Less Developed Countries, Tourism Investments and Local. Economic Development

Transcription

1 Less Develoed Countries, Tourism Investments and Local Economic Develoment Rainer Andergassen, Guido Candela Setember 7, 2009 Deartment of Economics, University of Bologna, Piazza Scaravilli 2, 4026, Bologna, Italy Abstract Te resent aer analyzes te forward linkages of a multinational s investment in a resort wic kicks off tourism activity in a LDC. We sow tat, under quite natural assumtions, overnigt stays are increasing in te number of tourism related, differentiated goods and services sulied in equilibrium. Tese goods and services, if sulied by te local community, reresent forward linkages of FDI in tourism. We investigate te multinational s incentives to romote, reduce or eliminate tese forward linkages and te effectiveness of some olicy instruments available to a local government to leverage on te resence of FDI and to stimulate domestic entrereneursi. Keywords: Tourism FDI; multinationals; economic develoment; domination strategies; LDC JEL: L83; F23; O9; Te autors would like to tank Roberto Cellini and Antonello Scorcu for elful comments. Corresonding autor. addresses: rainer.andergassen@unibo.it, guido.candela@unibo.it

2 Introduction Many olicy makers view FDI as a way to furter te rocess of wealt creation in te ost country and to el to jum-start te develoment rocess in less develoed countries LDCs). Te imlications in terms of economic olicy are te creation of a favorable economic environment adat to attract FDI, consisting, for examle, of secial tax treatment, exemtion from imort duties or direct subsidies. Academic literature studies te welfare effects of FDI on bot te ost and te ome country, unveiling otential cannels of welfare gains as well as welfare losses. FDI create backward and forward linkages in te ost country Rodriguez-Claire, 996), tat is vertical and orizontal externalities 2. Te negative cometitive ressure effects of FDI are countered by roductivity gains owing to tecnology transfer and diffusion, access to knowledge caital and, on te account of learning by doing and learning by observing, skill transfer. Tecnological ugrade and efficiency gains may also be induced and accelerated by te cometitive ressure effects of FDI, wic drive inefficient firms in te ost country out of te market. Linkage effects on welfare may be ositive or negative, deending on te revalence of ositive roductivity-efficiency effects and negative crowding out effects of local firms. Te ost country may also benefit from market access sill-overs, wic are due to te knowledge acquisition of international marketing, te access to international distribution markets and te oening u of international markets. In te ome country, FDI may lead to roductivity increases because of economies of scale and learning curve effects, wic counter negative emloyment effects resulting from te international reallocation of roduction facilities. In te resent aer we study te effects of a multinational s tourism investment on local economic develoment 3. FDI is often considered to be one of te most effective engines for arnessing te develoment of caital, infrastructure and knowledge and for granting access to global marketing in tourism. Terefore, LDCs often view te attraction of suc investments as a leading tourism and economic develoment strategy but te teoretical and emirical imlications of tourism FDI on te ost country are not well studied and understood 4. FDI, wic is mostly concentrated in otels and restaurants, is most imortant for emerging tourism economies, wile it is relatively less imortant for develoed and mature tourism markets. UNCTAD For a survey see, for examle, Markusen 995), Blomstrom and Kokko 998), Hanson 200), Lisey 2004), Navaretti and Venables 2004). 2 FDI are often related to imerfection in te tecnology market, i.e. tacit knowledge, and framed in Dunning s OLI aradigm, tat is ownersi, location and internalization advantage see Dunning, 98, 993 and Markusen 995 for a discussion). 3 In a comanion aer by Antonelli and Candela 2008) te relationsi between multinationals and te develoment of a tourism destination is investigated. 4 Te literature on FDI in tourism consists mainly of case studies. See, for examle, UNCTAD 2007). 2

3 2007) documents te strong imact of FDI on consumer demand in new destinations. For examle, FDI layed a major role in te tourism take-off in Tunisia in te 970s, wen investors suc as Club Med invented te concet of a beac resort. In a similar vein, in te Dominican Reublic, desite governmental investments to start a tourism industry in te 970s, te real boom did not occur before te massive influx of FDI from te 980s onwards. UNCTAD 2007) documents also te case of Butan and te United Reublic of Tanzania were sar increases in tourist arrival numbers aear to be directly linked to FDI. FDI raise te olicy callenge of taking full advantage of tese investments, tat is to leverage on te resence of FDI to foster domestic entrereneursi. Tese olicies consists, for examle, in minimizing leakages 5, in imroving te tourism value cain and, more generally, olicies aimed at boosting te net value of tourism tat ave te effect of increasing tourists lengt of stay. Te role of FDI in tourism is more nuanced tan it is in some oter sectors of te economy, and it is valued because of te kick off to te tourism industry wic it can rovide, but it is also feared for its imact on economic and cultural indeendence, and its otential damage to te communities and te environment UNCTAD, 2007). Te country faces te risk of foreign domination, suc as enclaves and oter strategies were te multinational limits te benefits for te ost country and were te country is terefore unable to caitalize on suc investments. Tis is te case, for examle, of Punta Cana, a Dominican Reublic island, were transnational cororations built small enclaves equied wit teir own essential services suc as ower, waste and water management and access modes UNCTAD, 2007). Broman 996) argues tat foreign domination and external deendency often seriously reduce tourism s otential for generating broadly based growt and takes te case of te Caribbean as an examle, were a satial olarization witin bot te modern tourism industry, based on resort enclaves in te most desirable coastal locations, and te older agricultural-based economy emerged. Battilani 2002) documents te investment strategy of a ool of entrereneurs led by Karim Aga Kan IV in te tourism untouced Costa Smeralda in Sardegna Italy) during te 960 s, wo eld te develoment of tourism and in general of te economic region under tigt control acquiring a monooly osition on te otentially usable land and controlling te administration and te develoment of te region. We consider te case of a multinational wo builds and owns a resort in a new destination in a LDC and study te multinational s incentives to make te local community articiant of te tourism 5 Leakages in te tourism sense, tat is negative national balance-of-ayment effects suc as te imorts of goods and te reatriation of rofits. 3

4 develoment. We sow tat under quite natural assumtions, tourists overnigt stays are increasing in te degree of diversification of tourism related roducts and services. Terefore, te multinational s rofits are strictly related to te variety of tourism related goods and services and consequently te multinational as an incentive to control its suly. In articular, te multinational faces two otions. On te one and te multinational can try to control te variety of goods and services sulied by local firms and on te oter and it can create an enclave and suly directly te referred variety. In te former case, in wic tourism is based on culturally autentic goods and services, we assume tat eac differentiated roduct is roduced by a single firm and ence te number of tourism related roducts sulied is a roxy for te local economic develoment induced by te multinational s investments. In te latter case, in wic tourism is based on a fake but autentically staged local variety, te local community does not benefit from te tourism activity. Our analysis focuses on tese forward linkages of FDI, tat is te creation of new markets in a LDC troug te tourism roduct, and we neglect backward linkages establised by resorts, tat is suliers of intermediate consumtion goods. We consider a small oen subsistence economy. All firms are rice taker on te international tourism market. We assume tat te roduction of differentiated roducts necessitates a fixed set u cost, wic accounts for te sadow rice of land and te cost of turning unroductive land into roductive land 6. Available and otentially usable land is in erfect elastic suly u to a fixed level and its suly is assumed to be erfectly inelastic afterwards. In Section 2 we caracterize te reresentative consumer s demand function consisting of overnigt stays, consumtion of differentiated, tourism related roducts and non-tourism related roducts. We sow ow te multinational s rofits from overnigt stays and te suly of differentiated roducts are related and tat, from te multinational s viewoint, tere exists an otimal degree of roduct diversification. Terefore, if tourism roducts are sulied by te local community, were te local variety is determined by a free entry condition, ten tere may be too many or too few roducts from te multinational s viewoint. As a consequence, te multinational may ave an incentive to eiter foster or curb local economic develoment or eventually to substitute te local autentic variety of tourism goods wit an autentically staged variety sulied by te multinational. We sow tat, for te multinational, subsidizing local develoment, tat is subsidizing te local firm s set u cost, is 6 As argued above, tourism related FDI is most imortant for new destinations and in emerging economies. Once te destination develos and becomes mature, te relative imortance of FDI diminises. Terefore, in tis context crowding out effects of FDI are negligible and we accordingly assume tat te sadow rice of land, wic accounts for alternative uses of land, is unaffected by te multinational s investment strategy. 4

5 never otimal. On te oter and we sow tat te multinational may curb local develoment, by buying all te available land and reselling art of it at a ositive margin. We furter sow tat from te multinational s rofits viewoint tis is equivalent to creating an enclave were te local variety is autentically staged by te multinational. Te difference between tese two strategies is tat in te former case te variety is truly autentic, wile in te second case it is an autentically staged variety. We call tese two strategies domination strategies since te multinational controls tourism develoment in te ost country 7. In Section 3 we discuss te effectiveness of olicy instruments available to a local government wo s aim is to maximize local economic develoment. We sow tat taxing te multinational s activity as well as subsidizing directly local economic develoment by subsidizing te firm s set u cost may well be counterroductive since it increases te multinational s incentives to engage in a domination strategy. On te oter and, we sow tat infrastructural investments aimed at increasing te otentially available land may be beneficial since tey reduce te multinational s gains from domination strategies. Section 4 concludes. All roofs are in te Aendix. 2 Te model In te first art of tis section we describe te reresentative consumer s tourism demand function. We ten caracterize te suly side. We calculate te equilibrium number of differentiated, locally sulied, tourism related roducts wic is determined by a free entry condition and te otimal suly of tourism related roducts from te multinational s oint of view. Finally, we caracterize te multinational s otimal domination strategy. 2. Tourism demand In tis Section we caracterize tourism demand consisting of nigts sent in a destination, tat is overnigt stays in te resort, and consumtion of tourism related roducts. We consider a reresentative consumer endowed wit a constant elasticity of substitution CES) utility function exibiting 7 A furter difference between te two strategies is tat in te case of an enclave te roduction rocess is a vertically integrated one, were te multinational acts as a coordinator of te suly, wile in te case of a autentic local variety te lack of coordination among roducers may lead to a subotimal outcome from te destination s viewoint. 5

6 Dixit and Stiglitz 977) love of variety references for differentiated tourism related goods U y,, x,..., x i,..., x n )= { y β + n ) ] } γ β β γ γ + i= xα α i ) were y indicates a comosite, non-touristic good, indicates overnigt stays and x i, for i =,..., n, indicates differentiated tourism related goods. We assume tat at least one variety as to be offered suc tat tourism is viable, i.e. n. We call T te tourism roduct, consisting of overnigt stays ) and differentiated tourism related roducts {x i } n i= ), i.e. T =, {x i} n i= ). Te reresentative consumer faces te budget constraint y + + n i= ix i = R 2) were R is is income, is te rice of a single overnigt stay, i te rice of x i and were te rice of te non-touristic good y as been normalized to. Trougout te aer we assume te following. Assumtion i) 0 < β <, ii) < γ < 0, iii) 0 < α <, iv) α > β. Assumtion i) imlies tat te non-touristic good y and te tourism roduct T are gross substitutes; for β tey are erfect substitutes. Assumtion ii) imlies tat overnigt stays and tourism related roducts are gross comlements, were for γ tey are erfect comlements. Assumtion iii) imlies tat goods/services x i, i =,...,n, are gross substitutes. Assumtion iv) states tat te gross substitutability of tourism related goods/services is greater tan te gross substitutability between te non-touristic goods y) and te tourism roduct T ). We define λ β = = β β 0, ), γ γ, 0), λ α = α α 0, ). We assume symmetry on te suly side were i = and terefore in equilibrium x i = x, for i =,...,n. Assumtion 2 Prices and and income R are suc tat te following inequalities are satisfied: i) ) ] + λ ) ] β +λ β ) + >Rz > + λ β ii) >and < λ ) γ λ β 6

7 Assumtion 2 i) guarantees tat te below defined rofit maximization roblem of te multinational as an internal maximum. In articular, if te first inequality is not satisfied ten te referred number of tourism related roducts to be sulied, if te multinational does not engage in a domination strategy, is te minimum ossible, i.e. n =. If te second inequality is not satisfied, ten te solution diverges to infinity. Assumtion 2 ii) states tat te rice of an overnigt stay is larger tan te rice of tourism related roducts and furter tat te substitutability between te non-touristic good y) and te tourism roduct T ) is sufficiently low. Maximizing ) subject to te budget constraint 2) we obtain te following result. Lemma Te otimal number of overnigt stays is strictly increasing in n, i.e. n) > 0; ii) te demand of differentiated roducts is strictly deceasing in n, i.e. x n) < 0; iii) te demand of te non-touristic good is strictly decreasing in n, i.e. y n) < 0. Lemma relates te degree of differentiation of te tourism roduct n) to te number of overnigt stays. In articular, it states tat te more "soisticated" is te tourism roduct, tat is te greater is n, te more days tourists send in a given destination and ence te greater is teir exenditure on overnigt stays. A sufficient condition for tis result to old is tat eiter overnigt stays and tourism related roducts are gross comlements, or tat te non-touristic good and te tourism roduct T are gross substitutes 8. Te assumtion about te gross substitutability of non-touristic goods wit te tourism roduct T imlies tat an increase in n decreases demand for and exenditure on non-touristic goods and terefore increases te consumers exenditure on touristic goods. Moreover, Assumtions i) and iv) guarantee tat, as n increases, te demand of x decreases increases) at a faster ace tan if β =0. Te assumtion of greater substitutability between tourism related goods tan between te non-touristic good and te tourism roduct T leads consumers to send relatively more on overnigt stays and relatively less on tourism related roducts as n increases. 2.2 Suly side For simlicity s sake we assume tat available and otentially usable land is in erfect elastic suly u to a fixed level N and its suly is assumed to be erfectly inelastic afterwards. Te firm s set u cost is c. 8 If β =0, tat is te non-touristic good y and te tourism roduct T are indeendent, and γ =0, tat is if overnigt stays and te differentiated tourism goods are indeendent, ten te demand for overnigt stays is indeendent of n. 7

8 In tis section we focus on te case were differentiated roducts and services are roduced by local firms, tat is te case of a culturally autentic local variety, and sow ow te multinational s interests may be conflicting wit tose of a local government wose aim we assume to be te romotion of local economic develoment. We assume tat eac differentiated roduct/service is roduced by a single firm and tat roduction cost are nil 9. Te free entry condition for local firms is x n L) = c 3) wic imlicitly defines te number of firms oerating in te tourism sector n L. Notice tat n L if c is sufficiently low, tat uniqueness is guaranteed by te fact tat x n) is strictly decreasing in n and tat n L is decreasing in c. We also assume tat N nl. In order to calculate te otimal degree of tourism roduct differentiation from te multinational s viewoint, we calculate te otimal resort caacity in terms of overnigt stays. Assuming a quadratic cost function for te roduction of overnigt stays z 2 2, te multinational s rofits are 0 π = z 2 2 Since overnigt stays deend on n we can caracterize n M, wic is te otimal n from te multinational s viewoint. First order condition yields n M, wic is imlicitly defined by n M) = z. Assumtion 2 togeter wit n) > 0 guarantee tat tere exists a unique n M wic is strictly increasing in z. Te following Lemma establises te relationsi between n M and n L. Lemma 2 Given z, and, let c e,n e ) be te unique solution to x n e )= ce n e )= z 4) ten: i) for c = c e, n M = n L = n e ; ii) for c > c e, n M n L > 0 and were te difference is increasing in c; iii) c < c e, n L n M > 0 and were te difference is decreasing in c. 9 Assuming in addition to fixed setu costs constant marginal costs of roduction would not cange te qualitative results. 0 We normalize te multinational s fixed costs or set u costs), to zero, and consider only oerating revenues and costs. Taking entry cost into account would affect te multinational s decision to enter te destination or not. 8

9 According to Lemma 2, deending on te value c takes, te local community s interest, tat is to romote local economic develoment, and te multinational s, wic lies in te exloitation of rofit oortunities, may eiter be conflicting or aligned. In articular, tere exists a critical cost level c e above wic te degree of roduct differentiation rovided by te local community is too low comared wit te multinational s otimal level and below wic it is too large. Tese conflicting interests raise te roblem tat te multinational may try to control te rovision of tourism related roducts, tereby sustaining or amering local economic develoment. 2.3 Te multinational s domination strategy We consider te case were a) te multinational finances local develoment by subsidizing local firms set u cost wic becomes c + s, s c, 0) and b) te multinational buys all te available land N at te rice c, becoming in tis way a monoolist, and resells it at a cost c + s, s 0, ). Te former case may be relevant if te local community sulies too few differentiated roducts comared wit te multinational s otimal level, wile te latter case addresses te issue of an excess-differentiation. Later on we argue tat, from te multinational s viewoint, te latter situation is equivalent to creating an enclave, te only difference being tat if tourism related roducts are roduced by local firms, ten tey are culturally autentic goods wile in te case of an enclave tey consist of fake, but autentically staged roducts. Given te value of s, if differentiated goods and services are roduced by local firms, ten te free entry condition 3) reads x n) =c + s 5) wic imlicitly defines te equilibrium number of local firms n s), wit n s) < 0. In te resent context, because of 5), coosing s is equivalent to coosing n. Terefore, te multinational s roblem is to coose n wic maximizes rofits from its core business, tat is rofits from overnigt stays, and rofits or losses related to te direct or indirect control of te degree of roduct differentiation. Consider first te multinational s incentives to subsidizes local develoment. For tis urose we ave to comare te multinational s rofits if it subsidizes local develoment, tat is max s 0 π s), were π s) = π n s)) + sn s), wit tose if it does not intervene, i.e. π n L). Te following roosition can be roved. Proosition Subsidizing local develoment is never otimal. 9

10 Te intuition for tis result is as follows. Because of te assumtion of greater gross substitutability between te single tourism related goods tan between te non-touristic good and te tourism roduct T Assumtion iv)), te demand x n) decreases at a faster ace as n increases. As a consequence, it is more costly to increase n and te gains from an increase in n are outweiged by te cost of doing so. Notice tat a consequence of Proosition is tat, if an enclave is otimal, ten te degree of roduct differentiation will never be larger tan te one rovided by te local community. Te intuition for tis is as follows. As argued below, te multinational s rofits are te same if it creates an enclave or if it buys all te land and resells it at a margin or discount to local roducers. In subsidizing local firms, te multinational oerates as if it buys all te requested land for a given, announced s<0 and resells it to local firms at a rice c + s. Since subsidizing is never otimal, it follows tat it is never otimal to buy more tan te required amount of land for an announced s<0and to resell it at a discounted rice c + s. Terefore, if subsidizing te set u cost is never otimal, ten it is never otimal to create an enclave wit more differentiated roducts. Let us consider te case were te multinational amers local develoment or creates an enclave. Te multinational acquires a monooly osition by buying all te available land N and ten resells it at an increased rice c + s, s 0, ). Since Proosition states tat s c, 0) is never otimal, te multinational s roblem can be written as max Πs) =π n s)) + c + s) n s) cn 6) s or using 5), equivalently max Πn) =π n)+x n) n cn 7) n Te first element of 6) and/or 7) are rofits related to te firm s core business, wile te remaining two elements are associated wit te buying and selling of land i.e. domination strategy). Notice tat, in te case of an enclave, te multinational s rofits are given by 7). Te reason for tis is as follows. To create an enclave te multinational as to acquire a monooly osition, tat is to eliminate all otential cometition from local roducers tird element in 7)). Selling te autentically staged variety to te tourists yields te multinational roceeds x n) n, te second element in 7), wile te We imlicitly assume tat te tourist s demand is indeendent of weter tourism related goods are autentic or fake, but autentically staged, and tat te roduction costs are te same. Te qualitative results stated in te aer old also if tis assumtion is to be relaxed, tat is if tourists refer one over te oter, and/or if te roduction costs differ. 0

11 first element in 7) reresents te multinational s roceeds from te core business. In coosing n, or equivalently s, te multinational as to trade off gains/costs from its core business against gains/costs related to te domination strategy. We first sow tat a unique n = arg max n Πn) to te roblem 7) exists and, using 5), we caracterize te corresonding s. We ten derive conditions suc tat restraining local develoment, or equivalently creating an enclave, is otimal. Lemma 3 Tere exists a unique n, wit z n < 0. Since s and n are related troug 5), Lemma 3 states tat tere exists a unique s wic maximizes te multinational s rofits. Moreover, it states tat s is increasing in te cost arameter z. Te reason for tis result is tat te larger are te costs associated wit its core business, te more rofitable becomes te domination strategy and ence it is otimal to increase s. Moreover, from 6) and Lemma 3 we observe tat n is indeendent of c and terefore, by using 5), we ave tat s is decreasing in c. Te following result caracterizes s. Lemma 4 For c c e, and also for some values of c > c e, s > 0. Lemma 4 states tat te multinational may ave an incentive to engage in a domination strategy even toug tis worsens its core business. Te intuition for tis result is te following. For c < c e, were n L >n M, a ositive s leads to gains in te core business as well as gains from te domination strategy. Consequently, s > 0 is otimal. If c = c e, ten from Lemma 2 we know tat te free entry condition leads to an equilibrium wit maximizes te multinational s rofits in te absence of domination activity, i.e. n L = n M = n e. Because of te local concavity of te rofit function π, increasing s leads to a second order loss in its core business and to a first order gain in its domination activity. As a consequence, engaging in a domination strategy is otimal for c = c e. By continuity te first order gain from a domination strategy is larger tan te second order loss in te core business also for some values of c > c e, were n L <n M. Summing u, we observe tat if n L n M, ten s > 0. Moreover, te larger te difference between n L and n M, i.e. te lower is c see Lemma 2), te larger is s and te larger are te gains associated wit te domination strategy. For some n L <n M, s > 0 is still otimal and increasing te difference between n L and n M, tat is increasing c see Lemma 2), decreases s. Te next roosition caracterizes te multinational s incentives to curb local develoment, or equivalently to create an enclave.

12 Proosition 2 Tere exists a tresold N suc tat for all N>N engaging in a domination strategy s > 0) is never otimal, wile for N < N engaging in a domination strategy is otimal. N is decreasing in c and increasing in z. Proosition 2 states tat te lower is te availability of otentially useful land N) and/or te lower are te set u costs of a local variety c), te more likely it is tat te multinational engages in a dominating strategy. Te intuition for tis result is tat te lower is N and/or c, te ceaer it is for te multinational to acquire a monooly osition and terefore to control local economic develoment. On te oter and, te larger is z, te larger are te costs associated wit its core business and terefore te more rofitable becomes te domination strategy. 3 Economic olicy In tis section we address te issue of te effectiveness of some economic olicy otions available to a local government wo aims at leveraging on te multinational s investment. In articular, we consider olicies wic consist in trying to reduce te enclave risk and to maximize local economic develoment. We analyze te consequences of i) subsidizing te set u cost c, ii) taxing multinational s activity in articular its inuts) and iii) increasing te otentially available land N. In case i) te government tries directly to foster local develoment, subsidizing te local firms set u costs. In case ii) we consider taxes on te multinational s urcases and activities for examle, taxing te multinational s imorts), wic we model by considering in te above analysis z instead of z, were z = z + t) and t>0 is te marginal, constant, tax rate 2. Te government may try in tis way to cas in on te multinational s activity. In case iii) te government invests in infrastructure, tereby increasing te amount of land wic may be ut to roductive use at unit cost c). Case i). We ave to distinguis between te case were engaging in a dominating strategy is and is not otimal. Consider first te case were te multinational does not engage in a dominating strategy N >N). Tourism roducts are sulied by local firms and subsidizing te firm s set u cost increases te degree of roduct differentiation and fosters local economic develoment. But since N is decreasing in c, a too strong subsidy leads te multinational to take a dominating osition. Once N<N, Lemma 3 states tat an otimal n exists, wic is indeendent of c and ence te effects of furter decreases in c are offset by increases in s. 3 Terefore, for N<N, tis olicy instrument is 2 Te range of variation of te tax rate t is restricted by Assumtion 2. 3 We imlicitly assume tat te government cannot tie in te granting of te subsidy wit a no-sale clause of te land. 2

13 ineffective. Consequently, economic olicy aimed at directly sustaining economic develoment may be artly, comletely or more tan offset by a multinational s domination strategy. Subsidies may ave te erverse effect of curbing instead of sustaining economic develoment. To see tis, consider te case of N being larger tan, but very close to N. A decrease in c may lead to N<N and terefore to a domination osition s > 0), wic may well lead to an overall reduction in n. Case ii). From Lemma 2, considering z instead of z, we observe tat c e is increasing in z, were, for c > c e, n M n L > 0 wit n M n L decreasing in z, wile for c < c e, n L n M > 0, wit n L n M increasing in z. 4 Lemma 4 and te ositive relationsi between N and z imlies tat te larger are taxes, tat is te larger is z, te greater are te multinational s incentives to engage in a domination strategy. Moreover, from Lemma 3 we know tat s is increasing in z. Te intuition for tese results is as follows. Increasing taxes t increases te multinational s costs associated wit its core business and ence a domination strategy becomes more attractive. As a consequence, as long as N is larger tan N, small taxes on te multinational s activity lead to tax yields and do not induce te multinational to engage in a domination strategy. But if taxes are too large and N < N, ten te multinational engages in a domination strategy, were larger taxes imly a stronger domination osition and terefore a stronger restrain on economic develoment. Terefore, te government as to trade off welfare benefits from tax yields against teir costs owing to a reduced economic develoment. If te government wants to romote economic develoment troug tourism, a comlete or artial tax exemtion for te multinational may be referable. Case iii). From te analysis above it follows tat increasing N, by increasing te multinational s costs of engaging in a domination strategy, decreases te rofitability of suc investments. Tis olicy unambiguously decreases te multinational s incentives to engage in a domination strategy and terefore increases te ossibility for te local community to benefit from tourism FDI. It remains an oen question ow to finance suc investments. Te government may well try to ursue a mixed strategy, tat is financing infrastructural investments troug taxes on te multinational s activity, under te constraint tat N>N. Moreover, te government may also try to tie in tourism FDI wit infrastructural investments, tat is to arove FDI only if tese are associated wit infrastructural investments aimed at avoiding domination strategies. 4 To see tis consider te case were c = c e. Given tat an increase decrease) in z leads to an increase decrease) in c e we are, for a constant value of c, in case Lemma 2 iii) in case Lemma 2 ii)) were, since n L is indeendent of z, n L = n e >n M n L = n e <n M ). 3

14 4 Conclusion Te resent aer analyzed te forward linkages of a multinational s investment in a resort tat kicks off te tourism sector in a LDC. We sowed tat, under Assumtion, overnigt stays are increasing in te number of differentiated tourism related goods and services, wic, if sulied by local firms, is a roxy for te economic develoment induced by FDI. We terefore argued tat te multinational as an incentive to control te variety of tese goods sulied in equilibrium. We sowed tat it is never otimal for a multinational to subsidizes local develoment and sowed tat te multinational may engage in a domination strategy aimed at reducing or even eliminating te ost country s benefits from te tourism activity. Tis result is in keeing wit te istorical evidence tat te resence of enclaves is associated wit a low degree of differentiation of tourism related roducts. We analyzed te effectiveness of some olicy instruments available to a local government wose goal is to leverage on te multinational s investments and we sowed tat domination strategies can be avoided by engaging in infrastructural investments designed to increase te otentially available land or by reducing taxes on multinationals acquisitions and exenses. Tis latter oint comlements te traditional argument in favor of tax reductions or exemtions for FDI based on te resumtion tat te suly of international caital is elastic relative to domestic caital see, for examle, Hanson, 200). Subsidizing directly local develoment may be counterroductive since it increases te multinational s incentives to adot a domination strategy. Tese olicy conclusions are in keeing wit some olicies undertaken by governments of LDCs. Endo 2006) reorts tat many LDCs grant te multinational duty free imorts. Even toug tis increases leakages, according to our analysis it decreases te multinational s incentives to engage in domination strategies and terefore it increases te ossibility for te ost country to leverage on FDI in te tourism sector. UNCTAD 2007) documents tat te majority of financial incentives to attract tourism related FDI consists in tax incentives tax rebates) and duty-free imorts of inuts. Battilani 200) documents tat starting from 978 te government as accomanied te strong influx of FDI in some small, uninabited islands of te Maldives wit infrastructural investments aimed at increasing te tourists mobility. According to our analysis, tese investments increased te multinationals costs of creating enclaves and terefore it increased te cances for te ost country to benefit from te tourism activity. In a similar vein, tourism in Punta Cana in te Dominican Reublic, wic was initially confined to small enclaves, took off only after investments in resorts by Club Med were tied to te construction of a new local international airort see UNCTAD, 2007), oening u tourism also 4

15 in oter areas and aving te way for te region s economic develoment. References ] Antonelli, L. and G. Candela 2008). Multinationals and te develoment of a tourism destination, mimeo. 2] Battilani, P. 200). Vacanze di oci vacanze di tutti, Il Mulino, Bologna. 3] Battilani, P. 2002). Rimini and Costa Smeralda: ow social values sae recreational sites. In S.C. Anderson, B. Tabb Ed.), Water Leisure and Culture, Oxford, Berg Publisers. 4] Blomstrom, M. and A. Kokko 998). Multinational Cororations and Sillovers, Journal of Economic Surveys, 2, ] Broman, J. 996). New directions in tourism for tird world develoment, Annals of Tourism Researc, 23, ] Dixit, A. K., J. E. Stiglitz 977). Monoolistic cometition and Otimum Product Diversity, American Economic Review, 67, ] Dunning, J.H. 98). International Production and te Multinational Enterrise. London, George, Allen and Unwin. 8] Dunning, J.H. 993). Te Globalization of Business. London, Routledge. 9] Endo, K. 2006). Foreign direct investment in tourism - flows and volumes, Tourism Management, 27, ] Hanson, G.H. 200). Sould Countries Promote Foreign Direct Investment? G-24 Discussion Paer Series, United Nations. ] Lisey, R.E. 2004). Home and Host Country Effects of FDI. In R.E. Baldwin and L.A. Winters, Eds., Callenges to Globalization, , University of Cicago Press, Cicago. 2] Markusen, J.R. 995). Te Boundaries of Multinational Enterrises and te Teory of International Trade, Journal of Economic Persectives, 9,

16 3] Navaretti, B.G. and A.J. Venables 2004). Multinational Firms in te World Economy. Princeton University Press, Princeton. 4] Rodriguez-Clare, A. 996). Multinationals, Linkages and Economic Develoment, American Economic Review, 86, ] UNCTAD 2007). FDI in Tourism: Te develoment dimension, New York and Geneva: United Nations. 5 Aendix Proof. Proof of Lemma ]Using Lagrange, te first order conditions read: { y β + n γ + i= xα i ) γ α ] β } β γ y β = λ 8) { y β + { y β + n γ + i= xα i n γ + i= xα i ) γ α ) γ α ] β } β γ n γ + i= xα i ] β } β γ n γ + i= xα i for i =,..., n were λ is te Lagrange multilier. ) γ ] β γ α n ) γ ] β γ α γ = λ 9) i= xα i ) γ α x α i = λ i, 0) Using te assumtion tat all local firms are symmetric we ave i = and ence obtain x i = x, for eac i =,..., n. From 9) and 0) we obtain x = n ) γ γ α ) wile from 9) and 8) we obtain = γ + n γ α x γ ) β γ γ y β wic, using ), reads as ) γ ] y = β +n γ α γ α γ β ) γ β 2) 6

17 Finally, we calculate substituting ) and 2) into te budget constraint 2) = R k n) 3) were k n) = +n γ α γ α ) γ γ ] β β +n γ α γ α ) γ ] β γ β γ γ +. 4) We can rewrite 4) as ) ] k n) = +n λα +λ β λ ) ] β +n λα, 5) wit ) k n) = λ α n λα + λ β λ β λ ) ] β +n λα. 6) Using Assumtion, wic imlies tat λ β > 0, < < 0 and tat λ α > 0, we ave tat k n) < 0, and tus n) is strictly increasing in n. Moreover we can write x = R ) ] n λα + n + λ β n λ β λ ) ] λ α β + n λ β wic, since Assumtion iv) imlies tat λ α λ β <, is strictly decreasing in n, and y = R + λ β λβ ) ] +n λα wic is decreasing in n. Proof. Proof of Lemma 2] Te result follows from system 4). An intuitive roof is as follows. Suose we are initially in te equilibrium c = c e, were n M = n L = n e. An increase in c leads to a reduction in n L wile n M remains unaffected. Hence, n M = n e >n L. A decrease in c leads to an increase in n L and terefore n M = n e <n L. 7

18 Proof. Proof of Proosition ] We establis te result by sowing tat s<0. Using 5), te rofit function π reads n π n) < 0, for eac π n) = n) 2 zn)2 +x n) c] n We rewrite te relationsi between x n) and n) in ) as follows ) x n) = n) n λα 7) wic allows us to write te multinational s rofits as π n) = n) z n)+n 2 λα ) ] c n Taking te first derivative wit resect to n we obtain n π n) = n) z ) ] ) n)+n λα + λ α n λα n) c Using n) = n) k n) kn), and substituting te exressions for k n) and k n), we obtain after rearranging terms z R n π n) = n) kn)] 2 λ α n λα λ β +n λα ) ] λ β + λ β ) λ β λ β λ ) ] β +n λα ) ] +n λα k n) c Te first term in curled brackets is negative and terefore a sufficient condition for nπn) < 0 to old is tat λ ) ) ] β n) k n) λ λα αλ β n λ β +n λα < c Substituting te exression for k n) and rearranging terms tis inequality reads ) n) λ α λ β n λα c ] λ β λ ) ] β +n λα < c 8

19 A sufficient condition for tis inequality to old is tat ) λ α λ β n) n λα < c wic, after using 7), simlifies to x n) < c λ α λ β Using Assumtion, wic imlies tat λ α λ β <, and te free entry condition 5) x n) =c + s, tis inequality reads s < c λ αλ β λ α λ β Tis last inequality is always satisfied for s<0 and terefore n π n) < 0 for all s<0 wic imlies tat s<0 can never be otimal. Proof. Proof of Lemma 3] Rewriting Πn) using te relationsi between x n) and n) as in 7) we obtain Πn) = n) z n)+n 2 λα ) ] c N Taking te first derivative wit resect to n we obtain n Πn) = n) z ) ] ) n)+n λα + λ α n λα n) wic, after using n) = n) k n) kn), can be written as { ) ] n) nπn) = kn) k n) z n)+n λα + λ α n λα ) k n) } Substituting te exression for k n) and k n) and rearranging terms we obtain Πn) n ) = n) λ kn)] 2 γ λ α n λα λ β +n λα z R λ β +n λα ) ] λ β λ β + λ β ) ] } +n λα k n) + λ ) ] β 8) 9

20 were te term in te first line, multilying te two terms in curled brackets, is ositive; te first term in curled brackets in te second line) is negative, decreasing in n, converges to zero in te ) limit of n 0 and to z R λ β + λ β for n ; te second term in curled brackets in te tird line) is ositive, decreasing in n, converges to for n 0 and, since lim n k n) = ) ) + λ β, it converges to λ β + λ β for n. Consequently, a unique maximum exists if z R ) λ β + λ β > λ β λ β + λ β ), i.e. for R z + λ β ) ) < λ β + λ β sufficiently large. Notice tat, since + λ β ), Assumtion guarantees tat a unique maximum exists. Moreover, te rofit function Πn) is locally concave at n. Since 2 z n Πn) < 0, z n < 0 and terefore z s > 0 Proof. Proof of Lemma 4] We sow tat, for c = c e and n = n M, Πn) n=n n < 0 and terefore M s > 0 follows. Using te relationsi between x n) and n) in 7), we can write rofits as Πn) = π n)+ n) n λα ) c N Taking te first derivative and evaluating te exression at n = n M and c = c e, and taking into account tat π n M) =0, we obtain ) n Πn) n=n M,c=c = n M ) λ α n M ) + λ e γ λ α n M) n M ) ] or ) n Πn) n=n M,c=c = n M ) λ α n M ) n M ) k n M) ] e k n M ) nm + λ α Using k n ) M and k n), and rearranging terms we obtain tat n Πn) n=n M,c=c < 0 if e λ β n M ) λ α ) ] λ β λ + n M ) ) ] β λ α < Since te first term in squared brackets is decreasing in n M and since n, a sufficient condition for 20

21 te inequality to old is tat λ ) β < wic, because of Assumtion 2, is always satisfied. By continuity, it follows tat s > 0 is otimal also for some values of c > c e. Proof. Proof of Proosition 2] For n <N, N Π < 0 and, using te enveloe teorem, c Π < 0. Consequently, for N sufficiently large Π <π n L). Let us next caracterize te tresold N. Notice tat we restrict te analysis to te case c < c e, were s > 0. For c < c e, c π n L) > 0, wile c Π < 0. As a consequence, N is decreasing in c. Using te enveloe teorem, we observe tat Π z < 0. For c < c e, z π n L) = 2 n L) 2, wile Π z = 2 n ) M 2. Since n) > 0, n M <n L and n M is increasing in z, Π z < z π n L), tat is rofits in te case of a domination strategy decrease at a lower ace tan rofits witout a domination strategy. As a consequence N is increasing in z. 2