve
stm
ents
fro
m c
ountr
ies
that
tax
fo
reig
n inco
me
and p
rovi
de
fore
ign t
ax c
redits.
!The
seco
nd g
roup s
hould
be
less
sen
sitive
to s
tate
ta
x diffe
rence
s, w
her
eas
the
firs
t gro
up s
hould
be
extr
emel
y se
nsi
tive
.!
For
exam
ple
, Fr
ench
inve
stm
ent
should
be
more
at
trac
ted t
o T
exas
, w
her
e th
ere
is n
o s
tate
co
rpora
te t
ax,
than
should
British
inve
stm
ent,
si
nce
the
U.K
. ta
xes
the
Tex
as inco
me.
!The
evid
ence
(H
ines
, �A
lter
ed S
tate
s,�
Am
eric
an
Eco
nom
ic R
evie
w,
1996)
is c
onsi
sten
t w
ith t
his
.
Evi
denc
e in
Hin
es (1
996)
.
!Com
par
ing o
nly
zer
o-t
ax s
tate
s to
the
hig
hes
t-ta
x st
ates
, in
vest
ors
fro
m e
xem
ption c
ountr
ies
tend
to c
once
ntr
ate
thei
r U
.S.
inve
stm
ents
in t
he
zero
-ta
x st
ates
. J
apan
ese
and B
ritish
inve
stors
, on t
he
oth
er h
and,
diffe
r lit
tle
bet
wee
n t
hes
e st
ates
.!
The
regre
ssio
n a
nal
ysis
indic
ates
that
1%
sta
te
tax
rate
diffe
rence
s ar
e as
soci
ated
with 9
-11%
diffe
rence
s bet
wee
n t
he
inve
stm
ent
inte
nsi
ties
of
exem
ption inve
stors
and t
hose
fro
m c
ountr
ies
that
tax
worldw
ide
inco
me.
!The
sim
ple
st e
mpiric
al e
xerc
ise
is s
imply
to look
at n
um
ber
s of
busi
nes
ses
that
fore
ign inve
stors
ow
n in e
ach s
tate
. H
ere
too t
her
e ar
e im
port
ant
tax
effe
cts:
1%
hig
her
tax
es a
re a
ssoci
ated
with
3%
few
er a
ffili
ates
of
exem
ption c
ountr
y in
vest
ors
com
par
ed t
o w
orldw
ide
inve
stors
.
Are
tax
conc
essi
ons
to fo
reig
n in
vest
ors
effe
ctiv
e?
!Pa
per
: H
ines
, �T
ax s
par
ing a
nd d
irec
t in
vest
men
t in
dev
elopin
g c
ountr
ies,
� 2001.
!�T
ax s
par
ing�
is t
he
pra
ctic
e of
gra
nting f
ore
ign
tax
cred
its
for
taxe
s th
at w
ould
hav
e bee
n p
aid
in
the
abse
nce
of
spec
ial co
nce
ssio
ns
to f
ore
ign
inve
stors
. �
Tax
spar
ing�
incr
ease
s th
e im
pac
t of
loca
l ta
x co
nce
ssio
ns
to inve
stors
fro
m c
ountr
ies
that
tax
worldw
ide
inco
mes
.!
Com
par
e Ja
pan
ese
and A
mer
ican
inve
stm
ent.
"Ja
pan
offer
s ta
x sp
arin
g,
the
U.S
. does
not.
"Ja
pan
ese
firm
s ap
pea
r to
rec
eive
tax
conce
ssio
ns
in
countr
ies
with s
par
ing a
gre
emen
ts.
"G
reat
er J
apan
ese
inve
stm
ent
conce
ntr
atio
n r
esults.
"Effec
ts p
ersi
st w
hen
usi
ng U
K t
ax s
par
ing a
s an
in
stru
men
t fo
r Ja
pan
ese
tax
spar
ing.
AVER
AGE
TAX
RATE
0
0.1
0.2
0.3
0.4
0.5
0.6
NO S
PARI
NGSP
ARIN
GJA
PANE
SE T
AX T
REAT
Y PR
OVIS
ION
ARITHEMETIC MEAN OF AGGREGATE AVERAGE TAX
RATES
US JAPA
N
"TAX
SPA
RIN
G"
AND
FD
I
0
0.00
2
0.00
4
0.00
6
0.00
8
0.01
0.01
2
0.01
4
0.01
6
NO
SPA
RIN
GSP
ARIN
G
JAPA
NES
E TA
X TR
EATY
PR
OVI
SIO
N
SHARE OF FDI
US
JAPA
N
Hom
e co
untry
tax
effe
cts
on F
DI.
!Pa
per:
�Inter
natio
nal ta
xatio
n and
mult
inatio
nal fi
rm
locati
on de
cision
s,� O
xford
Cen
tre fo
r Bus
iness
Ta
xatio
n wor
king p
aper
08/25
, Octo
ber 2
008.
!Au
thors:
Salv
ador
Bar
rios (
EC),
Harry
Huiz
inga
(Tilb
urg U
niver
sity),
Luc L
aeve
n(IM
F), a
nd G
aetan
Nico
deme
(EC)
.!
Data
(Ama
deus
) on E
urop
ean m
ultina
tiona
ls in
33
Euro
pean
coun
tries,
1999
-200
3.!
Focu
ses o
n the
impa
ct of
home
tax r
egim
es.
The
met
hod
of a
naly
sis.
!Th
e pap
er ta
kes t
he sa
mple
of all
new
foreig
n su
bsidi
aries
estab
lishe
d betw
een 1
999 a
nd 20
03,
and a
sks:
"W
hat fa
ctors
deter
mine
whe
re th
e sub
sidiar
y is
locate
d?"
Wha
t facto
rs de
termi
ne th
e hom
e cou
ntry o
f the p
aren
t of
that s
ubsid
iary.
!Th
e stat
istica
l pro
blems
are i
mmen
se, a
nd th
e pa
per t
akes
only
a sim
ple pa
ss at
them
; no
nethe
less,
the re
sults
are i
ntere
sting
.
Find
ings
.
!1%
high
er lo
cal ta
xes r
educ
e the
prob
abilit
y of a
subs
idiar
y be
ing lo
cated
ther
e by 0
.6% (c
ontro
lling f
or ec
onom
ic an
d po
litica
l con
dition
s, ge
ogra
phic
prox
imity
, othe
r fac
tors).
!Cr
oss-b
orde
r tax
es (d
ivide
nd w
ithho
lding
taxe
s and
home
co
untry
taxe
s on f
oreig
n inc
ome)
have
sign
ifican
t and
sli
ghtly
larg
er ne
gativ
e effe
cts (1
% lo
wer p
roba
bility
for 1
%
highe
r tax
es) t
han l
ocal
effec
tive t
axes
.!
Turn
ing th
e meth
od ar
ound
, the p
aper
cons
iders
who o
wns
an af
filiate
. 1%
high
er ta
xes o
n own
ersh
ip of
an af
filiate
re
duce
s by 3
% th
e like
lihoo
d tha
t a pa
rent
will o
wn it.
Are
tax
effe
cts
unifo
rm?
!The
stat
istica
l an
alys
is t
ypic
ally
im
pose
s th
at a
giv
en
tax
rate
diffe
rence
has
the
sam
e ef
fect
eve
ryw
her
e;
put
diffe
rently,
em
piric
al w
ork
looks
for
aver
age
tax
effe
cts.
!It
is
unlik
ely
to b
e th
e ca
se t
hat
tax
rat
e diffe
rence
s hav
e th
e sa
me
effe
cts
in a
ll ci
rcum
stan
ces.
!D
har
map
ala
and H
ines
(D
ay 4
rea
din
gs)
fin
d t
hat
ta
x ef
fect
s hav
e m
uch
str
onger
effec
ts o
n U
.S.
fore
ign d
irec
t in
vest
men
t am
ong c
ountr
ies
that
sc
ore
wel
l on W
orld B
ank
gove
rnan
ce indic
ators
.
Doubtles
s th
is r
efle
cts
that
inve
stors
may
fee
l th
at
low
tax
rat
es d
o t
hem
little
good in s
om
e par
ts o
f th
e w
orld.
!W
e nee
d m
ore
anal
ysis
of w
hen
and h
ow
tax
rat
es
hav
e diffe
rential
effec
ts o
n inve
stm
ent.
Ratio
of U
S FD
I to
GDP
for 4
Gro
ups o
f Cou
ntrie
s
0 0.1
0.2
0.3
0.4
0.5
0.6
0.7
Poor
ly-Go
vern
ed, L
ow T
ax P
oorly
-Gov
erne
d, Hi
gh T
axW
ell-G
over
ned,
Low
Tax
Well
-Gov
erne
d, Hi
gh T
ax
Ratio
of T
otal F
DI to
To
tal G
DP
Pro
fit s
hifti
ng.
!Ther
e is
consi
der
able
evi
den
ce t
hat
the
loca
tion o
f ta
xable
inco
me
is s
ensi
tive
to n
atio
nal
tax
rat
e diffe
rence
s.!
This
evi
den
ce c
om
es in s
ever
al form
s."
Hig
her
rep
ort
ed p
rofit
rate
s in
low
-tax
loca
tions.
"G
reat
er r
oya
lty
and inte
rest
pay
men
ts t
o r
elat
ed p
arties
(a
nd a
lso t
o u
nre
late
d p
arties
) in
hig
h-t
ax loca
tions.
"Fo
reig
n a
ffili
ates
in low
-tax
loca
tions
run lar
ge
report
ed
trad
e su
rplu
ses
with r
elat
ed p
arties
, co
mpar
ed t
o a
ffili
ates
in
hig
h-t
ax loca
tions.
"D
irec
t ev
iden
ce t
hat
price
s ch
arged
for
rela
ted p
arty
tr
ansa
ctio
ns
seem
to b
e af
fect
ed b
y ta
x ra
te d
iffe
rence
s."
None
of
the
stat
istica
l ev
iden
ce p
rove
s th
at a
nyo
ne
is
viola
ting t
he
law
, th
ough it
is c
onsi
sten
t w
ith m
odel
s of
avoid
ance
that
hav
e th
at f
eatu
re.
Pro
fit ra
tes
and
taxe
s.
!Sev
eral
stu
die
s lo
ok
at p
rofit
rate
s of
fore
ign
affilia
tes
as f
unct
ions
of
loca
l ta
x ra
tes.
!The
conce
pt
is t
hat
, if f
irm
s re
allo
cate
tax
able
in
com
e, t
hen
hig
her
tax
rat
es s
hould
be
asso
ciat
ed w
ith low
er r
eport
ed p
rofit
rate
s.!
The
difficu
lty
is t
hat
you d
o n
ot
know
what
pro
fit
rate
s w
ould
be
in t
he
abse
nce
of m
anip
ula
tion.
But
it is
poss
ible
to o
bse
rve
inputs
such
as
capital
an
d lab
or,
contr
ol fo
r th
ese,
and s
ee t
o w
hat
ex
tent
the
diffe
rence
s th
at r
emai
n a
re c
orr
elat
ed
with t
ax r
ates
.!
Eve
ryone
(e.g
., G
ruber
tan
d M
utt
i, 1
991;
Hin
es
and R
ice,
1994;
Huiz
inga
and L
aeve
n,
2008)
report
s th
at h
igh t
axes
go w
ith low
pro
fit
rate
s.
Hui
zing
a an
d La
even
stud
y.
!A s
imple
model
of ta
x av
oid
ance
thro
ugh
inco
me
shifting.
!D
ata
for
firm
s in
33 E
uro
pea
n c
ountr
ies
in
1999;
they
look
at t
he
det
erm
inan
ts o
f ea
rnin
gs
bef
ore
inte
rest
and t
axes
, co
ntr
olli
ng for
oth
er inputs
.!
They
rep
ort
sig
nific
ant
effe
cts
of ta
xes
on
report
ed p
rofits
, co
nsi
sten
t w
ith (
though
smal
ler
than
) th
e re
sults
of H
ines
and
Ric
e (1
994)
for
U.S
. dat
a.
Mor
e ev
iden
ce o
n pr
ofit
shift
ing.
!Kim
Cla
usi
ng
(2001)
finds
that
10%
low
er a
ffili
ate
tax
rate
s ar
e as
soci
ated
with 4
.4%
gre
ater
tra
de
surp
luse
s w
ith t
he
U.S
. par
ent
com
pan
y;
furt
her
more
, af
filia
tes
in low
-tax
countr
ies
hav
e gre
ater
rep
ort
ed e
xport
s to
rel
ated
aff
iliat
es.
!Cla
usi
ng
(2003)
finds
that
1%
low
er a
ffili
ate
tax
rate
s ar
e as
soci
ated
with 2
% h
igher
price
s fo
r U
.S.
goods
import
s (r
elat
ive
to p
rice
s pai
d t
o
unre
late
d p
arties
for
the
sam
e ca
tegory
of
goods)
an
d 1
.8%
low
er e
xport
price
s.!
Har
ris,
Morc
k, S
lem
rod a
nd Y
eung
report
that
U
.S.
firm
s w
ith t
ax h
aven
affili
ates
hav
e lo
wer
U
.S.
taxe
s/U
.S.
sale
s th
an d
o o
ther
firm
s.
Use
of d
ebt.
!It
can
not
com
e as
a s
urp
rise
that
firm
s use
gre
ater
deb
t in
hig
h-t
ax loca
tions.
!Evi
den
ce o
n f
inan
cing is
report
ed in D
esai
, Fo
ley
and H
ines
, �A
mulit
nat
ional
per
spec
tive
on c
apital
st
ruct
ure
choic
e an
d inte
rnal
cap
ital
mar
kets
,�
Journ
al o
f Fi
nan
ce,
Dec
ember
2004.
!Evi
den
ce for
U.S
. firm
s in
1982-1
994.
!10%
hig
her
loca
l ta
x ra
tes
are
asso
ciat
ed w
ith
2.8
% g
reat
er d
ebt/
asse
t ra
tios
within
the
sam
e firm
s.!
The
elas
tici
ty o
f deb
t/as
sets
is
gre
ater
for
borr
ow
ing f
rom
U.S
. par
ent
com
pan
ies
(0.3
5)
than
borr
ow
ing f
rom
unre
late
d p
arties
(0.1
9).
Mor
e on
deb
t.
!D
esai
, Fo
ley
and H
ines
rep
ort
that
U.S
. firm
s use
par
ent
com
pan
y deb
t to
subst
itute
for
unav
aila
bili
ty o
r co
stlin
ess
of
3rd
par
ty d
ebt
in
man
y co
untr
ies.
!1%
hig
her
loca
l in
tere
st r
ates
due
to leg
al a
nd
capital
mar
ket
conditio
ns
are
asso
ciat
ed w
ith
1.3
% r
educe
d e
xter
nal
borr
ow
ing a
nd 0
.8%
gre
ater
par
ent
lendin
g (
both
as
a fr
action o
f as
sets
).!
Part
of th
e ben
efit o
f a
multin
atio
nal
str
uct
ure
is
the
abili
ty t
o d
eplo
y ca
pital
aro
und t
he
world t
o
ove
rcom
e lo
cal pro
ble
ms.
This
in t
urn
car
ries
tax
im
plic
atio
ns,
though m
ay b
e only
par
tly
tax
motiva
ted t
o s
tart
with.
An
inte
rest
ing
debt
epi
sode
.
!Le
slie
Pap
ke(2
000)
docu
men
ts t
he
U.S
. han
dlin
g
of th
e ta
x tr
eaty
bet
wee
n t
he
U.S
. an
d t
he
Net
her
lands
Antille
s.!
Prio
r to
1984,
the
U.S
. tr
eaty
with t
he
N.A
. pro
vided
zer
o w
ithhold
ing o
n p
ort
folio
inte
rest
, m
akin
g t
he
N.A
. th
e id
eal co
nduit f
or
borr
ow
ing
by
issu
ing b
eare
r bonds
(!)
on t
he
Euro
dolla
r m
arke
t. The
only
oth
er w
ay t
o g
et z
ero
withhold
ing w
as w
ith t
reat
y par
tner
s w
ho
iden
tified
the
lender
s.!
Aft
er 1
984,
the
U.S
. re
pea
led w
ithhold
ing t
axes
on p
ort
folio
inte
rest
for
ever
yone.
!As
a re
sult,
borr
ow
ing t
hro
ugh t
he
N.A
. al
l but
dis
appea
red,
to b
e re
pla
ced b
y direc
t borr
ow
ing.
How
effe
ctiv
e ar
e m
ultin
atio
nal f
irms
at
avoi
ding
taxe
s?
!They
obvi
ousl
y dev
ote
ener
gy
and
reso
urc
es t
o a
void
ing t
axes
.!
But:
corp
ora
tions
around t
he
world
(incl
udin
g in t
he
U.S
.) p
ay q
uite
a bit in
taxe
s.
So it
appea
rs t
hat
ther
e ar
e si
gnific
ant
limits
to t
hei
r ab
ilities
to a
void
.!
Note
that
this
is
consi
sten
t w
ith t
he
inve
stm
ent
evid
ence
: if a
void
ance
wer
e co
stle
ss a
nd p
erfe
ct,
then
ther
e sh
ould
be
no e
ffec
t of ta
xation o
n F
DI.
Rep
atria
tions
.
!The
U.S
. sy
stem
of ta
xing fore
ign inco
me
on r
epat
riat
ion c
reat
es ince
ntive
s to
ch
oose
the
tim
ing o
f re
pat
riat
ions
sele
ctiv
ely
in o
rder
to m
inim
ize
asso
ciat
ed
tax
oblig
atio
ns.
"This
consi
der
atio
n b
ecom
es m
ore
im
port
ant
at
low
er fore
ign t
ax r
ates
."
Note
that
, if r
epat
riat
ion t
axes
are
not
tim
e-va
ryin
g,
then
(per
hap
s co
unte
rintu
itiv
ely)
the
repat
riat
ion t
ax d
oes
not
affe
ct t
he
tim
ing o
f re
pat
riat
ions.
!Am
eric
an firm
s re
pat
riat
e ro
ughly
hal
f of
thei
r fo
reig
n p
rofits
on a
vera
ge
each
yea
r.
Tax
effe
cts.
!H
ow
should
we
thin
k ab
out
avoid
ance
?"
Altsh
ule
rand G
ruber
t(2
003)
argue
that
fir
ms
ought
to b
e ab
le t
o a
void
rep
atri
atio
n t
axes
altoget
her
with f
inan
cial
tr
ansa
ctio
ns.
"N
ote
that
, ev
en if
firm
s su
cces
sfully
avo
ided
all
repat
riat
ion t
axes
, it w
ould
not
follo
w t
hat
the
repat
riat
ion
taxe
s w
ere
not
burd
enso
me
to t
hem
.!
Do A
mer
ican
firm
s del
ay r
epat
riat
ion t
o a
void
the
asso
ciat
ed t
axes
?"
Hin
es a
nd H
ubbar
d (
1990)
and D
esai
, Fo
ley
and H
ines
(2
001)
report
that
fir
ms
are
leas
t lik
ely
to r
epat
riat
e $$
from
low
-tax
fore
ign loca
tions;
DFH
als
o f
ind t
hat
this
ap
pea
rs f
or
fore
ign s
ubsi
dia
ries
but
not
for
fore
ign
bra
nch
es,
consi
sten
t w
ith t
ax ince
ntive
s."
The
stro
nges
t ev
iden
ce c
om
es f
rom
the
2004 H
IA,
a one-
tim
e 85%
exc
lusi
on o
f div
iden
ds
rece
ived
in 2
005.
Wha
t hap
pene
d in
200
5?
!A s
urg
e in
rep
atriat
ions:
wen
t fr
om
about
$60b/y
ear
to $
362b in 2
005.
!The
Dhar
map
ala,
Fole
y an
d F
orb
es a
sk w
hat
firm
s did
with r
epat
riat
ed funds.
"The
HIA
req
uired
that
the
funds
be
use
d for
per
mitte
d inve
stm
ents
: in
vest
men
t, R
&D
, new
em
plo
ymen
t, c
erta
in a
cquis
itio
ns.
"H
IA funds
could
not
be
use
d for
div
iden
ds,
shar
e buyb
acks
, or
exec
utive
com
pen
sation.
!D
FF f
ind t
hat
rep
atriat
ions
wer
e not
asso
ciat
ed
with g
reat
er inve
stm
ent
or
emplo
ymen
t ex
pen
ses,
but
with p
ayouts
to s
har
ehold
ers:
$1 o
f re
pat
riat
ions
was
ass
oci
ated
with $
0.9
1 o
f sh
are
buyb
acks
and $
0.0
8 o
f div
iden
ds.
How
do
we
know
wha
t firm
s w
ould
ha
ve d
one
in th
e ab
senc
e of
HIA
?
!O
ne
of th
e pro
ble
ms
of in
fere
nce
is
that
firm
s th
at
wan
ted t
o d
o lots
of sh
are
buyb
acks
in 2
005
would
norm
ally
be
expec
ted t
o r
epat
riat
e m
ore
fr
om
thei
r fo
reig
n s
ubs
than
do o
ther
firm
s.!
DFF
use
an inst
rum
enta
l va
riab
les
stra
tegy
in
whic
h t
hey
pre
dic
t th
e lik
elih
ood o
f re
pat
riat
ion in
2005 o
n t
he
bas
is o
f:"
Hold
ing h
igh lev
els
of ca
sh a
bro
ad.
"U
se o
f ta
x hav
en h
old
ing c
om
pan
ies.
!The
resu
lts
appea
r to
be
sensi
tive
to t
he
use
of
thes
e in
stru
men
ts.
Wha
t do
exec
utiv
es s
ay?
!G
raham
, H
anlo
n a
nd S
hev
linsu
rvey
of firm
s w
ith
mem
ber
of th
e Tax
Exe
cutive
Inst
itute
; 744 o
ut
of
2806 r
esponded
(th
is is
a gre
at r
esponse
rat
e),
and a
fter
var
ious
excl
usi
ons
411 f
irm
s.!
They
rep
ort
that
ove
r 60%
of
repat
riat
ed f
unds
cam
e fr
om
ove
rsea
s ca
sh h
old
ings.
!The
two m
ost
com
mon r
eport
ed u
ses
of
the
funds
wer
e fo
r pay
ing d
ow
n d
ebt
and r
epurc
has
ing
shar
es;
the
third m
ost
com
mon u
se w
as f
or
U.S
. in
vest
men
t.!
Ver
y lit
tle
report
ed u
se o
f re
pat
riat
ed f
unds
for
div
iden
ds
or
shar
e buyb
acks
.
How
can
we
reco
ncile
thes
e re
spon
ses
with
the
econ
omet
ric e
vide
nce?
!The
trac
ing im
plic
it in t
he
HIA
does
little
to p
reve
nt
avoid
ance
.!
Gra
ham
, H
anlo
n a
nd S
hev
linas
ked
resp
onden
ts w
het
her
they
use
d �
free
d
up�
cash
to r
epurc
has
e sh
ares
, an
d m
any
report
that
they
did
(th
ough t
hey
rep
ort
oth
er u
ses
too,
incl
udin
g U
.S.
inve
stm
ent.
!N
ote
that
the
table
pre
sents
just
fra
ctio
ns
answ
erin
g �
yes,
� not
amounts
.
Effe
cts
of th
e H
IA o
n fu
ture
act
ions
.
!O
ne
of th
e pote
ntial
cost
s of a
�one
tim
e�
polic
y lik
e th
e H
IA is
that
it
may
dis
coura
ge
repat
riat
ions
as firm
s w
ait
for
the
nex
t one.
!G
raham
Han
lon a
nd S
hev
linfind t
hat
m
any
firm
s re
pat
riat
ing u
nder
the
HIA
to
ok
actions
to a
void
rep
atriat
ing
subse
quen
t fo
reig
n p
rofits
.!
Als
o,
man
y firm
s an
tici
pat
e th
at t
he
HIA
m
ight
be
repea
ted.