The Political Economy of Services Trade Agreements

Transcription

1 The Politicl Economy of Services Trde Agreements Mtteo Fiorini Mthilde Lebrnd CESIFO WORKING PAPER NO CATEGORY 8: TRADE POLICY MAY 2016 An electronic version of the pper my be downloded from the SSRN website: from the RePEc website: from the CESifo website: Twww.CESifo-group.org/wpT ISSN

2 CESifo Working Pper No The Politicl Economy of Services Trde Agreements Abstrct Why do governments sign services trde greements? This pper focuses on the role of interntionl greements in the context of trde in services when services re used s intermedite inputs in downstrem industries. Compred to goods, services inputs re mostly non-trdble nd complementry to other fctors of production. We build theoreticl trde policy frmework in which firms use foreign investment to contest foreign mrkets in services sectors nd governments cn restrict the entry of multintionls. Commitment helps governments to void politicl pressures tht would result in protectionist mesures leding downstrem industries to inefficiently reduce their production. First we show tht the role of services s complementry inputs is centrl to explin governments commitment to services trde liberliztion. Second we provide new results on the influence of lobbying by both ntionl firms nd foreign multintionls on trde policies nd the gins from commitment. Finlly we discuss how the brgining power of the government, the size of ntionl services sectors nd the difference in vlution between ntionl nd foreign contributions ffect the willingness of the government to sign services trde greement. JEL-Codes: D430, F130, F210, L800. Keywords: trde in services, trde greements, FDI, lobbying. Mtteo Fiorini Robert Schumn Centre for Advnced Studies ( EUI) Itly Sn Domenico di Fiesole (FI) mtteo.fiorini@eui.eu Mthilde Lebrnd Deprtment of Economics Europen University Institute Itly Florence mthilde.lebrnd@eui.eu My 27, 2016 We wish to thnk Bernrd Hoekmn, Pol Conconi, Andre Mttozzi, Fernndo Veg- Redondo nd Piero Gottrdi for their constnt guidnce nd support. We lso thnk prticipnts t the ETSG 2014 in Munich, the XII ELSNIT Conference in Florence nd the 7th FIW Reserch Conference in Vienn, s well s seminr prticipnts t EUI nd University of Montrel for their useful comments.

3 1 Introduction Triff brriers hve lmost completely disppered for mnufctured goods but significnt restrictions remin for interntionl services trnsctions. Restrictive policies re observed for professionl nd trnsporttion services in both industril nd developing countries, nd for most sectors in some of the fstest growing countries in Asi nd the oil-rich Gulf sttes (Borchert et l., 2012). 1 Due to the intngibility nd non-storbility of mny services, their interntionl provision requires the estblishment of commercil presence in the importing country. As consequence, restrictions on foreign cquisitions, discrimintion in licensing nd other investment restrictions represent de fcto brriers to trde in services. Liberliztion in service sectors hs become key element of the interntionl trde gend nd services trde greements hve imed t promoting integrtion in the services mrkets. The Trde in Services Agreement (TiSA) is recent inititive to push for multilterl negotitions. Services re lso the min topic of the Mrket Access chpter tht is negotited in the Trnstlntic Trde nd Investment Prtnership (TTIP) between the United-Sttes nd the Europen Union. A key stylized fct here is tht interntionl greements hve done very little to estblish higher ccess to service mrkets. 2 Why the current frmework of services trde greements is not conducive to the desired policy outcomes is question tht deserves creful considertion nd, bove ll, clls for deeper understnding of the motives behind commitment to services trde liberliztion through n interntionl greement. While there is little in the wy of theoreticl investigtion on the rtionles/motives behind services trde greements, severl hypothesis hve been dvnced regrding the pplicbility to services of the min explntions for the existence of interntionl trde greements. First, it hs been rgued tht the predominnce of foreign direct investment (FDI) in interntionl services trnsctions wekens the terms-of-trde rtionle for most services sectors (Hoekmn, 2008 nd Frncois nd Hoekmn, 2010). The terms-of-trde motive is prominent rgument to explin the existence of trde greement s wy out of prisoner dilemm where ech government imposes high brriers in order to improve its terms of trde t the expense of others (Johnson, 1954 nd Bgwell nd Stiger, 1999). With FDI, protectionist mesures tht ffect world prices benefit both ntionl firms nd foreign ffilites which ll produce in the sme country nd sell t the sme price. 3 Second, the commitment motive developed by Mggi nd Rodriguez-Clre (1998) rests on ssumptions (perfect competition, trde in finl goods) nd 1 According to the informtion provided by the World Bnk Services Trde Restrictiveness dtbse (the richest source of services trde policy outcomes in terms of country coverge) the highest brriers to services trde re observed in the high-income GCC countries, South nd Est Asi, the Middle Est nd North Afric. Ltin Americ, Estern Europe nd OECD countries re reltively more open while Sub-Shrn Afric shows n intermedite level of services trde openness. 2 In prctice, unilterl policy reforms remin the prime chnnel through which the mjor (though incomplete) steps towrd liberliztion hve been mde so fr. This key stylized fct is discussed in detils in Frncois nd Hoekmn (2010). 3 An dditionl rgument revisiting the terms-of-trde motive is given in Blnchrd (2010) where the termsof-trde motive is wekened by the rise of interntionl ownership. Governments hve less incentives to set high triffs given tht some domestic firms mke profits from importing goods nd suffer from high triffs. In the cse of services, interntionl ownership mostly ims t ccessing mrket when cross border exports re not possible or not optiml. 1

4 mechnisms (protectionism s source of mislloction given countries comprtive dvntges) tht do not fit the specificities of services. In their pper, n interntionl greement serves s commitment device for the government to distnce itself from domestic politicl pressures which would induce inefficient investment in the sector without comprtive dvntge. Protectionism in services sectors does not primrily reduce the gins from comprtive dvntge mechnism (s it hppens in tht pper), but mostly reduces the benefits from more competition in sectors where monopolies, lrge entry costs or stte-owned firms re predominnt. Moreover, these uthors only consider finl goods wheres services re mjor inputs for downstrem producers (Frncois, 1990). The lck of theory tht mtches the specificities of services clls for further work. The present pper constitutes first ttempt in rtionlizing the motives behind services trde greements ccounting for the chrcteristics of services production nd trde. We build new model whose ssumptions fit some key stylized fcts bout services. First, services re importnt inputs tht re often complementry to other fctors; second, trde in services often requires commercil presence; third, services sectors re chrcterized by significnt mrket imperfections. These premises re shown to be empiriclly importnt for trde in services s well s less relevnt for trde in goods. A new model tht relies on these three fetures is required to understnd the politicl economy of trde in services. Our frmework fetures n open economy in which perfectly competitive downstrem sector produces finl consumption good using service nd non-service intermedites s complementry inputs. Service inputs cn be imported by llowing foreign multintionls to estblish commercil presence (engge in FDI) nd operte domesticlly. Ntionl services providers nd foreign firms ffilites operte under regime of oligopolistic competition. The government chooses the number of foreign providers llowed to contest the services mrket, with free entry representing the welfre-mximizing policy option. Firms cn form lobbies nd give contributions in order to hve n impct on the government s choice. We show the existence of commitment motive for services greements. The government uses trde policies s substitute for domestic reforms in order to ttrct foreign multintionls. Imperfect competition in the services sector leds ntionl providers to exert politicl pressures nd lobby for protection ginst foreign entry. However higher services prices due to protectionist policies leds to underinvestment in the complementry non-services inputs nd lower production of the consumption good. Similrly to Mggi nd Rodriguez-Clre (1998), we revel time inconsistency problem for the government given tht the long run losses due to the protection of the services sector - which triggers n inefficient scle of the finl good industry - re not compensted by the politicl rents. As consequence, the government cn end up worse off when subject to politicl pressures thn under free entry. In this cse there is scope for commitment vi n interntionl greement s tool for the government to credibly distnce itself from the specil interest groups nd to prevent their lobbying ctivity. Our first contribution is to extent their rgument to the cse of trde in services nd show which chrcteristics of services production nd trde re necessry to rech similr conclusions. The literture on services trde policies (Hoekmn, 2008 nd Frncois nd Hoekmn, 2010) supports the existence of such commitment motive in which greements ct s commitment 2

5 device to void domestic politicl pressures even in the bsence of reciprocity chnnels. One of our contributions is to show tht the complementrity between services nd non-services inputs in downstrem production is the key ssumption to support the existence of commitment motive for services. An dditionl contribution of our pper is to consider the politicl influence of foreign firms on the choice of trde policies. We llow foreign providers to colesce into lobby nd to offer contributions directly to the government. However we ssume tht governments tend to vlue unit of foreign contributions less thn the sme unit of domestic contributions (Gwnde et l., 2006). Two interesting results emerge. First we show tht the foreign lobby s optiml policy is not free-entry. The foreign lobby supports lower brriers, i.e. more entry, till the gins from n dditionl entry re less thn the losses due to lower individul profits for ll ffilites. We show tht whether the foreign lobby pushes for higher or lower mrket ccess depends on the number of ntionl services providers. We find tht the foreign lobby supports more entry when the number of ntionl firms is lrge. The economic intuition is simple. With high number of ntionl firms, the contribution bse of the ntionl lobby is lrge nd high brriers tend to be chosen. The gins from n dditionl foreign firm re higher thn the sum of the loses for ll foreign firms. In tht cse foreign lobbying is welfre improving s it countercts the effect of ntionl lobbying. Second we find tht the government s vlution of contributions from foreign firms compred to those from ntionl firms shpes the commitment choice of the government in non trivil wy. In our new frmework, we study how the brgining power of the government, its vlution of foreign contributions nd the number of ntionl services providers ll interct to explin commitment. The pper reltes to the literture in severl respects. First, we provide the existing debte on the motives behind services trde greements (Hoekmn, 2008 nd Frncois nd Hoekmn, 2010) with forml nlysis of the commitment rtionle for the specific cse of trde in services through commercil presence. In doing so, we dd to the theoreticl work on liberliztion in producer services (see for instnce Frncois, 1990b, Frncois, 1990 nd Mrkusen et l., 2005). From these models, we keep the ssumptions on the role of services s intermedite inputs nd the relevnce of investment brriers for services trde. We dd politicl economy dimension to nlyze the trde policy outcomes nd the role of greements in cse of politicl pressures. 4 Our contribution is to develop model tht relies on these ssumptions. Some ppers hve focused on the effect of the rise of trde in intermedite inputs for commercil policy nd the design of trde greement for goods (Antràs nd Stiger, 2012 nd Blnchrd, 2015). In our pper we go further by focusing on the complementrity between services nd other inputs nd studying how this ffects the willingness of the government to commit to n greement. We lso contribute to the theoreticl literture on the motives behind interntionl greements. We drw from the seminl work on the commitment motive (Mggi nd Rodriguez-Clre, 1998) mking it compelling for the cse of trde in services. One of our contributions is to trnslte their rgument into model more suitble for services. We do this by deprting from compr- 4 A similr politicl economy frmework cn be found in Fung nd Siu (2008). These uthors however look t the role of politicl economy forces only for the contingent trde policy outcome while our focus is on the decision to commit vi n interntionl greement. 3

6 tive dvntge model of trde in finl goods, by introducing imperfect competition nd, most importntly, by focusing on entry brriers insted of triffs s the relevnt policy instruments. We consider model with intermedite services tht re used for downstrem production. Wheres the time inconsistency problem in Mggi nd Rodriguez-Clre (1998) relies on firms choice of sectors in which to initilly invest, our time inconsistency mechnism relies on firms initil investment in non-services inputs tht re complementry to services inputs. Under-investment in complement non-services inputs becomes constrint for the governement s ction s inefficient investments in sectors with comprtive disvntge ws in their pper. The mechnism is similr but the economic story differs in the cse of services nd relies on the complementrity ssumption of services inputs. Furthermore, the present pper speks to existing nlysis of foreign lobbying s determinnt of trde policy outcomes (see for instnce Gwnde et l., 2006). Compred to the existing studies, we consider tht the foreign lobby cts s third gent in the brgining gme. His objective differs from the objectives of both the government nd the lobby of ntionl firms. In line with Antràs nd Pdró i Miquel (2011), we study how foreign lobbying ffects the willingness of governments to commit. Finlly, our frmework hs implictions for the empiricl study of commitment to services trde liberliztion through interntionl greements (Egger nd Lnz, 2008). It suggests tht the degree of services input intensity, the complementrity between intermedite services nd other fctors of production s well s the impct of foreign lobbying nd the size of the ntionl services sector should be considered s determinnts of prticiption into services trde nd investment greements. The rest of the pper is orgnized s follows. Section 2 discusses some key empiricl fcts tht cpture the specificities of services production nd trde. Section 3 presents our bseline theoreticl model nd Section 4 extends it to llow for foreign lobbying. Section 5 offers conclusions. 2 Stylized fcts In this prt we detil the specificities of services on which we bse our modeling ssumptions to study the role of services trde greements. These specificities re shown to be empiriclly relevnt for trde in services wheres they re often less importnt for trde in goods. First services re specil inputs into modern production processes, from quntittive s well s qulittive perspective. Producer services (trnsporttion, communiction, ccounting, consulting, finncil, computer design services, etc) re importnt suppliers to the rest of the whole economy. 5 Good finncil nd trnsporttion services re necessry to expnd the production 5 Averging cross the 63 countries covered by the OECD Inter Country Input Output (ICIO) dtbse, for ech dollr of ntionl production in 2011, 4.7 cents cover the costs of R&D nd business services s intermedite inputs, 3.7 cents the cost of trnsporttion services, 3.2 cents tht of finncil services. Additionlly, there is often n incresing trend in the evolution of services input intensity from 1995 to 2011: this is the cse especilly for ICT services, whose cost s intermedite inputs reltive to one dollr of totl output goes from 0.43 cents in 1995 to 0.86 cents in 2011.OECD ICIO tbles cover OECD countries, the BRICS nd severl other high nd middle income countries. For detiled country coverge of the OECD ICIO tbles see http: 4

7 nd trde of finl goods. As consequence, improving the ccess nd the qulity of producer services is crucil step for the rest of the economy to expnd. This is supported by the empiricl studies tht nlyze the effect of reforms in the services sector on downstrem mnufcturing industries (Arnold et l., 2011, Fernndes nd Punov, 2011). The min result of those studies is tht higher services productivity, better ccess to producer services nd pro-competitive reforms trgeting the service industry hve positive impct on the economic performnce of sectors downstrem the supply chin, notbly mnufcturing. Wht is interesting for our pper is tht the entry of multintionls in the services sectors is key to get positive effect. 6 Services inputs re therefore key to improve the productivity of the whole economy. However the qulittive difference between services inputs nd other types or inputs is tht services re often complement inputs in the production function. Overll, services help to coordinte nd control specilized opertions involving the other fctors of production (Frncois, 1990 nd Mrkusen et l., 2005). This is modeled in Frncois (1990) s complementrity reltionship between producer services nd other inputs in the production function of the finl good. 7 This ssumption of complementrity (Leontief production function) is widely used in CGE models which ssume tht firms production function exhibits fixed proportions mong its inputs (Burfisher, 2011 nd UK.GOV, 2014). More recently Jones (2011), Boehm et l. (2015), nd Krishn nd Levchenko (2013) hve been working with the Leontief ssumption to discuss the role of input linkges in the propgtion of shocks. Hrrign et l. (2016) lso uses the complementry ssumption between ICT services nd skilled workers to study the impct of technology on job polriztion. In our model, the ssumption of complementrity is going to be centrl to understnd the motives for commitments in the cse of trde in services. An dditionl difference between services inputs nd goods inputs pertins to their reltive role in totl trde. For the cse of developed OECD countries nlyzed by Miroudot et l. (2009) cross the ten yers from 1995 to 2005, trde in intermedite goods inputs represents 56% of overll trde in goods. The sme figure increses by some 30% for services trde, where the flows in intermedite services inputs ccount for 73% of totl services trde. Moreover, trde in services minly hppens through horizontl FDIs, s opposed to cross-border trde. This is min difference between trde in goods nd trde in services which comes from the nture of services. Services re typiclly intngible nd non-storble (Jensen, 2011), such tht production nd consumption must occur t the sme plce nd time. In order to cover ll the possible wys to exchnge services, the GATS negotitors hve defined four mode of services provision: Mode 1 for cross-border trde; Mode 2 when customers move in the exporter s country; Mode 3 for commercil presence (lso clled horizontl FDI); finlly Mode 4 when the exporter s personnel move to the customer s country. In this pper we only focus on horizontl FDIs or Mode 3. This is relevnt becuse lrge shre of trde in service, especilly trde in producer // 6 These empiricl works include number of country cse studies with firm-level dt (such s Arnold et l., 2011, Fernndes nd Punov, 2011, Duggn et l., 2013 nd Arnold et l., 2016) s well s few cross-country nlysis, often with dt t the sector-level (see for instnce Arnold et l., 2008, Brone nd Cingno, 2011, Bourlès et l., 2013, Hoekmn nd Shepherd, forthcoming nd Beverelli et l., 2015). 7 More precisely, given tht the only fctor of production in Frncois (1990) is lbor, the complementrity is defined between the lbor directly employed for specilized production opertions nd the lbor employed in producer services. 5

8 yer yer services, hppens through horizontl FDI (mode 3). Figure 1 shows the evolution of trde in services through commercil presence (Mode 3 trde) for US exports nd imports respectively. Dt on trde in services through commercil presence is not cptured in the Blnce of Pyment Sttistics. A good mesure of mode 3 services trde is given by the Foreign Affilites in Trde in Services (FATS) dt. Mode 3 exports of ntionl economy re mesured s the sles (brod) of foreign ffilites of ntionl firms. Speculrly, mode 3 imports by reference country re given by the the sles of those services providers tht, while locted in the country, re foreign ffilites of non ntionl firms. From the beginning of the nineties when both mode 3 imports nd exports hd vlue of pproximtely 100 billion USD, these trde figures hve incresed to 800 billion USD for the cse of imports nd lmost to 1300 billion USD - which corresponds to pproximtely one twelfth of totl US GDP - for the cse of exports in This cnnot be explined just by decrese in brriers to FDI: indeed, lrge prt of services tht hve incresingly been trded re intngible nd cn be provided through FDI only. Figure 1: Mode 3 Services Trde in the US. 8 () Mode 3 Exports, US (b) Mode 3 Imports, US USD billions USD billions Figure 2 compres the figures of trde through commercil presence with those of the trde through cross border exchnge for both goods nd services in two mjor trde economies, the US nd Germny. Pnel 2 (2c) plots the rtio between mode 3 trde s mesured by turnover of US (Germn) multi-ntionl enterprises in the rest of the World over the vlue of US (Germn) gross-exports. For the cse of the US (pnel 2), the solid line shows tht the rtio is quite stble round vlue of 2 from 2002 to 2008 for mnufcturing goods. For services insted the sme rtio is lwys bove 4 nd remins pretty close to 5 from 2005 to When looking t Germn export this pttern is even more pronounced: in the cse of goods, the vlue of cross border exports (t the denumertor of the rtio) is lwys higher thn tht of mode 3 exports (the numertor) mking the mnufcturing rtio (solid line) lwys below 1. The sme rtio for services is insted often bove vlue of 5 showing tht mode 3 exports s opposed to cross border ones re significntly more relevnt for services rther thn merchndise trde. Anlogous conclusions bout the importnce of mode 3 flows - for services more thn for goods - cn be 8 Foreign Affilite Trde in Services dt used in Figure 1 come from the US Bureu of Economic Anlysis. 6

9 drwn from the ptterns of mode 3 nd cross border imports represented in pnels 2b nd 2d. Figure 2: Mode 3 VS Cross border trde (Goods VS Services). 9 () US Exports (b) US Imports MNEs outwrd turnover / Exports MNEs inwrd turnover / Imports Yer Yer Mnufcturing Services Mnufcturing Services (c) Germny Exports (d) Germny Imports MNEs outwrd turnover / Exports MNEs inwrd turnover / Imports Yer Yer Mnufcturing Services Mnufcturing Services Mode 3 trde is relevnt for services not only quntittively. An importnt finding from the empiricl trde literture is tht the presence of foreign firms is the key chnnel for successful liberliztion reforms (Arnold et l., 2011 nd Fernndes nd Punov, 2011). 10 There re two explntory chnnels: foreign firms bring know-how nd knowledge bout new products, nd foreign firms increse competition in sectors where monopolies nd high entry costs re predominnt. We focus on the lst explntion which is prticulrly relevnt in services sectors. Indeed services sectors re often chrcterized by strong degree of mrket concentrtion. In their nlysis of competition cross US industries, Epifni nd Gnci (2011) show tht services sectors hve mrk-ups which re from two to three times bigger thn the verge mrk-up in the mnufcturing industry. In the sme pper the uthors rgue tht the degree of mrket imperfec- 9 Figure 2 combines different sources: dt on turnover of foreign ffilites is from the OECD FATS Dtbse; dt on cross border (mode 1 nd mode 2) services trde is from OECD Trde in Services, EBOP Sttistics Dt on mnufcturing trde is tken from WITS. 10 Arnold et l. (2011) highlight tht foreign entry in services sectors is the key chnnel to improve performnce in the mnufcturing sectors. Foreign firms bring know-how nd knowledge bout new products. Their presence my lso led domestic providers to improve the qulity of their products. The positive role of foreign entrnts ws lso shown by Jvorcik et l. (2008). They focus on the Mexicn detergent industry nd find tht the entry of Wlmrt reduced the distribution cost for detergent mnufcturers. Finlly Fernndes nd Punov (2011) studies the impct of FDI inflows in producer service sectors on the productivity of Chilen mnufcturing firms. 7

10 tion is linked to the trdbility of n industry s output, highlighting the linkges between mrket structure nd trde policy. Anlogously, domestic mrket imperfections hve been identified s key determinnts for trde policy ction (Hoekmn, 2008 nd Frncois nd Hoekmn, 2010). In mny cses, more liberl trde policies ct s substitutes for domestic reforms to mke services sectors more competitive nd to improve the provision of intermedite services. The three fcts presented in this section hve importnt implictions for the politicl economy of services trde policy. First, protectionist mesures tht restrict foreign entry in services sectors directly ffect downstrem producers. Therefore lobbying from services firms to protect their interests ffect the behvior of ll producers tht use intermedite services nd thus crete inefficiencies in the provision of consumption goods. We study here how the government rects to politicl pressures nd chooses trde policies by weighting the interests of consumers nd of ll producers. Second, horizontl FDI is the min mode of services provision which mens tht the relevnt trde policies here re brriers tht restrict the entry of multintionls. Exmples of such policy instruments re foreign equity quots, limit on the number of licenses vilble, ntionlity requirements for key personnel nd discrimintory licensing criteri. 11 Moreover, the mrket power of services firms suggests tht ntionl providers re willing to buy protection for their oligopolistic profits ginst the entry of foreign providers. Given tht the role of foreign firms is crucil in improving domestic services sectors, foreign firms should lso be considered s importnt plyers in the politicl gme between the government nd services firms. Foreign firms will lso lobby in order to ffect the choice of the domestic government. In the next section we develop theoreticl frmework tht ccounts for the role of services s complementry inputs, the role of FDI in trde in services nd the mrket power of services providers. Our positive nlysis of services trde policy embeds the politicl economy forces described bove. 3 Theoreticl Model 3.1 The set up We consider one Home country (H), nd the rest of the world. The economic structure in H consists of 2 sectors: downstrem sector (D) nd n intermediry services sector (S). ssume representtive consumer receiving utility from the finl good domesticlly-produced by the downstrem sector nd from numerire tht represents the rest of production. 12 Formlly, utility is given by: U(x 0, x D ) = x 0 + u(x D ) st. x 0 + P D x D R (3.1) with x 0 nd x D being the consumption respectively of the numerire nd the finl good, nd R the consumer s income. The demnd is function of price nd is ssumed to be strictly decresing nd twice continuously differentible. The function u is qusi-liner nd defined by 11 For comprehensive discussion on the reltionship between modes of provision nd policy instruments see Frncois nd Hoekmn (2010). 12 The numerire is produced with constnt returns to scle with cost per unit of one which implies tht its price is equl to one. We 8

11 u(x) = vx x2 2. This implies tht the demnd for the finl good only depends on its price, not on the consumer s income. The downstrem sector (or finl-good sector) consists of unit mesure continuum of smll firms, operting under perfect competition. The firms produce finl good using services nd other inputs. We model ll non-services inputs s n input bundle denoted by k nd owned by consumers. Consistently with the coordintion role of services inputs, we ssume complementrity reltionship between services nd k. Downstrem producers operte through the following production function: F (k, d S ) = f(min{k, d S }) (3.2) with the usul ssumptions on f being incresing nd strictly concve. k nd d S re the demnds for the input bundle nd services from ech individul downstrem producer. 13 We focus here on prtil equilibrium model neglecting the impct of services trde policy for non-services upstrem mrkets. The supply of k is ssumed unlimited nd its price is fixed nd denoted by τ. The profit of downstrem producer is given by π D (k, d S ) = P D F (k, d S ) P S d S τk (3.3) with P D the price of the finl good nd P S the price of services inputs. All downstrem firms re identicl so tht the ggregte profit function is given by Π D = 1 0 πd dj = π D. Similrly, the ggregte demnd for the input bundle is K = k nd the ggregte demnd of services inputs is D S = d S. The services sector is chrcterised by finite number n of firms tht operte under regime of oligopolistic competition à l Cournot. We ssume tht services re non trdble through cross-border exports. Foreign services cn only be provided through commercil presence (FDI / Mode 3). 14 In order to ccess the H mrket, foreign firms hve to estblish commercil presence (open n ffilite) in the country nd employ domestic (H) specific fctors. We ssume no cost for opening n ffilite. The revenues from specific fctor ownership is given by the profits of the services sector. 15 Once foreign firms re llowed to contest the H mrket, there is no difference between Home firm nd Foreign firm ny more. Services cn therefore be provided by either Home ntionl firms or by ffilites of foreign services providers. We denote by n d the number of ntionl providers nd we tke it s n exogenous prmeter. 16 The remining n n d services firms operting within the domestic mrket re ffilites of foreign firm. Notice tht the owners of the specific fctor used by the ffilites of foreign providers own clims on the services sector s profits in the sme wy s those gents whose specific fctor is used by the ntionl services firms. All the n services providers re identicl in terms of production 13 Notice tht our ssumption on the competitive structure of sector D implies tht finl-goods producers tke inputs nd output prices s given. By doing so we prevent ny strtegic behviour mong downstrem firms. 14 We do not model trde in downstrem production. 15 We ssume tht the production of services requires only the use of sector-specific fctor tht is vilble in inelstic supply nd non-trdble. The sector-specific model hs been used in the literture s exemplified by Grossmn nd Helpmn (1994) or Buzrd (2014). 16 This implies tht our model hs short-term perspective tht does not consider the impct of new ntionl firms. 9

12 cpcity nd they ll tke the price of the finl good P D s given. The profits of services firm re given by: π S (q S ) = q S P S (Q S + q S ) (3.4) with P S ( ) the price of services given by the strtegic behvior of the firms in Cournot frmework. This price depends on the ggregte supply of services, which is equl to the individul supply of the firm q S plus the ggregte supply of ll the others, Q S. We now turn to the politicl structure of the economy. We ssume ntionl government in country H tht cn choose the level of mrket ccess in the services sector. More precisely, the government tkes s given the number of ntionl firms n d nd sets vlue for n which directly determines the number of foreign suppliers llowed to operte in the country. Firms in the services sector re ble to colesce into lobbies tht cn offer contributions to the government. The rtionle behind the lobbying ctivity comes from oligopolistic competition in the services sector. Indeed, services providers mke positive profits depending on the their totl number n. Importntly, two lobbies cn be formed: the ntionl lobby tht represents ntionl services providers nd the foreign lobby tht ggregtes the interests of foreign firms. Denoting with c nd c f the contributions of the ntionl nd the foreign lobby respectively, their objective function re given by: L(n, c) = n d π S (n) c (3.5) nd L f (n, c) = (n n d )π S (n) c f (3.6) In cse of inctive lobbies the government mximises the welfre given by the sum of the consumer s surplus, the services nd downstrem producers surplus nd the rents from the input bundle. 17 Given tht the n n d foreign ffilites use domestic specific fctors, the rents from the ntionl firms nd the foreign ffilites re eqully considered in the revenue of the representtive consumer. The socil welfre mximised by the government is given by: W (n) = H(n) + Π D (n) + Π S (n) + τk (3.7) where H(n) is the consumer surplus tht only depends on the price P D (n), 18 nd Π S (n) = n d π S (n) + (n n d )π S (n) ntionl firms foreign ffilites (3.8) When insted specil interest groups re ctive, ech lobby pys contribution to the government in order to hve n influence on the trde policy n. The government s objective is the weighted sum of the socil welfre nd contributions: G(n, c) = W (n) + c + γc f (3.9) 17 This results is given by the definition of the welfre in prtil equilibrium with qusi-liner utility function nd numerire tht represents the rest of the economy. 18 The consumer surplus is H(P D ) = (v P D )

13 where is the weight on the socil welfre. The prmeter γ vries between 0 nd 1 nd mesures to which extent foreign contributions re vlued with respect to ntionl ones. Following Gwnde et l. (2006), we ssume imperfect foreign lobbying: the government will never vlue one dollr of foreign contributions strictly more thn one dollr of ntionl ones. 19 When γ = 0, foreign lobbying is considered s forbidden or totlly inefficient. When insted γ = 1, foreign contributions re perfectly vlued by the government, i.e. there is no difference in vlution between ntionl nd foreign contributions. The timing of the economy is the following. At t = 0 there re n d domestic firms nd no foreign services providers. At the first stge, t = 1, the downstrem producers buy the non-services input bundle K. No production tkes plce. In the second stge, t = 2, the government chooses its trde policy by fixing n, the totl number of services firms llowed to operte in the mrket. We cll this the trde policy stge. Notice tht this timing reflects the ide tht both the government nd the lobbies lck commitment with respect to the downstrem producers. Relevnt downstrem economic decisions, exemplified by the purchse of the input bundle K, hve to be tken s given in the politicl gme. Finlly, t t = 3, production in the services sector tkes plce. The price of services P S is determined nd the downstrem firms buy services inputs. Production nd consumption of the finl goods tke plce. We re now redy to solve for the equilibrium of the model. We strt deriving benchmrk solution from the simplest cse of ntionl lobbying only. Section 4 extends the discussion to foreign lobbying. 3.2 The equilibrium with ntionl lobbying We proceed by bckwrd induction. We strt describing the incentives of the producers nd consumers in the finl stge (t = 3), given the level of K nd the trde policy decided in the previous periods. Then we move to the trde policy stge where, for given K, politicl gme between the government nd the ctive lobby determines the number of foreign firms to enter the domestic services sector. Finlly, we solve for the optiml ggregte level of the input bundle purchsed by downstrem producers t t = 1. In this section we ssume no foreign lobbying (γ = 0) The finl stge (t = 3) Optiml demnd for services inputs. At t = 3, downstrem producers hve lredy bought the input bundle k nd its cost τ is therefore sunk. 20 With k given, the demnd for services of 19 The rtionle behind γ 1 is linked to the risks run by policymkers when ccepting contributions from foreign entities, this being potentilly considered s non consistent with ntionl interests nd therefore punished by the voters. 20 If the two decisions on k nd services inputs hd tken plce simultneously, the mximistion problem of representtive downstrem firm would hve been given by mx P D f(min{k, d S }) P S d S τk {k,d S } with solution d S = k nd k determined from P D f (k) = P S + τ. 11

14 n individul downstrem producer solves the following mximistion problem mx P D f(d S ) P S d S τk subject to d S k (3.10) d S nd it is implicitly defined s function of the two prices d S (P S, P D ). Finlly, ll firms re similr so the ggregte demnd of services is D S (P S, P D ) with D S ( ) = 1 0 ds ( )di = d S ( ). D S cn be written s D S (P S, P D ) = (f ) 1 ( P S P D ) if (f ) 1 ( P S P D ) K K otherwise (3.11) Optiml behviour of the services firms. The problem of services firm is s follows: mx q S qs P S (Q S + q S ) (3.12) with P S (Q S + q S ) = P D f (Q S + q S ) given the oligopolistic frmework. We use the ggregte demnd in services goods (3.11) considering non binding constrint. 21 In ddition, we impose the functionl form f(x) = 1 e x A given by: with A positive constnt. 22 The ggregte services supply is Q S (P D ) = na (3.13) Optiml behviour of the representtive consumer finl good. The demnd function is The consumer only consumes the D D (P D ) = v P D (3.14) Mrket clering First, the ggregte supply Q S of services hs to mtch the demnd from ll downstrem producers Q S (P D ) = D S (P S, P D ) (3.15) Second, the ggregte production of the finl good hs to equlize the demnd D D ( ) of the representtive consumer f(k, D S (P S, P D )) = D D (P D ) (3.16) The equilibrium of production. Given the level of K nd the trde policy n, the equilibrium in the production stge is defined by the finl good price P D (K, n) nd the services price P S (K, n) tht stisfy the optiml behviour of the representtive consumer, the optiml behviour of the services nd downstrem firms nd the mrket clering conditions for the finl goods nd the service intermedite input. 21 The focus on the cse where (f ) 1 ( P S ) K will be rtionlised below when setting the vlue for τ. 22 P D This functionl form is chosen for trctbility resons. It results in constnt optiml individul supply function q S (P D ) = A. 12

15 Downstrem firms use services nd the complementry input bundle nd choose their production pln ccording to the prices of the two inputs (P S, τ). For simplicity, we ssume for the rest of the pper tht the price of k is low enough so tht the complementrity of inputs implies tht only the reltive price of the services ffects the finl firms decision. Therefore the initil ggregte purchse of the input bundle is not binding - K D S - nd the demnd function for the service input is simply given by D S (P S, P D ) = (f ) 1 ( P S P D ). This imply the following price equtions P D (K, n) = v f(na) (3.17) P S (K, n) = [v f(na)]f (na) (3.18) Finlly, stndrd comprtive sttics nlysis shows tht n increse in the totl number of services firms n (i) decreses the price of services inputs P S, (ii) decreses the finl price P D, nd (iii) decreses the profit mde by ech services firm The trde policy stge (t = 2) Trde policy n nd contributions c re determined through coopertive brgining gme between the government nd the lobby (Mggi nd Rodriguez-Clre, 1998). The thret point is given by the trde policy n 0, chosen by the government in cse of no politicl distortions, nd zero contributions. n 0 coincides with free entry for foreign services providers. 24 Following Binmore et l. (1986), we ssume tht the two plyers re different in terms of procedurl bilities nd preferences (time preferences or ttitudes towrds risk), which implies heterogeneity in brgining power. We denote the brgining weights σ nd 1 σ, for the government nd the lobby respectively. The solution (n N, c N ) mximises the Nsh product: 25 (n N, c N ) = rgmx (c,n) R R + [W (n K) + c W (n 0 K)] σ [n d π S (n K) c n d π S (n 0 K)] 1 σ (3.19) Notice tht the solution of the politicl gme (n N, c N ) with the ntionl lobby is efficient, i.e. it mximizes the joint surplus of the government nd of the ntionl lobby given by J N (n) = W (n K) + n d π S (n K) (3.20) A detiled derivtion of this result is given in Appendix A. Finlly, it is importnt to highlight tht the thret point gives the government utility W (n 0 K) which differs from the one in gme with no ctive lobbies. If the politicl distortions ssocited with specil interest groups re removed, downstrem producers do not expect ny form of protectionism to increse the price of services inputs. On the contrry, in the politicl gme 23 In order to keep these comprtive sttics results strightforwrd we ssume v > f( ). This llows the derivtive of π S with respect to n to be negtive for ny positive vlue of A. For nlogous technicl resons, pertining to subsequent derivtions, we further ssume v to be smller thn Techniclly this is given by setting the individul profits of services firms equl to 0: π S (n) = The superscript N denotes the solution of the politicl gme when only the ntionl lobby is ctive. 13

16 with lobbying, the ggregte purchse of the input bundle (K) is mde given the expecttions of politicl frictions over the trde policy decision The initil stge (t = 1) Downstrem firms re ssumed to be smll nd, s consequence, unble to individully ffect the ggregte demnd for the input bundle. Given the complementrity between services nd k, the optiml demnd for services determined in the finl stge (t = 3) represents n upper threshold for k s demnd. We cn write the initil stge mximistion progrm of downstrem producers using the expected services demnd from the finl stge d S = (f ) 1 ( P S P D ). mx P D f(k) P S d S τk subject to k (f ) 1 ( P S k P D ) (3.21) When the constrint is not binding the solution is given by the first order condition P D f (k) = τ, otherwise k is equl to the optiml unconstrined services demnd. This defines the ggregte demnd for k: K(P S, P D ) = min (f ) 1 ( τ P D ), (f ) 1 ( P S P D ) (3.22) The ssumption on τ (low enough to gurntee tht K (f ) 1 ( P S P D ) in the finl stge) reduces the ggregte demnd for k to K(P S, P D ) = (f ) 1 ( P S P D ) (3.23) Given our ssumption of n unlimited supply of k, ggregte demnd is lwys met. Importntly, our simple frmework implies tht the ggregte demnd for k is decresing in the reltive price of services. Therefore, if services become more expensive, the downstrem producers will demnd less of the other inputs Equilibrium We cn now define the equilibrium of the model when the government plys politicl gme with the ntionl lobby. Definition 1 (Equilibrium under ntionl lobbying) The equilibrium of the model with inctive foreign lobbying is defined by the vector E N = {n N, c N, K N, (P S ) N, (P D ) N } (3.24) tht stisfies the optiml behviour of the representtive consumer in (3.14); the optiml demnd of services nd k from downstrem firms given by equtions(3.11) nd (3.23); the optiml supply from services firms in (3.13); the optiml trde policy (3.19); nd ll mrket clering conditions, (3.15) nd (3.16). 14

17 The equilibrium reflects coming from the economic s well s politicl structure of our setting. The government wnts high entry of foreign firms in order to decrese the price of services. On the contrry, the number of ntionl firms is fixed nd their ggregte profits therefore only depends on the individul profit of ech firm. Individul profits in the services sector re decresing in the trde policy prmeter n which defines the number of domestic (ntionl nd foreign) services suppliers. Therefore, the rent owners vlue protection of the services sector ginst foreign entrnts. Their first-best trde policy would be closed domestic services sector (n = n d ). In order to influence the trde policy choice, the ntionl lobby offers contributions to the government through brgining process. The equilibrium trde policy results in number of domesticlly ctive foreign providers tht is non zero (closed economy) but lower thn the free-entry level (fully open economy). Techniclly, the equilibrium trde policy n N nd contributions c N re derived from the first order conditions of the Nsh product - eqution (3.19) - fter hving expressed ll other relevnt quntities s function of n. In prticulr, the complementrity ssumption between services nd k nd the mrket clering condition for services imply tht K = D S = Q S = na. The key properties of the equilibrium re given in the following Proposition 1 The equilibrium E N exists nd is unique. The equilibrium trde policy n N is given by n N = ln (1 v)( n d ) 2n d (3.25) which is incresing in the government s socil vlution () nd decresing in the number of ntionl services providers (n d ). Proof. See Appendix A. The equilibrium number of services firms n N increses in the government s vlution of socil welfre (the prmeter). Intuitively, given tht setting no entry restrictions is the welfre mximising trde policy, if the government puts less weight on politicl rents (contributions), it will reduce the brriers to foreign entry by choosing higher vlue of n. Insted, the totl number of services providers decreses in the exogenous number of ntionl services firms, n d. A lrger ntionl services sector mkes it more rewrding for the government to decrese the number of foreign entries. 3.3 The politicl gme versus unilterl commitment Similrly to Mggi nd Rodriguez-Clre (1998), we llow the government to commit to its optiml trde policy t t = 0, i.e. before ny economic decision is tken by the downstrem producers. We ssume tht government commitment t t = 0 is perfectly enforceble. As consequence, if the government commits t the beginning, the subsequent trde policy stge does not tke plce nd there is no politicl gme between the government nd the lobby. The welfre mximising trde policy n 0 is ssocited with no entry restrictions for foreign services firms. It is given by 15

18 the following zero-profit condition in the services sector π S (n 0 ) = 0 v f(n 0 A) = 0 n 0 = ln(1 v) (3.26) Notice tht positive contributions in the equilibrium with the politicl gme ssure tht n N < n 0 : indeed, the ntionl lobby is willing to offer contributions with the precise objective to increse protection beyond the welfre mximising policy n Importntly, the ggregte demnd for the input bundle under commitment is given by K 0 = n 0 A. This quntity is strictly bigger thn K N, the ggregte demnd for k in the equilibrium without commitment. Downstrem producers nticipte the distorted equilibrium trde policy when the government chooses not to commit. This implies higher services price thn under commitment. Given the complementrity of services input nd k, higher price for services leds to smller demnd for the input bundle. Formlly K 0 K N = Aln n d 2n d > 0 (3.27) The difference in the equilibrium ggregte input bundle is decresing in governments vlution of socil welfre nd incresing in the number of ntionl services firms n d. Formlly (K 0 K N ) 0 nd (K 0 K N ) n d 0 (3.28) The intuition behind this result is simple. First, government tht cres more bout socil welfre (higher vlue of ) is less influenced by contributions from services firms. The policy chosen in the politicl gme is then closer to its optiml when there is no lobbying. Therefore, downstrem producers expect the price of services to be closer to the free-entry benchmrk. This will llow them to buy more services t t = 3. Anticipting this higher demnd for services nd given the complementrity of services nd k, downstrem producers buy more input bundle t t = 1. Secondly, higher number of ntionl firms (higher n d ) implies tht the ntionl lobby is stronger nd tht the government cres more bout the individul profits of the ntionl firms. This leds the government to choose higher restrictions (lower vlue of n), which in turns imply higher services price. Downstrem producers will then buy less services t t = 3. The services-k complementrity in production nd the nticiption of lower demnd for services in the finl stge, led the finl good firms to buy less of k t t = 1. We cn now describe the optiml decision of the government t t = 0, whether to commit to n 0 or to do nothing, witing to ply the politicl gme with the lobby during the subsequent trde policy stge. Similr to Mggi nd Rodriguez-Clre (1998), the timing is importnt nd cretes time-inconsistency problem tht explins why the government might not lwys opt for the politicl gme. Indeed, when plying the politicl gme t t = 2, the government is constrined by the choice of the downstrem producers from t = 1. The optiml lloction gives 26 See the proof of Proposition 1 for the technicl restrictions on the prmeters implied by well defined equilibrium with ntionl lobbying. 16

19 the government his outside option plus shre of the surplus. However the outside option is not the first-best ny more. Indeed, the lower ggregte purchse of k by downstrem firms prevents the government to get W (n 0 ). The government gets its constrined first-best W (n 0 K) plus shre of the surplus tht depends on its brgining power nd therefore it be worse off with respect to the benchmrk cse where it obtins W (n 0 ). Under commitment insted, the government hs the opportunity to set its trde policy t t = 0 before ny relevnt economic decision by the downstrem producers. The enforcement is perfect such tht the policy chosen cnnot be renegotited. In other words, the government cn commit to free-entry (n 0 ) nd tie his hnds vis-à-vis the specil interest group. The trde-off between the rents from the lobby (cost of commitment) nd the lower input bundle purchse by the downstrem firms (cost of the politicl gme) pins down the vlue of commitment, defined s the difference between the objective of the government under commitment - G 0 - nd the one under the politicl gme, G N : Ω = G 0 G N = W (n 0, K 0 ) W (n N, K N ) c N (3.29) The following proposition detils the choice of the government t t = 0. Proposition 2 (i) If σ = 0, the government lwys benefits from commitment: Ω(σ = 0) > 0. (ii) There exists unique threshold V (0, 1) such tht v < V,! σ (0, 1) such tht Ω( σ) = 0. Moreover, when the government hs wek brgining power (σ < σ) it benefits from commitment (Ω(σ) > 0); when insted the government hs strong brgining power (σ > σ) it benefits from the politicl gme (Ω(σ) < 0). Proof. See Appendix A. Proposition 2 shows tht, when the government hs low brgining power, the rents it cn extrct from the lobby bove its reservtion utility re smll. 27 Therefore the cost of commitment (foregone contributions) is smller thn the cost of plying the politicl gme (lower purchse of k) nd the government optimlly chooses to initilly commit to unilterl trde policy. On the contrry, strong government cn extrct big politicl rents (bove its reservtion utility) nd therefore chooses not to commit t t = 0. Under technicl restriction on the prmeter v, there lwys exists level of brgining power bove which the cost of commitment is lrger thn the cost of plying the politicl gme. Crucil for the decision to commit is the ssumption of complementrity between services nd other intermedite inputs represented by k. While the time inconsistency setting is not new to the literture on the commitment motive, the inefficiency coming from input complementrity is, to the best of our knowledge, n originl contribution of our frmework. To further clrify the relevnce of our complementrity ssumption, let us tke the opposite stnce nd ssume perfect substitutbility between services nd k in downstrem production. In tht cse the government 27 The threshold σ hs the following nlyticl expression σ = [W (n0 K0) W (n0 KN )] Π(n N K N ) Π(n 0 K N ) tht comes from the loss of the government by not choosing its first-best when plying the brgining gme (numertor) nd from the gins of the lobby tht re to be shred (denomintor). 17