Taxes and Entry Mode Decision in Multinationals: Export and FDI with and without Decentralization

Transcription

1 Taxes and Entry Mode Decision in Multinationals: Export and FDI wit and witout Decentralization Yosimasa Komoriya y Cuo University Søren Bo Nielsen z Copenagen Business Scool Pascalis Raimondos z Copenagen Business Scool June 1, 2015 Abstract We use a duopoly model to simultaneously consider bot foreign direct investment (FDI) and decentralization decisions, featuring transfer pricing. To investigate te relationsip between corporate taxes and te optimal entry mode for te foreign market, we examine and compare rms pro t in eac of tree entry modes: export, FDI wit decentralization, and FDI wit centralization. We nd tat, even wen a ost country does not ave any advantage, FDI can be undertaken because of a strategic incentive. We also nd tat trade liberalization promotes decentralization of multinationals, and centralization is dominated by decentralization in free trade. Keywords: decentralization; foreign direct investment; transfer pricing JEL Classi cation: F21, F23, H25 Very preliminary! Tis is one of our ongoing researc projects undertaken wile Komoriya is visiting te Department of Economics in te Copenagen Business Scool as a visiting scolar. Komoriya would like to tank te department for its ospitality. y Faculty of Economics, Cuo University, Hacioji, Tokyo , Japan; komoriya@tamacc.cuou.ac.jp z Department of Economics, Copenagen Business Scool, Frederiksberg 2000, Denmark; sbn.eco@cbs.dk (Nielsen), pra.eco@cbs.dk (Raimondos)

2 1 Introduction Transfer pricing is one of te pro t-sifting instruments for multinationals. In addition to te pro t-sifting role, transfer pricing may ave oter roles 1. In oligopoly settings, multinationals ave an incentive to manipulate teir intra- rm prices for a ecting competition tat teir subsidiaries are facing. Tis manipulation of intra- rm price is denoted te strategic role of transfer pricing. Te strategic role of transfer pricing is widely known after Scjelderup and Sørgard (1997). Te motivation of teir researc is as follows: In te IO-literature on MNE beaviour bot delegation of autority to local a liates and strategic interaction are discussed. [t]e focal point in te IO literature on MNEs, owever, is on foreign direct investments, not on pro t sifting troug transfer pricing (Scjelderup and Sørgard, 1997, pp.278) 2. Wit tis motivation, Scjelderup and Sørgard (1997) nd tis new role of transfer pricing. However, foreign direct investments (FDI) is taken as a given in teir analysis. Terefore, in tis paper, we consider bot of FDI decision and pro t sifting, focusing on te strategic motive of transfer pricing. Decentralization is te key of te strategic role of transfer pricing. In te model of Scjelderup and Sørgard (1997), in addition to FDI, decentralization is also taken as a given. Altoug Nielsen et al. (2008) examine decentralization decisions, tat is, tey investigate weter multinationals decentralize or not, FDI is taken as a given even in teir analysis. Terefore, in tis paper, we simultaneously consider bot FDI and decentralization decisions of multinationals, featuring transfer pricing. In te literature, Porter (2012) analyzes e ects of a corporate tax on coices between FDI and export. In a duopoly setting, taking rms coices between export and FDI into account, Porter (2012) sows te possibility tat an increase in te ost country s corporate tax can create its desirable situation were te more e cient rm undertakes FDI and te less e cient rm does not. Altoug, as far as we know, Porter (2012) is one of te rare papers tat focus on e ects of corporate tax on coices between export and FDI, se does not consider pro t sifting and transfer pricing. Amerigi and Peralta (2010) investigate coices of modes for serving two countries. A rm as two alternatives in teir model. One is to ave a plant in one country and to serve bot te local and foreign markets from te plant. Te oter alternative is to ave two plants and to serve markets from te local plants. Tey combine te proximity-consentration trade-o wit a tax-saving incentive. Altoug, as far as we know, Amerigi and Peralta (2010) is te rst paper tat sows te relationsip between te pro t sifting and FDI incentive, tey do not focus on strategic incentives for undertaking FDI because tey use a monopoly model. In our paper, we rst examine multinationals pro t in eac of tree entry modes into a foreign market: export, FDI wit decentralization, and FDI wit centralization (witout decentralization). Second, to investigate FDI incentives in decentralization and centralization cases and a 1 See for, example, Göx and Sciller (2007). 2 See, for example, Horstmann and Markusen (1992) and Motta (1992). 2

3 decentralization incentive, we compare after-tax pro t between entry modes. Finally, we sow te relationsip between corporate tax rates and te optimal entry mode for te foreign market by simultaneously considering FDI and decentralization decisions. We nd tat, even wen a ost country does not ave any advantage, tat is, even wen tere is no di erence in marginal production cost between te ome and ost countries, tere is no tax di erence, and trade costs are not incurred on export, FDI can be undertaken due to a strategic incentive. We also sow tat centralized multinationals do not appear in free trade at least in our current setting and tat trade liberalization promotes decentralization of multinationals. Section 2 introduces our model and sows pro ts and welfare in eac of te tree modes and Section 3 investigates FDI incentives in decentralization and centralization cases and a decentralization incentive. Section 4 sows te relationsip between corporate tax rates and te optimal entry mode for te foreign market. Section 5 discusses e ects of trade cost and Section 6 o ers conclusions. 2 Basic Model In tis section, we introduce our basic model. Tere are two countries (ome and foreign) and two rms ( rm 1 and rm 2). Only rm 1 serves te ome market and bot rms serve te foreign market. Te two rms compete in Cournot fasion. Te output levels of rm 1 in te ome and foreign markets are, respectively, q 1 and Q 1, and te output level of rm 2 in te foreign market is Q 2. We use linear demand functions, p = a bq in te ome market and P = A BQ in te foreign market, were q q 1 and Q Q 1 + Q 2. p and P are, respectively, te market prices in te ome and foreign markets. Marginal costs for production of te two rms are c 1 and c 2. We assume c 1 = c 2 = 0 for simpli cation. Firm 1 as tree entry modes for serving te foreign market: Export, FDI wit decentralization, and FDI wit centralization. Te following subsections sow eac of te tree modes. 2.1 Export Case In tis mode, rm 1 is an exporting rm and it competes wit te rival ( rm 2) in te foreign market. Governments in te ome and foreign countries tax locally earned pro t at t and t f, respectively. After-tax pro ts of rm 1 and rm 2 are, respectively, 1 = (1 t ) [(a bq 1 )q 1 + (A BQ 1 BQ 2 )Q 1 Q 1 ] ; 2 = (1 t f ) [(A BQ 1 BQ 2 )Q 2 ] : is a trade cost and we assume = 0 for te time being 3. We nd 3 We relax tis assumption in section 5. q 1 = a 2b ; Q 1 = A 3B ; Q 2 = A 3B 3

4 in te equilibrium. Te after-tax pro ts are a 2 1 = (1 t ) 2 = (1 t f ) A 2 : We de ne (1) as EX 1 in te following part. 2.2 FDI wit decentralization 4b + A2 ; (1) In tis mode, rm 1 is a decentralized MNE 4. Before-tax pro ts from production and sales on te part of rm 1 in te ome and foreign countries and rm 2 are, respectively, 1 = (a bq 1 )q 1 + mq 1 ; f1 = (A BQ 1 BQ 2 )Q 1 mq 1 ; f2 = (A BQ 1 BQ 2 )Q 2 ; were, m is a intra- rm price (transfer price). Te transfer price m can deviate from te marginal production cost (of zero). If it is set di erent from zero, te company is assumed to incur transfer pricing cost. After-tax pro ts of rm 1 and rm 2 are, respectively, 1 = (1 t ) 1 + (1 t f ) f1 u 2 m2 F; 2 = (1 t f ) f2 : Te tird and fourt terms of 1 are te transfer pricing cost and te FDI xed cost. parameter u governs te cange in transfer pricing cost, if te deviation from zero is enlarged. In tis case, te eadquarters of rm 1 sets te transfer price before its subsidiary and rm 2 coose teir output levels. We solve tis model using backward induction. In te second stage, we ave In te rst stage, d dm d1 dm q 1 = a 2b ; Q 1 = A 2m 3B ; Q 2 = A + m 3B : d 1 dm = 1 A (4t f 3t 1) 4m 2t f 3t Bu ; 4 = 2t f 3t Bu < 0: In equilibrium, te level of te transfer price is m = A (4t f 3t 1) 4 2t f 3t Bu: Given corporate tax rates, te absolute value of te transfer price jmj decreases in u. If tax rates are not vastly di erent (t f is not muc greater tan t ), m will lie below marginal cost of zero, in 4 In tis subsection, we assume t f t > 1 4 (1 t 3 ) Bu for obtaining non-negative output levels of bot 4 rms. If Bu > 3, we can obtain non-negative output levels. 4 Te 4

5 tat case, te transfer price increases in u. Generally, te more pronounced are transfer pricing costs, te closer m is to zero. If t f = t, m = A 4 + 9ABu 16 1 t Bu > A 4 : Altoug tis equation sows te existence of te strategic motive of transfer pricing, te strategic motive is muted because of te transfer pricing cost 5. And te transfer pricing costs are present, even toug tere is no tax rate di erence. Equilibrium outputs are and te after-tax pro ts are A [2 (1 t ) + 3Bu] Q 1 = 4B 2t f 3t Bu = A A [8 (t f t ) + 3Bu] 2B 8B 2t f 3t Bu; Q 2 = A (4t f 5t Bu) 4B 2t f 3t Bu = A 4B + A [8 (t f t ) + 3Bu] 16B 2t f 3t Bu 1 = (1 t ) a 2 4b + i A (1 2 t ) 2 + 2Bu (1 t f ) 8B 2t f 3t F; (2) 4Bu 2 = (1 t f ) A 2 (4t f 5t Bu) 2 16B 2t f 3t Bu 2 > 0: If t f = t, 1 = (1 t ) " a 2 4b + A2 (1 8B 1 # t + 2Bu) t Bu F: (3) We de ne (2) as DC 1 in te following part. 2.3 FDI wit centralization In tis mode, rm 1 is a centralized MNE 6. Before-tax pro ts of rm 1 in te ome and foreign countries and rm 2 and after-tax pro ts of te two rms are te same as tose in te decentralization case. Unlike in te previous case, te eadquarter of rm 1 cooses bot its transfer price and output level, and rm 2 cooses its output level simultaneously. 5 Wen m = A 4, Q 1 is equal to te output level of te Stackelberg leader. 6 2 In tis subsection, we assume t f t + Bu 1 tf > 0, for non-negative output levels of bot rms. 5

6 We 1 = (1 t ) (a 2bq 1 ) = (t f t ) Q 1 1 = (1 t f ) (A 2BQ 1 BQ 2 ) + m (t f t ) 2 = (1 t f ) (A BQ 1 2BQ 2 ) ; and From F.O.C., @ 2 = u < 0; = 2B (1 t f ) < 0; = 2B (1 t f ) < 0: q 1 = a 2b ; m = (t f t ) Q 1 ; u Q 1 = (1 t f ) (A BQ 2 ) + m (t f t ) ; 2B (1 t f ) Q 2 = A BQ 1 : 2B Tus, in te equilibrium, transfer price and outputs are, respectively, m = Q 1 = A (t f t ) (1 t f ) 2 (t f t ) 2 + 3Bu (1 t f ) ; Au (1 t f ) 2 (t f t ) 2 + 3Bu (1 t f ) = A 3B + 2A (t f t ) 2 i; 3B 2 (t f t ) 2 + 3Bu (1 t f ) i A (t f t ) 2 + Bu (1 t f ) Q 2 = i B 2 (t f t ) 2 + 3Bu (1 t f ) = A 3B 3B A (t f t ) 2 i: 2 (t f t ) 2 + 3Bu (1 t f ) As already mentioned in Nielsen (2014), te output of rm 1 is consist of two parts. If t f = t, because m = 0, te output of eac rm equals a Cournot output level. 6

7 Te after-tax pro ts are 1 = (1 t ) a 2 4b 2 = + i A 2 u (1 t f ) 2 (t f t ) 2 + 2Bu (1 t f ) i 2 F; (4) 2 2 (t f t ) 2 + 3Bu (1 t f ) i 2 (1 t f ) A 2 (t f t ) 2 + Bu (1 t f ) i 2 > 0; B 2 (t f t ) 2 + 3Bu (1 t f ) If t f = t, We de ne (4) as C 1 in te following part. a 2 1 = (1 t ) 4b + A2 F: (5) 3 Comparison of tree modes In tis section, we investigate FDI incentives in eac of te two FDI modes and decentralization incentives of multinationals. 3.1 Export and FDI wit decentralization In tis subsection, we compare EX 1 and DC 1 for investigating te FDI incentive for decentralization case. Te condition for FDI is DC 1 > EX 1, (1 t ) a2 A2 4b + i (1 t ) 2 + 2Bu (1 t f ) 8B 2t f 3t Bu F > (1 t ) a2 4b + (1, A2 [ (1 t ) (16t f 15t 1) 18Bu (t f t )] 72B 2t f 3t > F; 4Bu, [ (1 t ) (16t f 15t 1) 18Bu (t f t )] 72 2t f 3t Bu > F (A 2 =B) : t ) A2 (1 t If (1 t ) (16t f 15t 1) 18Bu (t f t ) < 0, tat is, t f t > ) 2 2[8(1 t )+u], rm 1 cooses Export. Tis means tat rm 1 cooses Export wen t f is su ciently ig relative to t. If t f t < (1 t ) 2 2[8(1 t )+u] = 1 t 16 1 t 1 t Bu not depends on te relative size of te xed cost, F A 2 =B < 1 t 16, weter rm 1 cooses FDI or. Wen F is low enoug, wen A is ig enoug, and/or wen B is low enoug, a rm 1 cooses to be a decentralized multinational rm 7. 7 Because we use P A BX for te demand function in te foreign market, A(A=B) governs pro t opportunities in tat market. 7

8 Oterwise, rm 1 cooses Export. If t f = t, from (1) and (3), DC 1 > EX 1 ", A2 8B, # (1 t + 2Bu) 8 1 t Bu > F 9 A 2 (1 t ) 72B 1 t + 9 > F: 4Bu Tis means tat rm 1 cooses FDI wen F is low enoug and in particular wen F = 0. Wen A = B = 1, F = 0, t f = 0:3, and u = 1, te condition for FDI ( DC 1 > EX 1 ) is 184t + 75t (60t 77) Below, Figures 1, panels (a) and (b), as been drawn using tese values for A, B, F, t f, and u. > 0: FDI premium Figure 1 (a). Additional pro t of FDI wit decentralization (t f = 0:3) 8

9 FDI premium Figure 1 (b). Additional pro t of FDI wit decentralization (t f = 0:3) (enlarged view) In Figure 1, rm 1 cooses FDI wen t > et ' 0: Proposition 1 Even in a case witout te ost country s production cost and tax advantages and trade cost, FDI may take place. Even wen te tax rate in te ost country is iger tan tat in te ome country, FDI can be superior to export because of a (purely) strategic incentive, if te tax di erence is small. Wen t < t f, tere is no FDI incentive coming from tax-saving incentive. However, in te above gures, rm 1 cooses FDI if te tax di erence is small. Altoug Amerigi and Pelarta (2010) sows a tax-saving FDI incentive, tey do not consider tis strategic incentive for FDI. 3.2 Export and FDI wit centralization In tis subsection, we compare EX 1 and C 1 to investigate te FDI incentive for te centralization case. Te condition for FDI is C 1 > EX 1 i, (1 t ) a2 A2 2 u (1 t f ) (t f t ) 2 + 2Bu (1 t f ) 4b + i 2 F > (1 t ) a2 2 2 (t f t ) 2 4b + (1 + 3Bu (1 t f ), n A 2 (t f t ) t ) A2 o 8 (1 t ) (t f t ) 3 + 3Bu (1 t f ) [(3t f 8t + 5) (t f t ) 6Bu (1 t f )] i 2 > F: 18B 2 (t f t ) 2 3Bu (1 t f ) 9

10 If t f = t and F > 0, from (1) and (5), rm 1 does not prefer FDI wit centralization to Export because C 1 = EX 1 F. Wen A = B = 1, F = 0, t f = 0:3, and u = 1, te condition for FDI ( C 1 (10t 3) 15054t 7440t t t ( 15t + 25t 2 24) 2 > 0: Figure 2, panels (a) and (b), is drawn using tese values. > EX 1 ) is FDI premium Figure 2 (a). Additional pro t of FDI wit centralization (t f = 0:3) FDI premium Figure 2 (b). Additional pro t of FDI wit centralization (t f = 0:3) (enlarged view) 10

11 In Figure 2, wen t f = t, C 1 = DC 1 because of F = 0. And FDI wit centralization is better tan Export only in te case were te tax rate in te ome country is iger tan tat in te foreign country (t f < t ). 3.3 FDI wit decentralization and FDI wit centralization In tis subsection, we compare DC 1 and C 1 for investigating te decentralization incentive 8. Te condition for decentralization is DC 1 > C 1 i, (1 t ) a2 A2 (1 t ) 2 + 2Bu (1 t f ) 4b + 8B 2t f 3t Bu F i > (1 t ) a2 A2 2 u (1 t f ) (t f t ) 2 + 2Bu (1 t f ) 4b + i 2 F 2 2 (t f t ) 2 + 3Bu (1 t f ), A2 2t t 2 2Bu + 2But f 1 i A2 2 u (1 t f ) (t f t ) 2 + 2Bu (1 t f ) 8B 2t f 3t Bu > i (t f t ) 2 + 3Bu (1 t f ) i A 4 2 (1 t ) 2 (t f t ) 4 Bu (1 t f ), i 2 > 0; 8B 2t f 3t Bu 2 (t f t ) 2 + 3Bu (1 t f ) were 4 (1 t ) ( t f t + 2) (t f t ) 2 Bu (1 t f ) 8t f + 6t + t 2 f 6t 2 + 6t f t + 1. As Nielsen (2008) sowed, decentralization is preferable to centralization wen te two countries ave te same tax level. If t f = t, from (3) and (5), DC 1 C 1 = A2 8B = " (1 t + 2Bu) 8 1 t Bu 9 A 2 (1 t ) 72B 1 t + 9 > 0: 4Bu Tus, FDI wit decentralization is preferred to FDI wit centralization wen te two countries ave te same tax level. is Wen A = B = 1, F = 0, t f = 0:3, and u = 1, te condition for decentralization ( DC 1 > C 1 ) t 94316t t t t t (60t 77) ( 15t + 25t 2 24) 2 > 0: Figure 3, panels (a) and (b), illustrate te additional pro t from decentralization, using te parameter values above. 8 Tis subsection is based on Nielsen et al. (2008). # 11

12 Dec. premium Figure 3 (a). Additional pro t of decentralization (t f = 0:3) Dec. premium Figure 3 (b). Additional pro t of decentralization (t f = 0:3) (enlarged view) In Figure 3, rm 1 cooses FDI wit decentralization wen t > t ' 0:

13 4 Optimal entry mode for te foreign market We compare EX 1 = (1 t ) a 2 4b + (1 t ) A 2 ; DC 1 = (1 t ) a 2 A (1 2 t ) 2 + 2Bu (1 t f ) + 4b 8B 2t f 3t F; 4Bu C 1 = (1 t ) a 2 4b i A 2 u (1 t f ) 2 (t f t ) 2 + 2Bu (1 t f ) + i (t f t ) 2 + 3Bu (1 t f ) i F wen a = 2, b = 1, A = B = 1, F = 0, t f = 0:3, and u = 1 in tis section. We ave EX 1 = 10 (1 t ) ; 9 (6) 1 = 284t 125t ; 2 (60t 77) (7) DC C 1 = 21372t t t t t ( 15t + 25t 2 24) 2 : Figure 4 sows tese pro ts ( EX 1 is black tin straigt line, DC 1 is red tin line, and C 1 is blue bold line). Profit Figure 4 (a). Comparison of pro t in te tree modes 13

14 Profit Figure 4 (b). Comparison of pro t in te tree modes (enlarged view) Firm 1 prefers decentralized FDI to centralized FDI wen t > t ' 0:20860, prefers decentralized FDI to Export wen t > et ' 0:28254, and prefers centralized FDI to Export wen t > bt ' 0:3. Te order of te tree pro ts are EX 1 > C 1 > DC 1 for t < 0:20860, EX 1 > DC 1 > C 1 for 0:20860 < t < 0:28254, DC 1 > EX 1 > C 1 for 0:28254 < t < 0:3, and DC 1 > C 1 > EX 1 for 0:3 < t. Tus, we ave te following proposition. Proposition 2 Given te parameter values, in te case of no ost country s advantage and no trade cost, wen te rm cooses FDI, te rm also implements decentralization. Tis proposition implies no centralized multinationals appear in free trade. Because te FDI xed cost decreases DC 1 and C 1 (sifts te red tin line and blue bold line downward) and 14

15 it does not a ect te decentralization incentive, tis proposition would remain valid even wen F 6= 0. 5 Trade cost In tis section, we relax te assumption, = 0: For cecking and enforcing robustness, we investigate ow and to wat extent trade costs a ect te result in te previous section 9. Wen 6= 0, te after-tax pro t of rm 1 is Wit our parameters, Wen = 0:03, EX 1 = (1 t ) a 2 4b EX + (1 t ) A 2 : 1 = 2 (1 t ) : EX = (1 t ) : (8) Te following gure sows (6), (7), and (8) ( EX 1 is te black tin straigt line, DC 1 is te red tin line, and C 1 is te blue bold line). Profit Figure 5 (a). Positive trade cost case ( = 0:03) 9 In tis section, we consider a case were te trade cost is speci c. We will sow an ad valorem case in our future researc. We will also consider te case were rm 1 incurs trade costs even in FDI cases. 15

16 Profit Figure 5 (b). Positive trade cost case ( = 0:03) (enlarged view) Altoug, in tis case, te trade cost does not a ect te decentralization decision, te cost a ects FDI decisions. Firm 1 prefers decentralized FDI to Export wen t > et ' 0:20074 and prefers centralized FDI to Export wen t > bt ' 0: Te order of te tree pro ts are EX 1 > C 1 > DC 1 for t < 0:19782, C 1 > EX 1 > DC 1 for 0:19782 < t < 0:20074, C 1 > DC 1 > EX 1 for 0:20074 < t < 0:20860, and DC 1 > C 1 > EX 1 for 0:20860 < t. Tus, as te corporate tax in te ost country increases, rm 1 s entry mode to te foreign market sifts from Export to FDI wit centralization, to FDI wit decentralization. Tis implies FDI wit centralization is feasible wen te transport cost is su ciently ig. By comparing te results in te previous and current sections, we can nd tat trade liberalization accelerates decentralization In our future researc, we will discuss a relationsip between trade costs and decision ligts wit anecdotal 16

17 6 Conclusion We investigated FDI incentives, taking into account te strategic role of transfer pricing and decentralization. It as been found tat, even wen a ost country as neiter production cost advantage nor tax advantage, FDI can be undertaken by a (purely) strategic incentive. We also found tat centralized multinationals do not appear in free trade at least in our current setting and tat trade liberalization promotes decentralization of multinationals. For cecking and enforcing robustness, we sould examine ow and to wat extent an ad valorem trade cost a ect our ndings and also consider te case were rm 1 incurs trade costs even in FDI cases. We leave tose investigations to future work. In te model, we assume a quadratic form of transfer pricing cost. Tis cost implicitly re ects transfer pricing regulation implemented by governments. Investigating more speci cally te e ects of transfer pricing regulation on FDI incentives is also left to future work. References [1] Amerigi, Oscar, Susana Peralta, 2010, Te Proximity-concentration Trade-o wit Pro t Sifting?, Journal of Urban Economics 68, [2] Göx, Robert F., Ulf Sciller, 2007, An Economics Perspective on Transfer Pricing, In: Capman, Cristoper S., Antony G. Hopwood, and Micael D. Sields (Eds), Handbook of Management Accounting Researc, , Elsevier. [3] Horstmann, Ignatius J. and James R. Markusen, 1992, Endogenous Market Structures in International Trade (natura facit saltum), Journal of International Economics 32, [4] Motta, Massimo, 1992, Multinational Firms and te Tari -jumping Argument, European Economics Review 36, [5] Nielsen, Søren Bo, 2014, Transfer Pricing: Roles and Regimes, CESifo Working Paper Series No [6] Nielsen, Søren Bo, Pascalis Raimondos-Møller and Guttorm Scjelderup, 2008, Taxes and Decision Rigts in Multinationals, Journal of Public Economic Teory 10, [7] Porter, Lynda A, 2012, Asymmetric Oligopoly and Foreign Direct Investment: Implications for Host-Country Tax-Setiting, International Economic Journal 26, [8] Scejelderup, Guttorms and Lars Sørgard, 1997, Transfer Pricing as a Strategic Device for Decentralized Multinationals, International Tax and Public Finance 4, evidence. 17