Should I Stay or should I go? A note on employment protection, domestic anchorage, and FDI

Should I s ay o should I go? A no e on employmen p o ec ion, domes ic ancho age, and FDI Ge da Dewi Na ional Uni e si y o I eland Maynoo h Holge Gö g Uni e si y o No ingham and IZA Bonn Ca ia Mon agna* Uni e si y o Dundee and GEP Uni e si y o No ingham July 2003 Abs ac : This pape examines how employmen p o ec ion legisla ion a ec s loca ion decisions o mul ina ionals. Based on a simple heo e ical amewo k, we es ima e an empi ical model, using OECD-da a on bila e al FDI- lows and employmen p o ec ion indices. We ind ha , while an “un a ou able” employmen p o ec ion di e en ial be ween a domes ic and a o eign loca ion is inimical o o eign di ec in es men (FDI), a high domes ic le el o employmen p o ec ion ends o discou age ou wa d FDI. The esul s a e in line wi h ou conjec u e ha s ic employmen p o ec ion in he i m’s home coun y makes i ms eluc an o eloca e ab oad and keeps hem “ancho ed” a home. JEL Codes: D80, F23, J80. Key wo ds: Unce ain y, employmen p o ec ion, o eign di ec in es men , domes ic ancho age. Acknowledgemen s: We a e g a e ul o Daniel Mi za o making he OECD da a a ailable o us and o De mo Leahy and Hassan Molana o help ul commen s. Holge Gö g g a e ully acknowledges inancial suppo om he Le e hulme T us unde P og amme G an F114/BF and he Eu opean Commission unde G an No. SERD-2002-00077. * Co esponding au ho : Depa men o Economic S udies, Uni e si y o Dundee, 3 Pe h Road, Dundee, DD1 4HN, UK, el: (00)44-1382-344845, email: [email p o ec ed] 1 1. INTRODUCTION The aim o his pape is o shed some ligh on he e ec s o labou ma ke lexibili y (o he lack o i ) on he loca ion o economic ac i i y. The inc easing deg ee o economic in eg a ion and he libe alisa ion o o eign di ec in es men (FDI) policies wo ldwide ha e b ough he de e minan s o he loca ion o economic ac i i y o he o e on o policy deba es, wi h go e nmen s’ conce ns ocussing inc easingly on hei 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 laws and ins i u ions a e commonly ega ded as c ucial in de e mining he ela i e a ac i eness o loca ions o in e na ionally mobile i ms. Amongs hese, employmen p o ec ion laws a e pe cei ed as being pa icula ly ha m ul because hey educe he lexibili y wi h which i ms can adjus ou pu and employmen o e ol ing economic condi ions. Inc easingly, he subs an ial di e ences ha exis be ween economies (e en wi hin he Eu opean Union) in hi ing and i ing es ic ions 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. Al hough a subs an ial amoun o wo k exis s on he impac o employmen legisla ion on employmen ,1 li le esea ch has been done on he ela ionship be ween he o me and he loca ion o indus y.2 To ou knowledge, he limi ed body o a icles ha add esses his ela ionship ocuses on he ole o employmen p o ec ion in unde mining a loca ion’s abili y o a ac new oo loose indus y. Howe e , he deg ee o labou ma ke lexibili y also 1 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 1996, 2001), bu a e shown o educe job ealloca ion a es and employmen a ia ion o e he business cycle – e.g. Ben olila and Be ola (1990), Ga ibaldi e al. (1997), OECD (1996). 2 Haaland and Woo on (2002) and Haaland e al. (2003) p o ide ela ed heo e ical analyses, showing ha coun ies wi h mo e lexible labou ma ke s a ac inwa d in es men . By con as , Dewi e al. (2003) a gue ha labou ma ke in lexibili y may no necessa ily hinde a coun y’s abili y o e ain (and unde ce ain condi ions e en a ac ) economic ac i i y; since in lexibili y implies commi men powe , i ms may p e e an in lexible loca ion o e a lexible one, e en in an unce ain en i onmen . In an empi ical s udy, Gö g (2002) inds ha hos coun ies’ i ing cos s a e nega i ely ela ed o inwa d FDI om he US. Nicole i e al. (2003) ind ha employmen p o ec ion educes FDI in OECD coun ies. B own e al. (2003) is a gene al discu si e pape on he ela ionship be ween FDI and wo king condi ions in de eloping coun ies. 2 a ec s coun ies’ abili y o e ain hei exis ing indus ial base, wha we e e o as ‘domes ic ancho age’3. In dis ega idng his aspec , which is cen al o policy deba es, he exis ing li e a u e ails o cap u e he complexi y o he e ec s o employmen p o ec ion on indus y loca ion. In o de o ill his gap, we use a simple heo e ical model o examine how di e ences in employmen p o ec ion egula ions a ec (1) he loca ion decision o a new i m, when al e na i e loca ions a e possible, and (2) he eloca ion incen i es o an exis ing p oduc ion acili y. We hen es empi ically he p edic ions o he model using da a on bila e al FDI lows and employmen p o ec ion indices ob ained om he OECD. Ou empi ical analysis sugges s ha employmen p o ec ion laws a e likely o ha e di e en e ec s on i ms’ loca ion decisions: whils an ‘un a ou able’ employmen p o ec ion di e en ial be ween a domes ic and a o eign loca ion is inimical o inwa d FDI, a high domes ic le el o employmen p o ec ion ends o discou age ou low FDI. We also ind ha he e ec o employmen p o ec ion depends on he le el o in es men cos in he hos coun y. In pa icula , FDI cos s end o ein o ce he ancho age e ec o employmen p o ec ion. The es o he pape is o ganised as ollows. Sec ion 2 de elops a simple heo e ical amewo k. Sec ion 3 es ima es an econome ic model o bila e al FDI using OECD da a. Sec ion 4 concludes he pape . 2. THEORETICAL FRAMEWORK 2.1. Employmen p o ec ion and loca ion We i s ly analyse he decision o a monopolis conside ing whe he o se up a new p oduc ion acili y in i s home coun y h o o loca e in o eign coun y . Coun ies h and 3 A heo e ical analysis o he ‘a ac ion e sus e en ion’ e ec s o employmen p o ec ion in s a egic se ings is de eloped by Dewi , Leahy and Mon agna (2003). 3 ep esen an in eg a ed ma ke and p oduc ion cos s a e he same in each loca ion.4 We assume, howe e , he e o be an FDI cos which he i m will incu in he case i decides o loca e ab oad. The e a e wo pe iods. In pe iod one, he i m knows demand o ha pe iod bu aces unce ain y abou pe iod- wo demand, which is esol ed a he s a o pe iod wo. The in e se demands o pe iod one and wo a e gi en by 11 qap −= and uqap +−= 22 , whe e p and q deno e p ice and quan i y (wi h =1,2), a is a posi i e cons an , and [ ] uuu ,∈ is a s ochas ic demand componen , wi h mean 0 = Eu and a iance . Since u u e demand is unce ain, he i m alues lexibili y. Howe e , lexibili y may be hinde ed by employmen p o ec ion egula ions in he local labou ma ke . Employmen p o ec ion gi es ise o hi ing and i ing cos s when u u e ou pu needs o be adjus ed in line wi h he demand hen p e ailing. Suppose o al a iable cos s in loca ion i (i = h, ) a e gi en by ∑ = −+= 2 1 2 12 )( 2 i i qqqcC λ ,(1) whe e c is he cons an ma ginal cos o p oduc ion wi hou in e empo al ou pu adjus men s (i.e., i 12 qq =). The second e m in exp ession (1) e lec s adjus men cos s (e.g. Hame mesh, 1996), wi h i λ deno ing he loca ion-speci ic deg ee o employmen p o ec ion. To al cos s consis o he sum o a iable cos s and ixed se -up cos s ( i F), wi h h F= φ and δφ += F. Tha is, i he i m loca es ab oad, he se up cos includes an in es men cos δ > 0 ha can be hough o as e lec ing he ba ie s o he mobili y o capi al. Hence, δ will ypically all as he deg ee o ma ke in eg a ion inc eases. 4 These assump ions allow us o ocus on he e ec o employmen p o ec ion on loca ion choice, while abs ac ing om o he loca ion de e minan s, such as ma ke access and cos ad an ages whose impo ance o i ms’ loca ion decisions is well unde s ood. See, o ins ance, Ma kusen (2002). 4 Expec ed p o i s a e equal o 21 πππ EE += . Fo wa d looking, when choosing i s op imal ou pu in pe iod wo, he monopolis maximises second-pe iod p o i s gi en 1 q, implying 0)(2122 =−−+− qquqAi λ ,(2a) wi h caA − ≡ . Assuming ha he i m is isk neu al,5 i de e mines i s op imal ou pu in pe iod one by maximising expec ed p o i wi h espec o pe iod-one ou pu , yielding he ollowing i s -o de condi ion 0)(2121 =−+− qEqqAi λ .(2b) The i m’s op imal ou pu in pe iod one and in pe iod wo a e, espec i ely, gi en by 2 1 A qi=, (3a) i i uA q λ + += 22 2.(3b) Exp ession (3b) clea ly shows ha he i m will be mo e lexible in pe iod wo i he local deg ee o employmen p o ec ion is low ( dudqi/ 2 dec eases in i λ ). F om he in e se demand unc ions, he i s -o de condi ions in (2a) and (2b), and since 12 ii qEq =, maximised expec ed p o i s can be w i en as i i ii F qE− + += 2 2 1)2( )(2 λ π (4) Expec ed p o i s dec ease in he deg ee o employmen p o ec ion ha p e ails in he local labou ma ke ( 0/< ii ddE λπ ).6 Hence, he loca ion decision o he monopolis will be in luenced by he di e ence be ween he expec ed p o i s in he wo loca ions – which depends on he di e ence in employmen p o ec ion be ween hem and on he cos o FDI ( δ ). I he ba ie s o FDI lows a e p ohibi i ely high (i.e. δ is e y la ge), he monopolis will se up he new plan in Home, ega dless o he size and sign o he employmen p o ec ion 5 Risk a e sion would simply make loca ing in he coun y wi h he mo e lexible labou ma ke ela i ely mo e a ac i e. 5 di e en ial be ween home and o eign. Wi h e y low FDI cos s, i.e. wi h e y high deg ees o capi al ma ke in eg a ion, a new i m seeking o minimise u u e adjus men cos s will choose o p oduce in he loca ion cha ac e ised by he lowe deg ee o employmen p o ec ion. A in e media e le els o FDI cos s, he monopolis will always loca e in Home i 0)( >− h λλ , and will loca e in o eign i )( h λλ − is nega i e and su icien ly la ge7. Hence, o non-p ohibi i ely high FDI cos s, a gi en inc ease in )( h λλ − will be mo e likely o de e FDI om h o he la ge is δ . 2.2. Employmen p o ec ion and eloca ion The e ec s o employmen p o ec ion on loca ion decisions will di e in he case o a eloca ion o an exis ing plan . Suppose ha a i m wi h an ope a ing plan in h conside s eloca ing o . Jus like a new i m – p o ided ha he FDI cos s a e no oo high – ce e is pa ibus, his i m is mo e likely o p oduce ab oad i i expec s ha u u e adjus men s in i s wo k o ce will be less cos ly o ca y ou in han in h. Howe e , i he e is some deg ee o employmen p o ec ion in h, eloca ion implies ha he i m will incu exi cos s in he o m o se e ance pay o laid-o wo ke s in i s h-plan . Hence, e en ela i ely mode a e home employmen p o ec ion may discou age eloca ion and e ec i ely se e as a ‘domes ic ancho age’ de ice, hus hinde ing ou wa d FDI. The ancho age e ec o employmen p o ec ion will also be concei ably s onge he highe a e FDI cos s, i.e. a i m may be e en mo e eluc an o ace he occu ence o se e ance cos s associa ed wi h eloca ion in he p esence o high FDI cos s. To summa ise, ou heo e ical analysis sugges s ha bo h he di e ence in employmen p o ec ion be ween and h ( h λλ −) and he ac ual le el o employmen 6 No e ha , since p o i s a e con ex in u, expec ed p o i s inc ease in ν . 7 These conclusions hold because he i m is a monopolis . Wi h oligopolis ic i al y, i ms’ loca ion decisions a e mo e complex (see Dewi , Leahy and Mon agna, 2003). 6 p o ec ion in h ( h λ ) may be nega i ely ela ed o ou wa d FDI ( om h o ). On he one hand, as a de e minan o he home ma ke ‘ lexibili y di e en ial’ ( h λλ −), a high le el o home employmen p o ec ion will encou age eloca ion; on he o he hand, by de e mining he ex en o exi cos s, i will ac as an ‘ancho ’ o he domes ic indus y, hus making in e na ional eloca ion less likely. The ex en o capi al ma ke in eg a ion is also likely o a ec he esponsi eness o in es men lows o employmen p o ec ion: ce e is pa ibus, bo h a highe lexibili y di e en ial in a ou o he home coun y – i.e. a la ge ( h λλ −) – and a highe le el o home EP a e likely o be ha ing a s onge nega i e e ec on ou wa d FDI lows he la ge is δ . 3. EMPIRICAL ANALYSIS In his sec ion we b ing empi ical e idence o bea on he heo e ical p edic ions o he p e ious analysis. To his end, we es ima e an empi ical model o he de e minan s o ou wa d FDI lows om home coun y h o pa ne coun y using OECD da a. The eg ession equa ion is gi en by: h h h h h h h h XFDI εβδλλγδλγλλγλγα ++−++−++= ]*)[()*()()ln( 4321 (5) whe e λ h is a measu e o employmen p o ec ion (EP) in home coun y h a ime and ( λ – λ h ) is he di e ence in employmen p o ec ion a ime be ween hos and home coun y. Acco ding o ou heo e ical analysis, o he hings equal, he highe he le el o employmen p o ec ion in he o eign loca ion, he lowe will be he home coun y’s in es men in ha loca ion. We he e o e would expec he coe icien γ 2 o be nega i e. The coe icien γ 1 cap u es he ‘domes ic ancho age’ e ec desc ibed in he heo e ical model and is also expec ed o be nega i e. The heo y also sugges s ha he sensi i i y o FDI lows o employmen p o ec ion may depend on he le el o in es men cos s. In o de o cap u e his e ec , we include in e ac ion e ms o ou employmen p o ec ion a iables wi h in es men 7 cos , mul iplying λ h and ( λ - λ h ) by a dummy a iable δ equal o one i in es men cos s a e highe han he median in he sample.8 The ec o X cap u es a numbe o addi ional co a ia es ha po en ially a ec he loca ion o FDI. These a e: (1) he le el o pa ne coun y GDP, which is included in o de o con ol o he ma ke size o he hos economy. We would expec a posi i e ela ionship be ween his a iable and FDI in lows in o he pa ne coun y (see Culem, 1988). (2) he le el o home coun y GDP, included in o de o con ol o he size o he home coun y, which de e mines he supply o FDI (Blonigen, 1997). (3) a ade-weigh ed measu e o uni labou cos s, included o con ol o di e ences in he cos s o inpu s ac oss coun ies (see Culem, 1988). (4) a dummy a iable which is equal o one i home and pa ne coun y a e membe s o a ee ade a ea. This is included o cap u e he po en ially posi i e e ec o membe ship o a ade ag eemen on FDI (see, o example, Ba ell and Pain, 1999) The da a o all a iables come om he OECD. Da a on bila e al FDI lows a e aken om he OECD’s In e na ional In es men S a is ics Yea book. The index o employmen p o ec ion o each coun y – as desc ibed by Nicole i e al. (1999) – is based on measu es o p o ec ion a ec ing he coun y’s empo a y and egula employmen .9 The index – which akes on alues be ween 0 and 100 om he leas o he mos es ic i e – is a ailable o only wo yea s (namely 1989 and 1998), hus es ic ing conside ably he ime 8 One may a gue ha we should also include he le el o in es men cos on i s own in he equa ion. Ou conce n is ha his a iable is highly co ela ed wi h he wo in e ac ion e ms (wi h co ela ion coe icien s o 0.64 and – 0.25 espec i ely) which can po en ially lead o mul icollinea i y p oblems in ou sample. In ac , es ima ions o equa ion (5) also including δ on i s own, which a e no epo ed he e bu a e a ailable om he au ho s upon eques , yield s a is ically insigni ican coe icien s on he in e ac ion e ms bu nega i e and s a is ically signi ican coe icien s on δ , as expec ed. The coe icien s on he o he a iables a e e y simila o hose epo ed below. Hence, in o de o e lec ou emphasis on he in e ac ion be ween δ and employmen p o ec ion we epo only he esul s o equa ion (5) wi hou he le el o in es men cos . 9 The employmen p o ec ion indica o s we e cons uc ed “based on an in-dep h e iew o exis ing egula ions and laws a ec ing he hi ing and i ing o wo ke s along he wo dimensions o egula and empo a y con ac s” (Nicole i e al., 1999, pp. 40-41). 8 se ies dimension o ou analysis. The cos o in es men ( δ ) is a measu e o he cos o capi al o in es men s om home o pa ne and is also an index be ween 0 and 100 (as is he uni labou cos a iable). All nominal a iables, i.e. FDI and GDP, a e con e ed in o eal 1995 US dolla s. Table 1 epo s summa y da a on a coun y’s o al inwa d and ou wa d FDI lows and i s employmen p o ec ion index o he wo yea s 1989 and 1998. Apa om he o me communis coun ies Czech Republic, Hunga y and Poland which ha e employmen p o ec ion indices o 100 o 1989,10 he coun ies wi h he highes le els o employmen p o ec ion among OECD coun ies a e Sou h Ko ea, Po ugal and Spain. A he o he ex eme we ind he US, UK and Canada wi h e y low le els o employmen p o ec ion in 1989. I is no able ha o mos coun ies o which he e ha e been changes be ween 1989 and 1998, he le el o employmen p o ec ion has allen, wi h he excep ion o F ance. [Table 1 he e] A i s sigh , hese da a only poin o a loose connec ion be ween FDI and employmen p o ec ion: o ins ance, Canada (a coun y wi h a ai ly low le el o EP) had ela i ely low le els o inwa d and ou wa d FDI in 1989 bu much highe le els in 1998. The na u e o he ela ionship, howe e , canno be adequa ely cap u ed by hese summa y da a. A be e app ecia ion o he e ec s o employmen p o ec ion on FDI lows can be gained by es ima ing ou empi ical model desc ibed in equa ion (5) using bila e al da a. Gi en ha he employmen p o ec ion index is only a ailable o wo yea s, we es ima e wo independen c oss sec ions, using he da a o 1989 and 1998 sepa a ely. The esul s o he es ima ions o di e en speci ica ions o equa ion (5) a e epo ed in Table 2. Fo he yea 1989 we ha e 277 coun y pai s a ailable while he e a e 388 in 1998.11 10 No e ha hese coun ies a e no included in he econome ic es ima ions o 1989, gi en he absence o FDI da a o ha yea . 11 In he es ima ions we allow o he e oskedas ici y o he e o e m, as well as an unspeci ied co ela ion be ween e o e ms wi hin, bu no ac oss, pa ne coun ies. This allows o he possibili y ha he e may be 15 Table 3: Reg ession esul s wi h in e ac ion e ms (Dependen a iable: ln FDI lows om home o pa ne ) (1) (2) Pa ne GDP 0.601 0.658 (0.091)*** (0.089)*** 1998 * Pa ne GDP 0.033 -0.058 (0.113) (0.110) Home GDP 0.967 1.031 (0.105)*** (0.106)*** 1998 * Home GDP -0.080 -0.097 (0.118) (0.120) Pa ne uni labou cos 0.354 0.678 (0.622) (0.587) 1998 * Pa ne uni labou cos 0.260 -0.438 (0.857) (0.825) Home EP -0.016 -0.025 (0.010) (0.009)*** 1998 * Home EP -0.026 -0.041 (0.013)** (0.013)*** Home EP* In es men cos -0.022 -0.019 (0.007)*** (0.006)*** 1998 * (Home EP * In es men cos ) -0.019 -0.006 (0.009)** (0.009) EP di e en ial -0.023 -0.027 (0.006)*** (0.006)*** 1998 * EP di e en ial -0.002 -0.007 (0.009) (0.008) EP di e en ial * In es men cos -0.001 0.002 (0.009) (0.008) 1998 * (EP di e en ial * In es men cos ) -0.022 -0.024 (0.012)* (0.011)** FTA dummy 1.339 (0.282)*** 1998 * FTA dummy 0.152 (0.343) Obse a ions 665 665 F (in e ac ion e ms) 4.85*** 2.93*** R-squa ed 0.38 0.45 He e oskedas ici y consis en s anda d e o s, clus e ed by coun y-pai s, in pa en heses * signi ican a 10%; ** signi ican a 5%; ***signi ican a 1% eg ession includes cons an and (1998 * cons an ) e ms