Employmen p o ec ion and globalisa ion in dynamic oligopoly
Ge da Dewi De mo Leahy Ca ia Mon agna
Na ional Uni e si y o
I eland Maynoo h Uni e si y College Dublin Uni e si y o Dundee
Janua y 2003
Abs ac : We cons uc a model in which oligopolis ic i ms decide whe e o loca e.
Fi ms choose o loca e ei he in a coun y whe e employmen p o ec ion implies
cos ly ou pu adjus men s o in one wi hou adjus men cos s. Using a wo-pe iod
h ee-s age game wi h unce ain y i is demons a ed ha loca ion is in luenced by
bo h lexibili y and s a egic conce ns. We show ha he s a egic e ec s unde
Cou no wo k owa ds domes ic ancho age in he coun y wi h adjus men cos s while
hose unde Be and do no . S a egic agglome a ion can occu in he in lexible
coun y unde Cou no and e en unde Be and p o ided unce ain y and o eign
di ec in es men cos s a e low.
Keywo ds: Unce ain y, Flexibili y, Oligopoly, Employmen P o ec ion, Fo eign
Di ec In es men , Loca ion.
JEL Codes: D80, F23, L13
App oxima e Wo d Coun (including ables, igu es and e e ences): 7850
Acknowledgemen s: We a e g a e ul o Da id Collie, Pe e Nea y, Donal O’Neill,
Oli e Swee man, pa icipan s o he Eu opean T ade S udy G oup (B ussels,
Sep embe 2001), pa icipan s o he I ish Economic Associa ion Con e ence (Ap il
2002), pa icipan s o he Eu opean Economic Associa ion Con e ence (Augus 2002),
semina pa icipan s a Uni e si y College Dublin, a he Na ional Uni e si y o
I eland Maynoo h and a he Uni e si y o Ghen .
Co espondence: Ge da Dewi , Na ional Uni e si y o I eland Maynoo h, Depa men
o Economics, Maynoo h, I eland, el.: (+)353-1-7083776, ax: (+)353-1-7083934, E-
mail: [email p o ec ed]; De mo Leahy, Uni e si y College Dublin, Depa men
o Economics, Bel ield, Dublin 4, I eland, el.: (+)353-1-7168551, ax: (+)353-1-
2830068, E-mail: [email p o ec ed]; Ca ia Mon agna, Uni e si y o Dundee,
Depa men o Economic S udies, 3 Pe h Road, Dundee DD1 4HN, Uni ed Kingdom,
el.: (+)44-1382-344845, ax: (+)44-1382-344691, E-mail:
[email p o ec ed] .
1
1. In oduc ion
This pape con ibu es o he unde s anding o he complex in e ace be ween
globalisa ion and labou s anda ds by ocussing on he e ec s o employmen
p o ec ion on he in e na ional loca ion o economic ac i i y.
The pas decade has wi nessed a ema kable accele a ion o he p ocess o in eg a ion
o he wo ld economy – o en e e ed o as ‘globalisa ion’. The libe alisa ion o
o eign di ec in es men (FDI) policies wo ldwide, which has esul ed in in es men
lows be ween coun ies g owing much as e han ade lows, has led o an inc ease
in he ease wi h which i ms (and jobs) mo e ac oss na ional bo de s. As a esul ,
go e nmen s’ he o ic and policies inc easingly be ay conce ns abou hei coun ies’
abili y o p e en domes ic indus y om eloca ing ab oad and o a ac and/o e ain
o eign in es men .
Labou ma ke ins i u ions a e commonly ega ded as playing a c ucial ole in
de e mining he loca ion o economic ac i i y, no leas i hey in luence he lexibili y
wi h which i ms can adjus ou pu scale and employmen le els o e ol ing economic
condi ions. Employmen p o ec ion laws in pa icula a e iden i ied as a majo sou ce
o in lexibili y and, inc easingly, ecommenda ions a e pu o wa d ha he s a e-
manda ed edundancy paymen s – which we e in oduced in many Eu opean coun ies
om he la e 1950s o he ea ly 1970s – a e disman led. The eme ging consensus is
ha by o cing i ms o unde -p oduce du ing economic booms and o e -p oduce
when he economy slows down1, high hi ing and i ing cos s unde mine hei abili y o
adap o as changing compe i i e ma ke s. No only is employmen p o ec ion held
esponsible o he poo employmen pe o mance o many Eu opean coun ies (e.g.
Lindbeck and Snowe , 1988; and Lazea , 1990) bu also o hinde ing coun ies’
abili y o hold on o oo loose indus ies. I ollows ha , in a wo ld whe e coun ies
pe cei e hemsel es as being engaged in ie ce compe i ion o economic ac i i y, he
subs an ial di e ences ha exis be ween economies (e en wi hin he Eu opean
1 Ben olila and Be ola (1990) a gue ha i ing cos s a e likely o ha e educed employmen a ia ion
in Eu ope.
2
Union) in hi ing and i ing es ic ions2 a e seen as a sou ce o un ai ‘compe i i e
ad an age’ o hose loca ions wi h lowe cos s o employmen adjus men s.
Whils a subs an ial amoun o wo k has been ca ied ou o assess he impac o
hi ing and i ing es ic ions on employmen 3, o ou knowledge (and despi e i s
p ominence in policy deba es) he lexibili y o e ed o i ms by a gi en loca ion in
adjus ing o changing economic condi ions4 has ecei ed ela i ely li le a en ion in
he heo e ical li e a u e on FDI5. Ins ead, he s udy o indus y loca ion has ended o
ocus on ma ke access and local cos s o p oduc ion as he cen al de e minan s o a
coun y’s abili y o a ac FDI and e ain domes ic i ms6. These ac o s a e o cou se
impo an . Howe e , by ocusing on he ela ionship be ween employmen p o ec ion
and he loca ion o indus y his pape ills an impo an gap in he li e a u e. To
explo e his ela ionship, we combine ideas om he indus ial o ganisa ion and he
labou li e a u e, and apply hese o a se -up in which i ms’ loca ions a e endogenous.
We a gue ha , con a y o he con en ional wisdom, labou ma ke in lexibili y may
no necessa ily hinde a coun y’s abili y o a ac and/o e ain economic ac i i y.
This iew inds heo e ical suppo in he heo y o indus ial o ganisa ion ha
emphasises how commi men (i.e. in lexibili y) may be a sou ce o s a egic
ad an age (Ti ole, 1988). We shall he e o e in es iga e how egion-speci ic
lexibili y a ec s loca ion decisions when i ms a e oligopolis ic and ac s a egically.
In a non-s a egic se -up, lexibili y only en ails ad an ages o a i m. This is no
2 Acco ding o Laya d, Nickell and Jackman (1991), i ing cos s ha e anged om 0.48 mon hs sala y
in Denma k o 5.24 in F ance and e en o 15.86 in I aly.
3 Hi ing and i ing es ic ions a e ypically no ound o ha e a decisi e ole on o e all a es o
unemploymen (e.g. Nickell (1998)).
4 Fi m lexibili y in e ms o “loca ional po olio di e si ica ion” has been examined in he li e a u e.
Mo e speci ically, he ole o cos and exchange a e unce ain y in p o iding a a ionale o se ing up
plan s in di e en coun ies has been espec i ely s udied in de Meza and Van de Ploeg (1987) and in
Sung and Lapan (2000).
5 Cooke (1997) inds e idence ha hos coun ies’ es ic i e legisla ion go e ning layo s ha e had a
nega i e e ec on US o eign di ec in es men ab oad. Mo an (1998, p.89) summa ises e idence om
in es o su eys and men ions labou egula ions, in pa icula “ lexibili y in hi ing and laying o
wo ke s”, as one o he main conce ns o i m loca ion in economies o ansi ion and de eloping
economies. A cu so y look a a ailable da a (UNCTAD, 2001), howe e , does no seem o suppo he
con en ional wisdom ha employmen p o ec ion is unambiguously inimical o inwa d FDI. Amongs
he Eu opean coun ies wi h highe han a e age alues o he employmen p o ec ion index
cons uc ed by Nickell, Nunzia a and Quin ini (2001), a signi ican numbe also show highe han
a e age le els o inwa d FDI (e.g. Belgium, he Ne he lands, Po ugal, Sweden and Finland) – in
gene al, no clea co ela ion eme ges by compa ing he da a on sha es o wo ld FDI and employmen
p o ec ion le els.
6 See Smi h (1987), Ho s mann and Ma kusen (1987) and (1992). Fo a comp ehensi e su ey on
mul ina ionals and FDI we e e o Ca es (1996).
3
necessa ily ue when i ms ac s a egically, since lexibili y hen implies lack o
commi men powe . Fi ms p oducing in loca ions whe e employmen is less lexible
may he e o e bene i om po en ial ad an ages ob ained by he commi men powe
ha such in lexibili y implies7.
We use a wo-pe iod oligopoly model, in which i ms’ loca ion decisions depend on
s a egic and lexibili y conside a ions. Ou analysis will be d i en by wo subs an i e
ques ions. Fi s , do loca ion-speci ic sou ces o in lexibili y c ea e s a egic
ad an ages ha a ec a coun y’s abili y o e ain p oduc ion o in e na ionally mobile
i ms? As we a gued, while his ques ion has emained la gely unexplo ed in he
heo e ical li e a u e8, policy make s o en ci e s ic employmen p o ec ion as a
h ea o he ancho age o domes ic indus y. Second, when can we expec o ind
s a egic clus e ing in he same egions and when is s a egic geog aphical dispe sion
mo e likely? By ocussing on his ques ion, his pape complemen s he economic
geog aphy li e a u e, which is mainly conce ned wi h agglome a ion o ma ion in
non-s a egic se -ups9.
The model is ou lined in Sec ion 2. The non-s a egic de e minan s o loca ion a e
analysed in Sec ion 3, while he s a egic implica ions o in lexibili y a e discussed in
Sec ion 4. Sec ion 5 analyses he loca ion decisions o i ms and examines he e ec s
o inc easing globalisa ion on he s a egic incen i es o geog aphical agglome a ion.
Sec ion 6 concludes he pape .
2. The model
Two i ms plan o launch new p oduc s, which a e impe ec subs i u es, o be sold in
an in eg a ed ma ke . One i m, he Home i m, has i s headqua e s in he coun y
named “Home”, while he o he , e e ed o as he Fo eign i m, has i s headqua e s
in he coun y named “Fo eign”. Each has o decide whe e o loca e hei p oduc ion
plan : ei he in “Home” o in “Fo eign”. We assume ha he ixed cos s o se ing up
a plan a e su icien ly high o ensu e ha each i m chooses o ha e one plan only.
7 The e ec o adjus men cos s on s a egic beha iou has been discussed in se -ups wi hou loca ion
decisions (see Lapham and Wa e, 1994; Jun and Vi es, 2001).
8 The s a egic e ec s o p oduc ion cos di e ences on loca ion choice ha e ecei ed ample a en ion
in he li e a u e (see e e ences in oo no e 6).
9 See, o ins ance, K ugman (1991) and Fuji a, K ugman and Venables (1999).
4
Compe i ion akes place du ing wo pe iods, wi h i ms choosing “ma ke ac ions” –
ou pu s unde Cou no and p ices unde Be and compe i ion – in each pe iod. The
espec i e demand unc ions o he Home and he Fo eign i m o pe iod one a e
gi en by
*
111 eqqap −−= (1a)
1
*
1
*
1eqqap −−= (1b)
In pe iod wo, he i ms’ espec i e demand unc ions a e:
ueqqap +−−= *
222 (2a)
ueqqap +−−= 2
*
2
*
2(2b)
whe e 10
<
≤
e is an in e se measu e o p oduc di e en ia ion10, and a>0. The
Home i m’s p ice and ou pu a e deno ed by
p
and
q
, espec i ely. Va iables
e e ing o he Fo eign i m a e s a ed. Subsc ip s 1 and 2 indica e he ime pe iod.
In pe iod one, demand o ha pe iod is obse ed bu he e is unce ain y abou u u e
demand. Hence, a s ochas ic componen ,
u
, en e s he demand unc ion o pe iod
wo and is de ined o e he suppo
[,
]
u
u
, wi h mean Eu=0 and a iance 2
σ
. The
unce ain y is esol ed a he s a o pe iod wo11.
As men ioned abo e, each i m has o choose ei he Home o Fo eign as i s
p oduc ion loca ion. We assume ha hese coun ies di e in one impo an espec .
In Home, s ic labou ma ke egula ions, inspi ed by a conce n o employmen
p o ec ion, p e ail. These cause i ms o incu hi ing and i ing cos s i , a e an
unexpec ed change in demand, hey wan o de ia e om he pe iod-one p oduc ion
(and hence employmen ) le el. By con as , labou ma ke egula ions in Fo eign a e
lax, implying ha expansions o educ ions in p oduc ion can be ca ied ou wi hou
incu ing any adjus men cos s. The p o i unc ions o he Home (
π
) and he
Fo eign i m ( *
π
) a e espec i ely gi en by12
CRR −+= 21
π
(3a)
10 S ic ly speaking, he model could allow homogeneous p oduc s (e=1) wi h Cou no beha iou , bu
no wi h Be and beha iou .
11 We es ic he suppo ],[uu o gua an ee in e io solu ions.
12 We assume ha he discoun ac o is uni y.

5
**
2
*
1
*CRR −+=
π
(3b)
whe e
Rdeno es he Home i m’s e enue in pe iod (wi h =1,2) and Cs ands o i s
o al cos ; *
Rand *
C e e o he Fo eign i m. To al cos s consis o e ms ha a e
loca ion speci ic and o he s ha a e no . They depend on he loca ion chosen by he
i m and on whe he i engages in FDI o no . The ma ix in (4) gi es he cos
unc ion o each i m in each loca ion:
Home loca ion Fo eign loca ion
Home i m: C
ϕ
+Λ++ 21 cqcq
δϕ
+++ 21 cqcq
Fo eign i m: C*
δϕ
++Λ++ **
2
*
1cqcq
ϕ
++ *
2
*
1cqcq (4)
P oduc ion cos s a e no loca ion speci ic since he ma ginal cos o p oduc ion (c) is
assumed o be he same in bo h loca ions. This allows us o abs ac om loca ion-
speci ic cos di e ences. Adjus men cos s (
Λ
and *
Λ
) a e, howe e , loca ion
speci ic; hey a e paid in pe iod wo, bu only i he i m loca es in he in lexible
(Home) loca ion. They a e deno ed by 2
12 ))(2/(qq −≡Λ
λ
o he Home i m and by
2*
1
*
2
*))(2/(qq −≡Λ
λ
o he Fo eign i m13. The
λ
-pa ame e ( 0
>
λ
) measu es he
deg ee o in lexibili y. Fi ms ha loca e in Fo eign, he lexible loca ion, incu no
adjus men cos s. The i m’s ixed cos o se ing up a plan in i s na i e coun y is
deno ed by
ϕ
. Howe e , i i loca es ab oad, hen i s ixed cos s a e
δ
ϕ
+
, wi h
0
>
δ
. In o he wo ds,
δ
ep esen s he cos s associa ed wi h FDI. These may, o
ins ance, be due o he cos s incu ed in dealing wi h o eign languages, o eign laws
and a o eign axa ion sys em14. Mo e gene ally,
δ
can be hough o as e lec ing he
ba ie s o he mobili y o capi al, and will he e o e sh ink as he deg ee o ma ke
in eg a ion inc eases.
13 This adjus men cos speci ica ion is s anda d in he li e a u e. O he (e.g., linea , asymme ic)
speci ica ions would make he analysis echnically mo e edious wi hou changing he quali a i e
esul s (see Hame mesh (1996, Ch.6) o a ex book su ey o di e en employmen adjus men cos
speci ica ions).
14 This was i s o malised by Hi sch (1976).
6
Fi ms play a wo-pe iod h ee-s age game, ac ing simul aneously in each s age. The
sequence o decisions is shown in igu e 1. In pe iod one, p oduc ion loca ions,
Home (H) o Fo eign (F), a e chosen (s age one). The e a e ou possible loca ion
equilib ia: wo in which bo h i ms choose he same loca ion, (H,H) and (F,F), and
wo in which i ms choose a di e en loca ion, (H,F) and (F,H). Fo each loca ion
pai , he i s le e e e s o he Home i m’s loca ion choice, whe eas he second
indica es he Fo eign i m’s loca ion choice. No FDI occu s in he (H,F)-equilib ium,
while bo h i ms engage in FDI in he (F,H)-equilib ium. In s age wo, pe iod-one
ac ions a e de e mined gi en demand o pe iod one bu wi h unce ain y abou
demand in pe iod wo. In pe iod wo, he unce ain y is esol ed and i ms choose
ma ke ac ions a e ha ing obse ed ac ual demand o ha pe iod (s age h ee).
[Figu e 1 abou he e]
In his model, a i m’s loca ion decision is in luenced bo h by non-s a egic and
s a egic ac o s. The non-s a egic aspec s o he p oduc ion loca ion choice a e
examined i s .
3. The non-s a egic dimension o he loca ion decision
To ocus on he non-s a egic loca ion de e minan s, we ini ially abs ac om
s a egic beha iou by conside ing he limi case o e=0 in which one i m’s p oduc
is su icien ly di e en om i s i al’s ha each i m e ec i ely becomes a
monopolis . In he absence o s a egic beha iou , only cos and lexibili y
conside a ions will de e mine i ms’ loca ion choices.
The loca ion choice o a Fo eign monopolis is simple: i will always choose o
p oduce in Fo eign since his choice en ails maximum lexibili y wi hou incu ing he
cos o FDI. Fo a Home monopolis , he loca ion decision in ol es a ade-o
be ween he cos s o FDI and he lexibili y bene i s associa ed wi h p oducing in
Fo eign. Due o demand unce ain y, he i m an icipa es i will ace adjus men cos s
in Home, while he e will be no adjus men cos s when i p oduces in Fo eign.
No e ha he e is a c i ical le el o unce ain y abo e which he Home monopolis
will choose o p oduce in Fo eign and below which i will choose o loca e in Home.
7
This c i ical 2
σ
-le el dec eases in he deg ee o labou ma ke in lexibili y in Home
(
λ
) and inc eases in he deg ee o in e na ional capi al immobili y (
δ
).
4. S a egic implica ions o employmen p o ec ion
When p oduc s a e subs i u es, i.e. e>0, i ms beha e as duopolis s and hei loca ion
decisions in ol e bo h s a egic and non-s a egic conside a ions. Sol ing he game
backwa ds, in his sec ion we explo e he s a egic implica ions o employmen
p o ec ion a ixed loca ions (loca ion choice is discussed in sec ion 5). Since i ms
in e ac di e en ly unde Cou no and unde Be and compe i ion, we s udy he
s a egic implica ions o he p oduc ion loca ion in bo h o ms o oligopoly. Fo
exposi ional cla i y, we explain he na u e o he s a egic e ec s in de ail using he
case in which each i m p oduces domes ically, ha is, he Home i m p oduces in
Home and he Fo eign i m in Fo eign (i.e., (H,F)). The s a egic beha iou in he
o he possible loca ion equilib ia will be discussed a he end o each subsec ion,
using able 1, which epo s he s a egic e m in all possible loca ion combina ions.
4.1. Cou no compe i ion
Fi ms’ p oduc ion loca ions a ec hei ma ke ac ions. We i s conside pe iod wo,
in which loca ions and pe iod-one ou pu s ha e al eady been chosen. Pe iod- wo
eac ion unc ions unde Cou no compe i ion a e ob ained by maximising pe iod- wo
p o i s, 2
π
and *
2
π
, wi h espec o ou pu s. When each i m p oduces domes ically,
(i.e., in he (H,F)-case), we ha e Λ−−= 222 cqR
π
and *
2
*
2
*
2cqR−=
π
. Fi ms’
espec i e second-pe iod eac ion unc ions a e gi en by
(
)
λλ
++−+= 2/)( 1
*
22 qequAqwi h caA
−
≡
(5a)
2/)( 2
*
2equAq−+= (5b)
Exp essions (5a) and (5b) clea ly sugges ha a i m’s loca ion has implica ions o i s
lexibili y. The Home i m’s eac ion unc ion esponds less o unexpec ed demand
shocks han i s i al’s does ( uquq ∂∂<∂∂ // *
22 om (5a) and (5b)). The i m in
Home is also less esponsi e o changes in i al ou pu (
2
/
)
2
/(
e
e
<
+
λ
). In o he
wo ds, he labou ma ke in lexibili y in Home a ec s he slope o he Home i m’s
8
second-pe iod eac ion unc ion. Finally, due o adjus men cos s, he Home i m’s
eac ion unc ion depends posi i ely on i s own pas ou pu , as cap u ed by he e m in
1
q.
We now u n o s age wo. Being unce ain abou he demand in pe iod wo, i ms
simul aneously de e mine hei ou pu s o pe iod one by maximising o al expec ed
p o i s wi h espec o i s -pe iod ou pu s. Wi h ),,,(*
22
*
11 qqqq
ππ
= and
),,,(*
22
*
11
** qqqq
ππ
=, i s -o de condi ions a e gi en by15
0)]/([ 1
*
2
*
2
1=+ dqdqEE q
q
ππ
(6a)
0)]/([ *
12
**
2
*
1=+ dqdqEE q
q
ππ
(6b)
Subsc ip ed a iables deno e pa ial de i a i es. In (6a) and (6b), he second e m on
he le -hand-side cap u es he s a egic e ec s. The s a egic e m in (6a) is posi i e
(0
2
*
2<−= q
q
π
and 0/1
*
2<dqdq ), implying ha he i m in Home s a egically o e -
p oduces in pe iod one (o , 0
1<
q
E
π
). By choosing a high ou pu le el in pe iod one,
he Home i m is o ced o keep i s p oduc ion in he nex pe iod a a ela i ely high
le el, since changing i s p oduc ion hen is cos ly16. This commi men o keep
p oduc ion high in pe iod wo o ces he i al i m o cu back i s ou pu . Meanwhile,
he e is no s a egic beha iou by he i m in Fo eign (i.e., he s a egic e m is ze o,
since, om exp essions (5a) and (5b), 0/*
12 =dqdq ).
E iden ly, he (H,F)-case is no he only possible loca ion combina ion ha can a ise.
The uppe hal o able 1 p esen s he s a egic e ms o bo h i ms unde Cou no
beha iou o all possible loca ion combina ions. I bo h i ms engage in FDI, (F,H),
only he Fo eign i m s a egically o e -p oduces in pe iod one. I bo h i ms p oduce
in Fo eign, (F,F), nei he i m ac s s a egically. Finally, i bo h i ms p oduce in
Home, (H,H), hen each i m beha es s a egically and o e -p oduces in pe iod one.
4.2. Be and compe i ion
15 F om he en elope heo em we know 0
*
22 *
2
2== q
q
ππ
, and hus 0
*
*
2
2== q
qEE ππ .
16 Diag amma ically, he Home i m’s expec ed second-pe iod eac ion unc ion has shi ed ou .
15
a ac i eness o he Fo eign loca ion inc eases). Gi en ha he Fo eign i al p oduces
in Fo eign, we hen need o examine whe e he Home i m will se up i s plan .
Figu e 2 shows he loca ion pa e n ha eme ges in ),(2
λσ
-space. The equilib ium
loca ion combina ions unde Cou no and unde Be and a e indica ed by supe sc ip s
C and B espec i ely22.
[Figu e 2 abou he e]
Since compe i ion is oughe unde Be and han unde Cou no beha iou , p ice
compe i ion makes i ha de o i ms o ca y he cos s o FDI. Based on hese
conside a ions only, one would he e o e expec he a ea wi h domes ic ancho age o
be la ge unde Be and han unde Cou no compe i ion. In ac , i is clea om
igu e 2 ha he opposi e is ue: he egion in which domes ic ancho age occu s is
la ge unde Cou no (a eas I and II) han unde Be and compe i ion (a ea I only).
This seemingly coun e in ui i e esul can be explained by he s a egic
conside a ions unde lying i ms’ loca ion decisions23. Unde Be and compe i ion,
gi en i al p oduc ion in Fo eign, he Home i m is, om a s a egic poin o iew,
be e o p oducing in Fo eign ( B
FF ),(in a ea II) han in Home ( HFFF
θθ
>, see
exp ession (13a)). By p oducing in Fo eign, i a oids he massi e i s -pe iod p ice
unde cu ing by i s i al ha would occu i he Home i m we e o p oduce in Home.
In con as , unde Cou no compe i ion, he Home loca ion holds s a egic ad an ages
(FFHF
θθ
>, see (12a)), implying ha he Home i m will p oduce domes ically
(gi ing (H,F)C) in a ea II.
As p oduc di e en ia ion inc eases (i.e. as e alls), s a egic beha iou is diminished.
As a esul , he a ea wi h domes ic ancho age sh inks unde Cou no compe i ion
22 In he igu es, he maximum alue o
λ
is limi ed o ensu e he exis ence o all possible equilib ia.
A e=0.75,
λ
- alues smalle han (app oxima ely) 2.5 a e needed o he exis ence o he (H,F) and
(F,H) equilib ia unde p ice compe i ion (see also oo no e 17). Unde quan i y compe i ion, alues o
λ
do no need o be es ic ed. Howe e , no hing changes quali a i ely a
λ
- alues abo e hose
depic ed.
23 In ac , i i ms do no beha e s a egically in s age wo, and hence
θ
akes he same (open-loop)
alue a each loca ion combina ion (see oo no e 21), he a ea wi h domes ic ancho age is la ge unde
Be and hen unde Cou no compe i ion.

16
whe eas i expands unde Be and compe i ion. Impo an ly, he a ea wi h domes ic
ancho age is always la ges unde Cou no beha iou .
5.2. Globalisa ion and s a egic agglome a ion
The p ocess o globalisa ion implies ha capi al mobili y inc eases. In ou model his
is cap u ed by alling FDI-cos s (
δ
). As ba ie s o capi al mobili y a e p og essi ely
lowe ed, o he loca ion equilib ia, beside (H,F) and (F,F), s a o eme ge. Figu es 3
and 4 show he loca ion pa e n unde quan i y and p ice compe i ion espec i ely.
E en a b ie glance a hese igu es con i ms ou p e ious inding ha domes ic
ancho age emains s onges in indus ies wi h Cou no beha iou : he egion in
which he Home i m p oduces in Home is much la ge unde Cou no (a eas Ia and
Ib in igu e 3) han unde Be and (a ea I in igu e 4).
Wi h inc easing globalisa ion, he cos o FDI (
δ
) alls and so i is easie o i ms o
in es ab oad. Thus, he cos o he Home i m o acqui ing lexibili y by in es ing in
Fo eign is now lowe . In igu es 3 and 4, he h eshold unce ain y locus below which
he Home i m p oduces domes ically is now lowe han in igu e 2. The a ea in
which he Home i m is loca ed in Home has sh unk (a eas Ia and Ib in igu e 3; a ea I
in igu e 4), meaning ha domes ic ancho age is less easily sus ained, while FDI om
he Home i m in o he Fo eign loca ion becomes ela i ely mo e impo an . In bo h
igu es his is e lec ed in an enla gemen o a ea II, in which (F,F) is he unique
equilib ium. Inc eased globalisa ion educes he c i ical le el o unce ain y abo e
which i ms p oduce in Fo eign o ob ain high p oduc ion lexibili y.
When low FDI cos s a e combined wi h low unce ain y, s a egic conside a ions in
he loca ion decision become ela i ely mo e impo an . Unde Cou no compe i ion
his s a egic conce n encou ages inwa d FDI a he in lexible loca ion (in a ea Ib in
igu e 3), leading o he (H,H)-equilib ium. The Fo eign i m ac ually has o jump a
ba ie and se up i s plan in he Home loca ion o a oid ending up in he wo s
possible s a egic posi ion (HFHH **
θθ
>, see exp ession (12b)). The (H,H)-
equilib ium is an example o s a egic agglome a ion, in he sense ha he
agglome a ion would no occu in he absence o s a egic beha iou .
17
Unde Be and, globalisa ion combined wi h low unce ain y can lead o FDI by he
Home i m as i ies o loca e in he s a egically a ou able lexible Fo eign loca ion
( esul ing in (F,F)). As shown in igu e 4, p o ided ha he le el o employmen
p o ec ion in Home (
λ
) is su icien ly high, he (F,F)-equilib ium occu s e en a
ce ain y ( 0
2=
σ
); his oo is an example o s a egic agglome a ion. In his case, he
Home i m is willing o incu he FDI-cos pu ely o s a egic easons ( HFFF
θθ
>,
see exp ession (13a)).
[Figu es 3 and 4 abou he e]
As globalisa ion deepens and
δ
alls u he , egions in which domes ic ancho age
p e ails become smalle s ill. In he limi , wi h comple e globalisa ion ( 0
=
δ
), i ms
e ec i ely lose hei na ionali y. The espec i e loca ion pa e ns ha hen occu
unde quan i y and unde p ice compe i ion a e shown in igu es 5 and 6. E en hen,
i ms may wan o loca e in he in lexible Home loca ion. In ac , a su icien ly low
le els o unce ain y, (H,H) is he unique equilib ium unde Cou no compe i ion.
Mo eo e , (H,H) can also be an equilib ium ( oge he wi h (F,F)) unde Be and
compe i ion, bu only a e y high deg ees o employmen p o ec ion. This (H,H)-
equilib ium is also an example o s a egic agglome a ion because, a high
λ
,
HFHH **
θθ
>; gi en ha he Home i m p oduces in Home, he Fo eign i m se s up
p oduc ion in Home oo, hus a oiding an un a ou able s a egic posi ion.
[Figu es 5 and 6 abou he e]
The loca ion e ec s o deepening globalisa ion p o e o be ela i ely obus o
changing deg ees o p oduc di e en ia ion. I is wo h men ioning ha as p oduc
di e en ia ion inc eases (e alls), s a egic agglome a ion s ill occu s, bu only a
lowe le els o unce ain y.
6. Conclusions
18
We ha e explo ed how di e ences in labou ma ke lexibili y a ec loca ion
decisions when u u e demand is unce ain and i ms ac s a egically. When demand
unce ain y is e y high, i ms will clus e in coun ies whe e he labou ma ke is
ela i ely lexible, hus a oiding cos ly edundancy packages du ing economic
slowdowns and expensi e o e ime paymen s o hi ing cos s in economic booms.
Among coun ies wi h a high deg ee o capi al mobili y, his loca ion pa e n may
e en p e ail a ela i ely low le els o unce ain y.
Howe e , when i ms ac s a egically, hey may be willing o o ego lexibili y and
p oduce in coun ies whe e he labou ma ke is ela i ely in lexible in o de o ob ain
s a egic ad an ages. This is he case when he i ms engage in Cou no beha iou .
Unde quan i y compe i ion an in lexible loca ion allows a i m o commi o high
u u e ou pu , which makes he in lexible loca ion mo e a ac i e a low le els o
unce ain y. This s a egic ad an age helps o main ain domes ic ancho age o i ms
in loca ions wi h s ic labou egula ions. Unde p ice compe i ion howe e , a i m
loca ed in he in lexible coun y aces agg essi e p icing om i s lexible i al in
pe iod one. As a esul , he in lexible loca ion is un a ou able bo h om a s a egic
and a lexibili y pe spec i e. Hence, bo h s a egic and lexibili y incen i es wo k
agains domes ic ancho age unde Be and compe i ion.
We ha e shown ha deepening globalisa ion can lead o a g ea e endency o he
de elopmen o s a egic agglome a ion. This is he case unde bo h Cou no and
Be and compe i ion. Unde Cou no compe i ion, i ms acing low FDI-cos s clus e
in he in lexible loca ion when unce ain y is low ( his can also occu unde Be and
compe i ion a e y high le els o employmen p o ec ion and when unce ain y is
e y low). Such clus e ing has howe e a p isone ’s dilemma cha ac e wi h i ms all
p oducing highe ou pu and enjoying less lexibili y han hey would in a loca ion
wi h lowe labou adjus men cos s. Unde Be and compe i ion, when s a egic
agglome a ion occu s, i does so mainly in he lexible loca ion as i ms lee he
s a egically un a ou able in lexible loca ion.
When o mula ing he policy lessons om his analysis, one should p oceed wi h
cau ion. Th oughou his pape , we ha e assumed ha he le el o employmen
p o ec ion is exogenous. This is a easonable assump ion since he poli ical
unwillingness o change employmen p o ec ion egula ions, once hese a e in place,
19
is o en s ong. I does no , howe e , p eclude policy make s om using loca ion-
dependen iscal incen i es o inc ease he a ac i eness o hei egion. Ou analysis
sugges s ha coun ies wi h s ic labou egula ions will ind i less di icul o
achie e domes ic ancho age o key indus ies, by using iscal incen i es, when i m
beha iou is app oxima ed by Cou no a he han by Be and compe i ion24.
Finally, we ha e no de i ed op imal employmen p o ec ion le els in his pape . This
would equi e aking in o accoun he link ha ypically exis s be ween employmen
p o ec ion and labou cos s: employmen p o ec ion ends o push up labou cos – no
leas because i s eng hens wo ke s’ ba gaining powe – which would in u n a ec a
loca ion’s a ac i eness o in es o s. Whe he and how a go e nmen should design
i s labou s anda ds op imally in o de o achie e domes ic ancho age and o
maximise i s FDI-in low is a ques ion le o u u e esea ch.
24 Whe he i ms’ beha iou is be e desc ibed by Cou no o by Be and compe i ion would need o
be empi ically in es iga ed on an indus y-by-indus y basis. The e exis s a subs an ial empi ical
li e a u e on his issue ( o a ex book su ey, see Ma in (2002), Ch.7).
20
Re e ences
Ben olila, S. and G. Be ola (1990), “Fi ing cos s and labo demand: How bad is
Eu oscle osis”, Re iew o Economic S udies, 57, 381-402.
Ca es, R.E. (1996), Mul ina ional En e p ise and Economic Analysis, 2
nd ed.,
Camb idge Uni e si y P ess, Camb idge.
Cooke, W. N. (1997), “The in luence o indus ial ela ions ac o s on U.S. o eign
di ec in es men ab oad”, Indus ial and Labo Rela ions Re iew, 51, 3-17.
de Meza, D. and F. Van de Ploeg (1987), “P oduc ion lexibili y as a mo i e o
mul ina ionali y”, The Jou nal o Indus ial Economics, 35, 343-351.
Fuji a, M., P. K ugman and A. Venables (1999), The spa ial economy: Ci ies, egions
and in e na ional ade, Camb idge and London: MIT P ess.
Hame mesh, D. (1996), Labo Demand, P ince on: P ince on Uni e si y P ess.
Hi sch, S. (1976), “An in e na ional ade and in es men heo y o he i m”, Ox o d
Economic Pape s, 28, 258-269.
Ho s mann, I. And J. Ma kusen (1987), “S a egic in es men s and he de elopmen
o mul ina ionals”, In e na ional Economic Re iew, 28, 109-121.
Ho s mann, I. And J. Ma kusen (1992), “Endogenous ma ke s uc u es in
in e na ional ade (na u a aci sal um)”, Jou nal o In e na ional Economics,
32, 109-129.
Jun, B. and X. Vi es (2001), Incen i es in dynamic duopoly, Cen e o Economic
Policy Resea ch, Discussion Pape 2899.
K ugman, P. (1991), Geog aphy and ade, Camb idge and London: MIT P ess.
Lapham, B. and R. Wa e (1994), “Ma ko puppy dogs and ela ed animals”,
In e na ional Jou nal o Indus ial O ganiza ion, 12, 569-593.
Laya d, R., S. Nickell and R. Jackman (1991), Unemploymen , Ox o d: Ox o d
Uni e si y P ess.
Lazea , E. (1990), “Job secu i y p o isions and employmen ”, The Qua e ly Jou nal
o Economics, 699-726.
Lindbeck, A. and D. Snowe (1988), “Job secu i y, wo k incen i es and
unemploymen ”, Scandina ian Jou nal o Economics, 90, 453-474.
Ma in, S. (2002), Ad anced Indus ial Economics, second ed., Ox o d: Blackwell
Publishe s.
Mo an, T. (1998), Fo eign di ec in es men and de elopmen , Ins i u e o
In e na ional Economics, Washing on, DC.
Nickell, S. (1998), “Unemploymen : Ques ions and some answe s”, Economic
Jou nal, 108, 802-816.
Nickell, S., L. Nunzia a and G. Quin ini (2001), “The Be e idge cu e,
unemploymen and wages in he OECD om he 1960s o he 1990s”, mimeo,
London School o Economics.
Smi h, A. (1987), “S a egic in es men , mul ina ional co po a ions and ade policy”,
Eu opean Economic Re iew, 31, 89-96.
Sung, H. and H. Lapan (2000), “S a egic o eign di ec in es men and exchange- a e
unce ain y”, In e na ional Economic Re iew, 41, 411-423.
Ti ole, J. (1988), The Theo y o Indus ial O ganisa ion, Camb idge, Massachuse s:
The MIT P ess.
UNCTAD (2001), Wo ld In es men Repo , Uni ed Na ions, New Yo k and Gene a.

S age 1: Fi ms choose loca ion
Possible combina ions:
(F,F) ; (H,F) ; (F,H) ; (H,H)
S age 2:Fi ms choose 1s -pe iod ma ke ac ions:
• (q1, q*1) i Cou no compe i ion
• (p1, p*1) i Be and compe i ion
S age 3: Fi ms choose 2nd-pe iod ma ke ac ions:
• (q2, q*2) i Cou no compe i ion
• (p2, p*2) i Be and compe i ion
PERIOD 1
Unce ain y
PERIOD 2
Ce ain y
Figu e 1: The sequence o decisions
Table 1: The s a egic e ms in all possible loca ion combina ions
S a egic
e m (H,H) (H,F) (F,H) (F,F)
COURNOT
Home i m
1
*
2
*
2dq
dq
Eq
π
HH
HH Eq
e2
2
∆
λ HF
HF Eq
e2
2
∆
λ
0
0
Fo eign i m *
1
2
*
2dq
dq
Eq
π
HH
HH Eq
e*
2
2
∆
λ
0 FH
FH Eq
e*
2
2
∆
λ
0
BERTRAND
Home i m 1
*
2
*
2dp
dp
Ep
π
HH
HH Eq
e2
2
∇
−βλ HF
HF Eq
e2
2
∇
βλ FH
FH Eq
e2
2
2∇
−βλ
0
Fo eign i m *
1
2
*
2dp
dp
Ep
π
HH
HH Eq
e*
2
2
∇
−βλ HF
HF Eq
e*
2
2
2∇
−βλ
FH
FH Eq
e*
2
2
∇
βλ
0
No e: 0)2(22 >−+≡∆e
HH
λ
; 0)2(22>−+≡∆=∆e
FHHF
λ
0
)
1
(
)
2
(
222
>
+
−
+
≡
∇
βλ
βλ
e
HH ;
0
)
1
(
)
2
(
2
2
>
+
−
+
≡
∇
=
∇
βλ
βλ
e
FHHF
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5
λ
2
σ
I: (H,F)C,B
III: (F,F)C,B
II: (H,F)C;
(F,F),B
Figu e 2: The loca ion pa e n unde Cou no and Be and compe i ion:
High FDI-cos s (δδ =0.01; e=0.75; A=1)
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5
2
σ
λ
Figu e 3: The loca ion pa e n unde Cou no compe i ion wi h inc eased globalisa ion:
In e media e FDI-cos s (δδ =0.005; e=0.75; A=1)
Ia:(H,F) Ib:(H,H)
II: (F,F)
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5
2
σ
λ
Figu e 4: The loca ion pa e n unde Be and compe i ion wi h inc eased globalisa ion:
In e media e FDI-cos s (δδ=0.005; e=0.75; A=1)
I: (H,F)
II:(F,F)
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5
2
σ
λ
Figu e 5: The loca ion pa e n unde Cou no compe i ion wi h comple e globalisa ion:
No FDI-cos s (δδ =0; e=0.75; A=1)
I: (H,H)
II: (H,H);(F,F)
III: (F,F)
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5
2
σ
λ
Figu e 6: The loca ion pa e n unde Be and compe i ion wi h comple e globalisa ion:
No FDI-cos s (δδ =0; e=0.75; A=1)
I: (H,H);(F,F)
II: (F,F)