1 UNITED NATIONS CONFERENCE ON TRADE AND DEVELOPMENT UNITED NATIONS CENTER FOR INTERNATIONAL DEVELOPMENT HARVARD UNIVERSITY G-24 Dscusson Paper Seres Should Countres Promote Foregn Drect Investment? Gordon H. Hanson No. 9, February 2001
2 UNITED NATIONS CONFERENCE ON TRADE AND DEVELOPMENT CENTER FOR INTERNATIONAL DEVELOPMENT HARVARD UNIVERSITY G-24 Dscusson Paper Seres Research papers for the Intergovernmental Group of Twenty-Four on Internatonal Monetary Affars UNITED NATIONS New York and Geneva, February 2001
3 Note Symbols of Unted Natons documents are composed of captal letters combned wth fgures. Menton of such a symbol ndcates a reference to a Unted Natons document. * * * The vews expressed n ths Seres are those of the authors and do not necessarly reflect the vews of the UNCTAD secretarat. The desgnatons employed and the presentaton of the materal do not mply the expresson of any opnon whatsoever on the part of the Secretarat of the Unted Natons concernng the legal status of any country, terrtory, cty or area, or of ts authortes, or concernng the delmtaton of ts fronters or boundares. * * * Materal n ths publcaton may be freely quoted; acknowledgement, however, s requested (ncludng reference to the document number). It would be apprecated f a copy of the publcaton contanng the quotaton were sent to the Edtoral Assstant, Macroeconomc and Development Polces Branch, Dvson on Globalzaton and Development Strateges, UNCTAD, Palas des Natons, CH-1211 Geneva 10. UNCTAD/GDS/MDPB/G24/9 UNITED NATIONS PUBLICATION Copyrght Unted Natons, 2001 All rghts reserved
4 Should Countres Promote Foregn Drect Investment? PREFACE The G-24 Dscusson Paper Seres s a collecton of research papers prepared under the UNCTAD Project of Techncal Support to the Intergovernmental Group of Twenty-Four on Internatonal Monetary Affars (G-24). The G-24 was establshed n 1971 wth a vew to ncreasng the analytcal capacty and the negotatng strength of the developng countres n dscussons and negotatons n the nternatonal fnancal nsttutons. The G-24 s the only formal developng-country groupng wthn the IMF and the World Bank. Its meetngs are open to all developng countres. The G-24 Project, whch s admnstered by UNCTAD s Macroeconomc and Development Polces Branch, ams at enhancng the understandng of polcy makers n developng countres of the complex ssues n the nternatonal monetary and fnancal system, and at rasng awareness outsde developng countres of the need to ntroduce a development dmenson nto the dscusson of nternatonal fnancal and nsttutonal reform. The research carred out under the project s coordnated by Professor Dan Rodrk, John F. Kennedy School of Government, Harvard Unversty. The research papers are dscussed among experts and polcy makers at the meetngs of the G-24 Techncal Group, and provde nputs to the meetngs of the G-24 Mnsters and Deputes n ther preparatons for negotatons and dscussons n the framework of the IMF s Internatonal Monetary and Fnancal Commttee (formerly Interm Commttee) and the Jont IMF/IBRD Development Commttee, as well as n other forums. Prevously, the research papers for the G-24 were publshed by UNCTAD n the collecton Internatonal Monetary and Fnancal Issues for the 1990s. Between 1992 and 1999 more than 80 papers were publshed n 11 volumes of ths collecton, coverng a wde range of monetary and fnancal ssues of major nterest to developng countres. Snce the begnnng of 2000 the studes are publshed jontly by UNCTAD and the Center for Internatonal Development at Harvard Unversty n the G-24 Dscusson Paper Seres. The Project of Techncal Support to the G-24 receves generous fnancal support from the Internatonal Development Research Centre of Canada and the Governments of Denmark and the Netherlands, as well as contrbutons from the countres partcpatng n the meetngs of the G-24.
5 SHOULD COUNTRIES PROMOTE FOREIGN DIRECT INVESTMENT? Gordon H. Hanson Department of Economcs and School of Busness Admnstraton Unversty of Mchgan, Ann Arbor, USA G-24 Dscusson Paper No. 9 February 2001
6 Should Countres Promote Foregn Drect Investment? v Abstract Ths paper examnes whether polces to promote foregn drect nvestment (FDI) make economc sense. The dscusson focuses on whether exstng academc research suggests that the benefts of FDI are suffcent to justfy the knd of polcy nterventons seen n practce. For small open economes, effcent taxaton of foregn and domestc captal depends on ther relatve moblty. If foregn and domestc captal are equally moble nternatonally, t wll be optmal for countres to subject both types of captal to equal tax treatment. If foregn captal s more moble nternatonally, t wll be optmal to have lower taxes on captal owned by foregn resdents than on captal owned by domestc resdents. Absent market falure, there s no justfcaton for favourng FDI over foregn portfolo nvestment. In practce, countres appear to tax ncome from foregn captal at rates lower than those for domestc captal and to subject dfferent forms of foregn nvestment to very dfferent tax treatment. FDI appears to be senstve to host-country characterstcs. Hgher taxes deter foregn nvestment, whle a more educated work force and larger goods markets attract FDI. There s also some evdence that multnatonals tend to agglomerate n a manner consstent wth locaton-specfc externaltes. There s weak evdence that FDI generates postve spllovers for host economes. Whle multnatonals are attracted to hgh-productvty countres, and to hgh-productvty ndustres wthn these countres, there s lttle evdence at the frm or plant level that FDI rases the productvty of domestc enterprses. Indeed, t appears that plants n ndustres wth a larger multnatonal presence tend to enjoy lower rates of productvty growth over tme. Emprcal research thus provdes lttle support for the dea that promotng FDI s warranted on welfare grounds. Subsdes to FDI are more lkely to be warranted where multnatonals are ntensve n the use of elastcally suppled factors, where the arrval of multnatonals to a market does not lower the market share of domestc frms, and where FDI generates strong postve productvty spllovers for domestc agents. Emprcal research suggests that the frst and thrd condtons are unlkely to hold. In the three cases we examne, t appears that the second condton holds, but not the frst or thrd condtons. Ths suggests that Brazl s subsdes to foregn automoble manufacturers may have lowered natonal welfare. Costa Rca appears to have been prudent n not offerng subsdes n the case of Intel. There clearly s a need for much more research on the host-economy consequences of FDI. The mpresson from exstng academc lterature s that countres should be sceptcal about clams that promotng FDI wll rase natonal welfare. A sensble approach for polcy makes n host countres s to presume that subsdzng FDI s unwarranted, unless clear evdence s presented to support the argument that the socal returns to FDI exceed the prvate returns.
7 Should Countres Promote Foregn Drect Investment? x Table of contents Page Preface Abstract v I. Introducton... 1 Settng the stage: foregn versus domestc nvestment... 2 II. Promoton of FDI n practce... 3 III. FDI and host-country economc performance... 9 A. What explans multnatonal producton? B. What determnes the locaton of multnatonal producton? C. Does FDI generate postve spllovers for the host economy? IV. Evaluaton of FDI n practce A. A theoretcal model B. The promoton of FDI n practce: three cases C. Evaluaton of FDI promoton cases V. Concludng remarks Notes References Appendx: Dervaton of the welfare effects for the host economy... 28
8 Should Countres Promote Foregn Drect Investment? 1 SHOULD COUNTRIES PROMOTE FOREIGN DIRECT INVESTMENT? Gordon H. Hanson * I. Introducton There s a presumpton among many academcs and polcy makers that foregn drect nvestment (FDI) s somehow specal. 1 One common vew s that FDI helps accelerate the process of economc development n host countres. Optmsm about the economc consequences of foregn nvestment, coupled wth heghtened awareness about the mportance of new technologes for economc growth, has contrbuted to wde-reachng changes n natonal polces on FDI. Durng the last two decades, many emergng economes have dramatcally reduced barrers to FDI, and countres at all levels of development have created a polcy nfrastructure to attract multnatonal frms. 2 Standard tactcs to promote FDI nclude the extenson of tax holdays, exemptons from mport dutes, and the offer of drect subsdes. Snce 1998, 103 countres have offered specal tax concessons to foregn corporatons that have set up producton or admnstratve facltes wthn ther borders (Av- Yonah, 1999). Typcally, these concessons are appled to multnatonal enterprses but not to local frms n the same lnes of actvty. In ths paper, we examne whether polces to promote FDI make economc sense. Whle elmnatng barrers to foregn nvestment s a means of achevng global market ntegraton, promotng FDI goes one step further by favourng one form of ntegraton expanded foregn control of productve assets over others, such as ncreased trade n goods, more nternatonal lcensng of technology, or larger cross-border flows of portfolo captal. Assessng the consequences of promotng FDI for natonal welfare s a bg task and one we n no way pretend to complete n full. We focus on whether exstng academc research suggests that the benefts of FDI are suffcent to justfy the knd of polcy nterventons seen n practce. Ths wll help to dentfy a set of practcal gudelnes for when and where promotng FDI mght be welfare-enhancng. In the remander of the ntroducton, we frame the dscusson by outlnng the condtons under whch economc theory suggests that government polces favourng foregn over domestc captal are justfed. In secton II, we brefly revew the types of polcy ncentves that the Group of 24 (G-24) and other countres offer to multnatonal frms; 3 ths wll help to establsh notons of standard practce. In secton III, we survey the theoretcal and emprcal lterature on FDI, wth emphass on research whch examnes whether FDI s a source of postve exter- * I should lke to thank James Hnes, Mustafa Mohaterem and Dan Rodrk for helpful comments and dscussons, as well as Poh Boon Ung for provdng excellent research assstance.
9 2 G-24 Dscusson Paper Seres, No. 9 naltes for host countres. In secton IV, we develop a smple theoretcal model of FDI, whch we then use to evaluate three cases n whch developng-country governments have offered specal ncentves to multnatonal frms as n the cases of Ford and General Motors (GM) n Brazl and Intel n Costa Rca. The purpose of these case studes s to see whether economc theory and relevant emprcal lterature would suggest that polcy nterventon n favour of FDI was justfed. Fnally, n secton V we offer concludng remarks on factors to consder when tryng to determne whether promotng FDI wll rase host-country welfare. Settng the stage: foregn versus domestc nvestment To begn, t s helpful to specfy what we mean by promotng FDI. The benchmark one adopts depends on whether t s optmal for countres to subject foregn and domestc captal to equal tax treatment. Countres may have reason to tax domestc and foregn captal dfferently. For a small open economy facng an mmoble supply of labour and an nternatonally moble supply of captal, the optmal factor ncome tax falls entrely on labour (Gordon, 1986; Razn and Sadka, 1991). Not taxng captal s sensble because the mmoble factor bears the ncdence of any tax on factor ncomes, makng t more effcent to tax the mmoble factor drectly. 4 By extenson, f foregn captal s perfectly moble but domestc captal s not, then the optmal tax on factor ncomes falls on domestc labour and captal but not on foregn captal. More generally, f foregn captal s elastc n ts supply relatve to domestc captal, then t s optmal for countres to tax ncome from domestc captal at hgher rates (where the optmal tax rate on ncome from foregn captal may be postve f ts supply elastcty s less than nfnte). 5 If one presumes that domestc and foregn captal are equally elastc n supply (for example, f they are equally moble nternatonally), then promotng FDI means any polcy whch subjects FDI to favourable tax treatment relatve to domestc nvestment and foregn portfolo nvestment (whether n the form of debt or equty). If, on the other hand, one presumes that foregn captal s more elastc n supply than domestc captal, then promotng FDI means any polcy whch favours drect nvestment nflows over portfolo nvestment nflows, holdng constant a country s relatve tax treatment of domestc nvestment ncome and foregn nvestment ncome. For FDI to mert specal treatment, there needs to be market falure that s specfc to producton by multnatonal frms. 6 Asymmetrc nformaton between domestc and foregn nvestors s one commonly mentoned source of market falure, though one that s not specfc to FDI. If domestc nvestors are better nformed about domestc nvestment opportuntes than foregn nvestors, then, all else equal, a captal-mportng country would rase welfare by subsdzng foregn captal nflows (Gordon and Bovenberg, 1996). Could asymmetrc nformaton justfy favourng FDI over foregn portfolo nvestment? Razn et al. (1998) suggest the answer s no. Snce FDI, but not portfolo nvestment, gves a foregn nvestor a controllng nterest n a domestc frm, multnatonal frms are lkely to be at an nformatonal advantage relatve to foregn portfolo nvestors (though not relatve to domestc ones). In ths case, the optmal tax polcy s to subsdze foregn portfolo nvestment and to leave FDI untaxed. 7 There are other sources of market falure whch could justfy specal treatment of FDI (Caves, 1995). A much cted possblty s that FDI generates productvty spllovers for the host economy (Blomstrom and Kokko, 1998). One dea s that multnatonal enterprses possess superor producton technology and management technques, some of whch are captured by local frms when multnatonals locate n a partcular economy. 8 A related source of spllovers s forward and backward lnkages between multnatonal and host-economy frms (Rodrguez-Clare, 1996), whch may result from multnatonals provdng nputs at lower cost to local downstream buyers or by ther ncreasng demand for nputs produced by local upstream supplers. A further possblty s that FDI shfts rents earned by multnatonals to the host economy (Glass and Sagg, 1999; Janeba, 1996). Multnatonals may have global market power and may share monopoly rents wth managers and workers n ther varous operatonal unts. By attractng multnatonal frms, the host economy may capture a porton of the rents that these frms generate. 9 Whle these and other types of market falure are plausble, each s also subject to controversy. Spllovers assocated wth FDI are supported by casual evdence from many countres, but ther exstence and magntude are, as we shall see, dffcult to establsh emprcally. Indeed, mcro evdence from large samples of manufacturng plants n developng countres fals to support the exstence of postve productvty spllovers related to FDI. There s also reason to beleve that multnatonal enterprses tend
10 Should Countres Promote Foregn Drect Investment? 3 to have market power n ther respectve ndustres. Whether or not they share rents wth employees n ther foregn afflates s an emprcal queston. Attractng FDI may shft a porton of the rents that multnatonals earn to the host economy, but t may also reduce the profts of local frms that compete wth multnatonals at home or abroad. 10 Arguments for promotng FDI are based on clams concernng the economc envronment, whch can and should be subject to emprcal verfcaton. Before decdng to promote FDI, t s essental to evaluate possble sources of market falure assocated wth multnatonal frms. It s ths task to whch we devote much of the paper. II. Promoton of FDI n practce In ths secton, we summarze current government polces to promote FDI n G-24 and other countres. We begn wth a bref revew of corporate taxaton at the natonal level and then dscuss the range of tax and other ncentves whch countres offer to multnatonal enterprses. The source for all data, except where noted, are annual edtons of Corporate Taxes: A Worldwde Summary by Prce Waterhouse. Polces to promote FDI take a varety of forms. The most common are partal or complete exemptons from corporate taxes and mport dutes. These polces are typcally the result of formal legslaton or presdental decree, whch apply to all foregn corporatons that meet certan restrctons. These restrctons vary consderably across countres. In many cases they requre multnatonals to establsh producton facltes n the host country n specfed lnes of actvtes or desgnated regons, such as export-processng zones (EPZs), and to export output embodyng nputs mported duty-free. Drect subsdes and other types of concessons are often negotated between multnatonal frms and host governments on a case-by-case bass. Such ndvdualzed subsdes appear to be common, but are hard to document systematcally. Table 1 shows corporate tax rates n 1990 and 1998 for each G-24 country for whch data are avalable and averaged across regons for other selected countres. 11 Snce some countres have progressve tax rates (lower rates for smaller corporatons) or rates whch vary across sectors, we report the mnmum and maxmum tax rates whch apply to corporate ncome. In 1998 the maxmum tax rates on corporate ncome n the G-24 countres ranged from a hgh of 57 per cent n Iran to a low of 25 per cent n Brazl. Several countres ncludng Argentna, Columba, Guatemala, Peru, the Phlppnes and Sr Lanka tax corporate ncome at a flat rate, whle others ncludng Ghana, Iran, Mexco, and Trndad and Tobago tax ncome earned by small corporatons at rates much lower than for large corporatons. Between 1990 and 1998 most countres reduced ther maxmum corporate ncome tax rates, wth hgh-tax countres undertakng the largest cuts n absolute terms. Tax rates n ndvdual G-24 countres are roughly comparable to the averages for other countres n ther respectve regons. A few outlers are apparent. On a regon-by-regon bass, tax rates n 1998 were relatvely hgh n the Republc of the Congo, Inda, Iran and Pakstan. Tax rates n North Amerca, Oceana and Western Europe are on average smlar to those n Latn Amerca, and lower than those n Afrca and Asa. Tables 2 5 gve a bref descrpton of how G-24 and a sample of fve other countres treat foregn corporatons that operate wthn ther borders. 12 Most countres for whch data are avalable grant corporate ncome tax exemptons to foregn corporatons makng nward drect nvestments. Typcally, these exemptons last for less than a decade from the ntaton of a new project, though n some cases they are long-lved. Most countres also offer exemptons to foregn corporatons on mport dutes, where these tend to be restrcted to nputs that are used to produce exports or, n a few cases, captal goods. Exemptons from value-added taxes are a somewhat less common tax concesson that countres grant multnatonal frms. Smlar tax concessons are also avalable to domestc frms n some countres, though these concessons are for the most part ted to partcpaton n EPZs, export actvtes outsde of such zones, or producton n offcally desgnated prorty sectors or regons. Comparng 1990 wth 1998, there s a slght ncrease n the number of countres offerng exemptons from valued-added taxes and mport dutes and supportng EPZs. 13 Not ncluded n the tables are detals on drect subsdes whch host governments offer to multnatonal frms on a case-by-case bass. These arrangements are frequently unpublczed, but the practce appears to be relatvely common. Brazl s one country whch actvely pursues multnatonal frms and has offered generous subsdes n a number
11 4 G-24 Dscusson Paper Seres, No. 9 Table 1 CORPORATE TAX RATES FOR G-24 AND OTHER COUNTRIES, 1990 AND Mn. rate Max. rate Mn. rate Max. rate Group of 24 Congo Egypt Gabon Ghana Ngera Inda Iran Pakstan Phlppnes Sr Lanka Argentna Brazl Colomba Guatemala Mexco Peru Trndad & Tobago Venezuela Other countres (regonal averages) Afrca East Asa Eastern Europe Latn Amerca Mddle East North Amerca Oceana South East Asa Western Europe Source: Prce Waterhouse (1990). Note: Ths table shows mnmum and maxmum corporate ncome tax rates for selected countres. See text for detals. Data s more detaled for some countres than others. Approxmatons are made n certan cases. of nstances (see the GM and Ford examples n secton IV). For nstance, the country gves generous tax ncentves to frms that locate manufacturng facltes n the Amazon regon. Unspecfed government subsdes appeared to be mportant n lurng Multbras (a US-owned frm) to construct a $400 mllon plant to manufacture ar condtoners and mcrowave ovens n Manaus n Investment subsdes also appeared to be mportant n convncng Honda to buld a motorcycle plant n the area. In the absence of tax breaks, there appears to be lttle reason why multnatonals would locate n the regon. Poorer countres n Europe have also been aggressve n pursung multnatonal frms. To gve a few examples: n 1991 Portugal provded a lumpsum subsdy and promsed tax breaks on future
12 Table 2 TAX CONCESSIONS FOR INWARD FDI IN G-24 COUNTRIES, 1990 (I) (II) (III) (IV) Avalable to domestc corporatons? Corporate ncome Value-added Import duty EPZ Country tax exempton Perod Sectors tax exempton Items exempton Items provson (I) (II) (III) (IV) Egypt X 5 20 years a All b X Gabon X 0 10 years All X c Ngera X 0 5 years A, M, P X d E, M X e E, M, R X X X Congo X 5 15 years P Argentna X E, R X Brazl X E, P X E, P X f Guatemala X g P X h P X X j Mexco (tax credts) X E, P Peru X A, P * P X X Venezuela X A Inda X 5 years All k X E X X X Pakstan X 2 8 years K, A, P * l E, M X E X X X X X Phlppnes X 3 6 years All m X E, K, P, R X X X X Source: Prce Waterhouse (1990). Note: * = taxed at lower rate. A = agrculture. E = exported goods/exportng. K = captal. M = manufacturng. P = prorty companes/ndustres. R = raw materals. X = country offered concesson n ndcated year. Data s more detaled for some countres than others. Approxmatons are made n certan cases. a Maxmum of 20 years f t s a project for medum-szed and economcal housng, whose whole unts are leased vacant for dwellng. b In July 1989, a new nvestment law was ssued and t offered a projects profts to be exempt from tax on ndustral and commercal profts and from corporate tax. c More restrctons apply on domestc corporatons than on foregn corporatons. d Excse duty pad on export manufactures s refundable. e Refund of mport duty. f Excse and sales and servce tax exemptons are granted to exporters of manufactured goods. g The government may grant exemptons from dutes and taxes f the enterprse s classfed as ether basc, necessary or useful, or f t s to be located outsde the muncpalty of Guatemala. h See 7. Exemptons from ncome taxes and mport dutes (up to 100 per cent for a maxmum 10-year perod) may also be granted to ndustres orgnally located (or, f exstng, transferred) outsde of the Department of Guatemala, ste of the captal cty. j See 9. k New ndustral undertakng n free trade zones, or a 100 per cent export-orented undertakng s entrely exempt from ncome tax, subject to certan condtons. l Companes exportng goods manufactured n Pakstan can clam a rebate of 50 per cent on the tax attrbutable to such export sales. However, n respect of certan specfed goods, the tax rebate s avalable at 25 per cent or 75 per cent. m Income tax holday gvng full exempton from corporate ncome tax for sx years for poneer frms and four years for non-poneer frms from date of commercal operaton; expandng frms are gven three years. Should Countres Promote Foregn Drect Investment? 5
13 Table 3 TAX CONCESSIONS FOR INWARD FDI IN COMPARISON COUNTRIES, 1990 (I) (II) (III) (IV) Avalable to domestc corporatons? Corporate ncome Value-added Import duty EPZ Country tax exempton Perod Sectors tax exempton Items exempton Items provson (I) (II) (III) (IV) Chle X a P X b E X Ireland X All c Japan South Afrca Thaland X 3 8 years P X d E, R, K X X X Source: Prce Waterhouse (1990). Note: * = taxed at lower rate. A = agrculture. E = exported goods/exportng. K = captal. M = manufacturng. P = prorty companes/ndustres. R = raw materals. X = country offered concesson n ndcated year. Data s more detaled for some countres than others. Approxmatons are made n certan cases. a Tax benefts and other ncentves for companes operatng n northernmost and southernmost parts of the country. Tax benefts to forestry companes also. b Rembursement of taxes pad. c Companes that commenced operaton wthn the Shannon Free Arport before 1 January 1981 can obtan full exempton from corporaton tax untl 5 Aprl 1990 f the ncome s derved from the carryng on of certan actvtes, ncludng exportng goods and a wde range of servces. d Ether exempton or reducton. 6 G-24 Dscusson Paper Seres, No. 9
14 Table 4 TAX CONCESSIONS FOR INWARD FDI IN G-24 COUNTRIES, 1998 (I) (II) (III) (IV) Avalable to domestc corporatons? Corporate ncome Value-added Import duty EPZ Country tax exempton Perod Sectors tax exempton Items exempton Items provson (I) (II) (III) (IV) Côte d Ivore X 5 8 years All X All Egypt X 5 20 years All a X Gabon X 0 10 years All X b Ngera X 0 5 years P, E, A, M X All c X E, R X X X X X Congo X d 0 10 years A, P, M Argentna (tax credt bonds) X E X E, R X X Brazl X E, P X E, P X e Guatemala f X E X K, E, R X X X X Mexco (tax credts) X E Peru X All g X All h X X Inda X 5 years All X E X X X Phlppnes X 3 6 years All j X All k X All X X X X X Sr Lanka X P X P X X Source: Prce Waterhouse (1998). Note: * = taxed at lower rate. A = agrculture. E = exported goods/exportng. K = captal. M = manufacturng. P = prorty companes/ndustres. R = raw materals. X = country offered concesson n ndcated year. Data s more detaled for some countres than others. Approxmatons are made n certan cases. a An nvestment and guarantee law effectve as of 11 May 1997 offers the profts of a project formed under t to be exempt from tax on ndustral and commercal profts and from corporate tax. b More restrctons apply on domestc corporatons than on foregn corporatons. c 1998 budget abolshes payment of excse dutes. d Tax prorty status gvng exempton (or a reducton ) from varous taxes and custom dutes for up to 10 years can be obtaned by notfcaton of agreement. e Excse and sales and servce tax exemptons are granted to exporters of manufactured goods. f In general, exempton from payment of mport dutes on machnery and equpment and on raw and packagng materals and from ncome tax s avalable for those corporatons classfed as exportng companes. These exemptons also apply to free trade zones. g Industral enttes establshed n the jungle, fronter zones and free trade zones are exempt from ncome tax. h Exempton from value-added tax (VAT) s provded for ndustral enttes establshed n the jungle and fronter zones. New ndustral undertakngs satsfyng certan condtons establshed n a free trade zone, software technology park, or electronc hardware technology park, or a 100 per cent export-orented undertakng s entrely exempt from ncome tax. j Income tax holday gvng full exempton from corporate ncome tax for sx years for poneer frms and those locatng n less-developed area, and four years for non-poneer frms from the date of commercal operaton; expandng export-orented frms are gven three years. k Local purchases of goods and servces from VAT-regstered enttes are ether VAT exempt or zero rated. Should Countres Promote Foregn Drect Investment? 7
15 Table 5 TAX CONCESSIONS FOR INWARD FDI IN COMPARISON COUNTRIES, 1998 (I) (II) (III) (IV) Avalable to domestc corporatons? Corporate ncome Value-added Import duty EPZ Country tax exempton Perod Sectors tax exempton Items exempton Items provson (I) (II) (III) (IV) Chle X a X b E X Ireland * c Japan (tax credts) d X South Afrca X e X Thaland X 3 8 years P X f K, R, E Source: Prce Waterhouse (1998). Note: * = taxed at lower rate. A = agrculture. E = exported goods/exportng. K = captal. M = manufacturng. P = prorty companes/ndustres. R = raw materals. X = country offered concesson n ndcated year. Data s more detaled for some countres than others. Approxmatons are made n certan cases. a Tax benefts and other ncentves for companes operatng n northernmost and southernmost parts of the country. Tax benefts to forestry companes also. b Rembursement of taxes pad. c Reduced rate of corporaton tax of 10 per cent of profts from manufacturng operatons arsng between 1 January 1981 and 31 December 2010, regardless of whether the goods are exported. Defnton of manufacturng operatons s rather lenent. d A corporaton tax credt of 3.5 per cent or 7 per cent of the adjusted acquston cost (25 per cent to 100 per cent) of certan desgnated, energy-savng machnery and equpment, or 7 per cent of the acquston cost of certan desgnated machnery and equpment contanng electronc computer systems acqured by desgnated small or medum-sze corporatons s avalable. The credt s lmted to 20 per cent of the corporaton tax otherwse payable. e Tax holday granted at dscreton to an enterprse wth qualfyng assets n excess of R3 mllon, ncorporated on or after 1 October 1996, for the sole object of carryng out a qualfyng project. f Ether exempton or reducton. 8 G-24 Dscusson Paper Seres, No. 9
16 Should Countres Promote Foregn Drect Investment? 9 earnngs to Ford and Volkswagen n return for ther constructng a jontly operated automoble producton faclty n the country; n 1995 Ireland granted employment subsdes to IBM and Ctbank for locatng data-processng jobs n the country (and granted smlar subsdes to Berg Electroncs the followng year); and n 1996 Turkey entced Honda to buld a new automoble producton faclty n the country by easng tax rules on new plants and relaxng mport dutes on automoble parts. Subsdes are by no means lmted to relatvely low-ncome regons. Germany offered nvestment subsdes to Advanced Mrco Devces n 1995, after t decded to buld a sem-conductor plant n Saxony, and to Motorola n 1998, after t decded to buld a new faclty n Bavara. In the Unted States nvestment subsdes from state governments helped attract Mercedes-Benz to Alabama and BMW to North Carolna. Tables 2 5 ndcate that most countres do not follow the resdence prncple (see note 5) n settng tax polces. 14 Whle many countres tax domestc ncome earned by foregn corporatons at lower rates than domestc ncome earned by domestc corporatons, the former rates are n most cases above zero n the long run (though often not durng the frst few years followng the establshment of a project). Foregn tax credts, whch allow corporatons to deduct taxes pad to foregn governments from ther tax lablty on foregn ncome, complcates the pcture. As of the md-1990s, Japan, the Unted Kngdom and the Unted States granted foregn tax credts to multnatonal corporatons based wthn ther respectve borders, and many other hgh-ncome countres ncludng Australa, Canada, France, Germany, the Netherlands and Swtzerland exempted the foregn earnngs of ther frms from domestc taxaton (Hnes, 1996). Where foregn tax credts apply, and where a country s tax rate on domestc ncome earned by foregn corporatons does not exceed the home-country tax rate for these frms, taxng foregn corporatons merely shfts tax revenue from FDI source countres to FDI host countres and does not necessarly dstort nvestment. One mportant queston s whether the concessons offered to foregn corporatons shown n table 2 represent favourable treatment of nward FDI relatve to nward foregn portfolo nvestment. A full evaluaton of the ssue s beyond the scope of ths paper. Income from nward portfolo equty nvestment, portfolo debt nvestment, and drect nvestment are governed by complcated tax rules, whch vary consderably across countres. 15 We shall, however, hazard a few general comments. Table 2 shows that ncome from nward drect nvestment s subject to myrad tax breaks n G-24 and other countres. Wth regard to nward portfolo equty nvestment, the presence of captal gans taxes, whch vary from country to country, would tend to dsfavour ths vehcle relatve to FDI. 16 Wth regard to portfolo debt nvestment, the recent abolton of wthholdng taxes on portfolo nterest ncome for foregn resdents (Av-Yonah, 1999), whch has occurred throughout the OECD and n many developng countres as well, would tend to favour ths vehcle relatve to FDI. The absence of portfolo nterest ncome wthholdng taxes means that foregners do not pay tax on ncome they earn from corporate or government bonds, bank accounts or certfcates of depost n a country. To summarze brefly, t appears that many countres may have tax polces whch favour FDI relatve to some types of nward portfolo nvestment, but dsfavour t relatve to others. Other polces clearly do favour FDI. A country that offers exemptons to value-added taxes or mport dutes to foregn but not domestc corporatons favours FDI, snce domestc corporatons whch receve foregn bank loans, ssue bonds to foregners, or have a non-controllng porton of ther stock owned by foregners do not receve comparable tax breaks. EPZs do not favour FDI over foregn portfolo nvestment, so long as domestc frms have equal access to EPZs (whch s often the case) and are equally lkely to engage n export producton as are multnatonals (whch s less lkely to be the case). Drect subsdes to multnatonal frms (examples of whch we dscuss n secton IV) also favour FDI relatve to other forms of nward foregn nvestment. III. FDI and host-country economc performance There s mmense academc lterature on FDI and multnatonal frms. Snce our nterests are rather narrow, we focus here on emprcal research whch studes the mpact of FDI on host economes. Wthn ths lterature, we emphasze two strands: one whch examnes the determnants of where multnatonals locate producton facltes and another whch examnes sources of market falure related to FDI. We shall begn wth a bref revew of theores of multnatonal producton.
17 10 G-24 Dscusson Paper Seres, No. 9 A. What explans multnatonal producton? Followng Dunnng (1981, 1993), t s standard practce to vew multnatonal enterprses as arsng from three dstnct types of advantages. A frm must own or control a unque moble asset (e.g. a patent or trademark) t wshes to explot (the ownershp advantage); t must be cost effcent to explot the asset abroad n addton to, or nstead of n, the frm s home country (the locaton advantage); and t must be n the frm s nterest to control the asset s explotaton tself, rather than contractng out use of the asset to an ndependent foregn frm (the nternalzaton advantage). For nstance, GM wll engage n FDI when t has a desgn for a car whch could be manufactured abroad more effcently than at home and whose producton the frm wshes to control through ownershp of the factory n whch t s made. When GM wshes to sell cars n Brazl (a case we wll consder later), t wll choose FDI settng up ts own subsdary n the regon when ths opton domnates exportng to Brazl from GM plants n the Unted States (or elsewhere) and contractng out producton of GM cars (or lcensng ts technology and brand name) to an ndependent Brazlan frm. The general equlbrum theores of multnatonals attempt to explan how envronmental condtons arse whch favour producton by multnatonals over other forms of global market ntegraton. 17 Common to these theores s the dea that to produce a good a frm must ncur fxed costs such as R&D to generate a patent, advertsng to create a brand name, or corporate nvestments to establsh a management structure whch can support producton n many plants. A frm conssts of an upstream faclty, whch undertakes fxed-cost headquarters actvtes, and one or more downstream producton plants. A multnatonal s smply a frm wth upstream and downstream facltes located n multple countres. Suppose that headquarters actvtes are relatvely skll- or captal-ntensve and that producton s relatvely labour-ntensve. If factor prces are not equalzed across countres, then a frm has an ncentve to become a multnatonal n order to explot dfferences n factor costs between countres. It could do so by locatng ts headquarters n a captal-abundant (low-captal cost, hgh-wage) country and producton n a labour-abundant (hgh-captal cost, low-wage) country. Ths would gve rse to vertcal FDI the creaton of a multnatonal whose country operatons specalze each n a dstnct vertcal stage of producton (Helpman, 1984; Helpman and Krugman, 1985). Alternatvely, suppose that factor prces are equalzed across countres, or nearly so, but trade barrers or transport costs make t expensve to shp goods abroad. When trade costs are low, a frm wll produce all ts output n domestc plants and serve foregn consumers through exports. When trade costs are hgh, t s optmal for the frm to buld producton plants both at home and abroad, so that t serves domestc consumers from ts domestc plants and foregn consumers from ts foregn plants (Markusen, 1984). Ths s a case of horzontal FDI, n whch the multnatonal undertakes smlar producton actvtes n all countres. Actvtes at ts headquarters, however, reman concentrated n one country only. Recent work combnes cross-country dfferences n factor and trade costs to develop a general framework for determnng when multnatonals (versus purely natonal frms whch may or may not export) wll be n operaton (Markusen and Venables, 1998 and 1999a). 18 The creaton of multnatonal frms rases global welfare by leadng to a more effcent global allocaton of resources. 19 In the models descrbed above, the absence of dstortons specfc to FDI means that there s no polcy justfcaton for treatng multnatonals dfferently from domestc frms. Nevertheless, polces to promote FDI would encourage multnatonal producton by rasng the advantages of multnatonalty. From the perspectve of a multnatonal frms, subsdzng FDI () lowers producton costs and so rases the ncentve to create patents, trademarks, or other assets whch sustan headquarters actvtes; () enhances the relatve attractveness of locatng producton n the country offerng ncentves; and () rases the economc benefts of FDI relatve to exportng and arms-length producton n the host country. An emergng body of lterature consders how producton externaltes affect the behavour of multnatonals and the mpact of FDI on host economes. The exstence of externaltes assocated wth FDI rases the possblty that promotng FDI may be welfare-enhancng. One lne of work suggests that the arrval of multnatonal frms n an economy may help jump-start the process of ndustral development by ncreasng the scale of operatons n domestc upstream and downstream ndustres that s, by creatng forward and backward lnkages. In Rodrguez- Clare (1996), the arrval of multnatonals ncreases an economy s access to specalzed ntermedate n-
18 Should Countres Promote Foregn Drect Investment? 11 puts (whch are produced n more developed economes and accessble abroad only through multnatonals), whch n turn rases the economy s total factor productvty. The dea s that multnatonals n effect gve less developed economes access to the stock of knowledge captal (whch s one nterpretaton of specalzed ntermedate nputs) n more developed economes. Ths access makes labour and other factors n the host economy more productve. Multnatonals crowd domestc frms out of producton, whch rases the possblty that some domestc factors of producton may lose from FDI. Under some condtons, the demand for labour by the enterng multnatonals s weaker than that for the extng domestc frms, n whch case the arrval of multnatonals may lower natonal welfare. Thus, the lnkage effect of multnatonals on factor demand may be postve or negatve. There are other mechansms through whch multnatonals mpact ndustral expanson n host countres. Gao (1999) offers a model n whch the creaton of multnatonal frms spreads ndustry from more to less ndustralzed countres, thereby reducng ndustral concentraton n the former. By lessenng ndustry agglomeraton, multnatonals rase natonal welfare n less ndustralzed regons, but may lower t n more ndustralzed ones. Markusen and Venables (1999b) offer a smlar model n whch the catalyst effect of multnatonals on a host economy (through forward and backward lnkages) may spur domestc ndustry so much as to drve multnatonals out of the market. Addtonally, whle multnatonals create forward and backward lnkages, they also ncrease competton for local frms, and thus may redstrbute ncome away from some groups e.g. specfc factors n ndustral producton (Matouschek, 1999). If the mpact of multnatonals on the proftablty of domestc frms s suffcently negatve, FDI may lower host-country welfare, n whch case the optmal polcy towards FDI s a tax (Glass and Sagg, 1999). Another mechansm through whch multnatonals may create spllovers for local ndustry s worker tranng. If multnatonals brng new technology nto an economy, then local frms may beneft by beng able to hre workers whose tranng costs have n effect been pad by the multnatonal (Motta et al., 1999). Of course, multnatonals may bd to retan these workers, n whch case domestc labour captures the full beneft of worker tranng. In ether case, multnatonals rase natonal welfare. B. What determnes the locaton of multnatonal producton? A large emprcal lterature examnes the factors whch determne where multnatonal enterprses locate ther producton facltes. We dscuss ths lterature brefly n order to dentfy the potental effcacy of usng tax ncentves to attract FDI. One strand of lterature apples general equlbrum theores of multnatonals to data to see whether FDI s assocated wth varaton n trade costs or factor costs across countres. Theory predcts that frms wll penetrate foregn markets through FDI when trade costs are low, frm-level scale economes are hgh (.e. the fxed costs assocated wth headquarters actvtes are hgh), and plant-level scale economes are low (.e. the costs of havng plants both at home and abroad are low). Conversely, frms wll penetrate foregn markets through exports when trade costs are low and plant level scale economes are hgh. Theory also predcts that frms wll penetrate foregn markets through vertcal FDI when factor-cost dfferences between countres are large, and through horzontal FDI when countres are smlar n terms of market sze and factor cost. A common measure of how frms from a gven source country penetrate a gven host market s the level of sales n the host market by foregn afflates of frms from the source country (normalzed by total sales from the source country to the host country, whch equal afflate sales plus exports). 20 Usng data on exports by US manufacturng ndustres and sales by foregn manufacturng afflates of US multnatonals, Branard (1997) fnds that afflate sales n a gven ndustry and country are postvely correlated wth trade costs (freght rates, tarff rates) to the country and average frm sze n the ndustry, and negatvely correlated wth average plant sze n the ndustry. Branard nterprets these results to mean that, consstent wth theory, hgher trade costs and stronger frm-level scale economes encourage FDI relatve to exports, whle stronger plant-level scale economes dscourage FDI relatve to exports. She also fnds that hgher host-country taxes appear to encourage afflate sales over exportng, whch s counter-ntutve. Yeaple (1999) extends Branard s approach and fnds that for more skll-ntensve ndustres afflate sales are postvely correlated wth average educatonal attanment n the host country. He nterprets
19 12 G-24 Dscusson Paper Seres, No. 9 ths result to mean that countres wth larger supples of human captal are more lkely to attract FDI, especally n sectors whch are relatvely ntensve n the use of sklled labour. In related work, Markusen and Maskus (1999a and 1999b), usng data on aggregate sales by foregn afflates of US multnatonals and aggregate sales by afflates of foregn multnatonals n the Unted States, fnd that afflate sales are hgher when source and host countres have relatvely smlar market szes but are unrelated to dfferences n the relatve supply of sklled labour n source and host countres. They nterpret these results to mean that multnatonals have expanded ther global producton operatons more through horzontal FDI than through vertcal FDI. The emprcal work cted above s general equlbrum n orentaton, n that t s based on estmatng a reduced form relatonshp between exports and/or multnatonal sales and measures of ndustral technology, and country trade costs, sze and factor supples. All prces and outputs are mplctly endogenous. Other research takes a partal equlbrum approach n that t examnes the mpact of polcy or other host-country condtons on FDI, holdng constant at least some prces and sectoral outputs. Ths lne of work focuses on the country characterstcs whch appear to attract multnatonal frms. In a wdely cted paper, Wheeler and Mody (1992) examne outward foregn nvestments by US multnatonal enterprses. They fnd that US outward FDI s hgher n countres wth larger markets, a larger stock of ntal FDI, hgher qualty of nfrastructure, and more ndustralzed economes. These results are broadly consstent wth theory. The authors nterpret the correlaton between current and past FDI to ndcate that multnatonals are attracted to locatons wth a larger concentraton of ndustral frms that s, that there are agglomeraton economes assocated wth FDI. Other results are less consstent wth theory. FDI s slghtly hgher n countres wth hgher labour costs and corporate taxes. These fndngs, whch are representatve of other papers n ths lterature, rase a number of mportant questons. Is FDI truly subject to agglomeraton economes? Can t be that multnatonals do not take labour costs or tax rates nto account n makng locaton decsons? If the answer to both questons s yes, then host-country governments may have lmted ablty to nfluence multnatonal locaton decsons through tax polcy. There are several reasons to be cautous n drawng polcy conclusons from regressons lke those that Wheeler and Mody report. One common problem n emprcal work whch examnes the mpact of tax polcy or agglomeraton effects on frm locaton decsons s that t s often dffcult to control for the mpact of all relevant regonal characterstcs (e.g. effcency of local bureaucracy and local factor productvty). 21 Excluded characterstcs are omtted varables, whose exstence may contamnate regresson results. Calforna, for nstance, attracts multnatonals n part because t has a large pool of labour whch s hghly sklled (and hghly pad). The state s abundant resources may allow t to get away wth hgh corporate tax rates. If there are many nstances lke Calforna n the data (as well as opposte cases lke Arkansas whch s a low wage, low tax, and low FDI regon), we would fnd that FDI s postvely correlated wth tax rates, labour costs and ndustry agglomeraton. But these correlatons would just be pckng up the more fundamental relatonshp between FDI and the local supply of sklled labour. Snce Wheeler and Mody-style regressons do not control for ths sort of possblty, they are dffcult to nterpret. Recent work n publc fnance attempts to address these and other dentfcaton problems. Hnes (1997) summarzes research on the mpact of taxaton on FDI. Contrary to the mpresson gven by Wheeler and Mody (1992), Branard (1997), Yeaple (1999) and others, a growng tax lterature fnds that FDI s lower n regons wth hgher corporate taxes. The elastcty of FDI wth respect to the after-tax rate of return s approxmately unty. One study whch controls for omtted varables n a partcularly convncng manner s Hnes (1996), who examnes the allocaton of nward FDI across US states. He compares nvestment from countres whch grant foregn tax credts wth countres that do not n hgh-tax versus low-tax US states. Relatve to nvestors from countres whch grant foregn tax credts, nvestors from countres whch grant no tax credts should be more senstve to cross-state dfferences n tax rates. Hnes approach allows hm to control for unobserved factors whch nfluence the attractveness of a state to foregn nvestors (and whch are common to nvestors from dfferent countres). Hs results mply an elastcty of captal ownershp wth respect to state taxes of 0.6. Ths suggests that multnatonals are nfluenced by cross-country or cross-regon dfferences n tax rates. Wth regard to agglomeraton effects, Head et al. (1995) examne the locaton decsons of new Japanese manufacturng plants n the Unted States. They fnd that, controllng for the local concentra-
20 Should Countres Promote Foregn Drect Investment? 13 ton of US plants n the same ndustry (among other factors), Japanese plants are more lkely to choose a locaton where the exstng local concentraton of Japanese manufacturng plants n the same ndustry s hgher. Ths fndng s smlar to Wheeler and Mody s that frms are attracted to large concentratons of other ndustral frms. But, by focusng on Japanese plants and controllng for the locaton of overall US manufacturng actvty, ths approach goes further than prevous studes n controllng for the effects of unobserved, ste-specfc characterstcs. 22 Ths suggests that multnatonals (at least the Japanese frms n Head et al. s data) are attracted to locatons wth other frms n ther own or related lnes of actvty. 23 C. Does FDI generate postve spllovers for the host economy? That multnatonal frms are dfferent from purely domestc frms s abundantly clear. Across countres and tme, several emprcal regulartes are apparent. Relatve to ther domestc counterparts, multnatonals are larger, pay ther workers hgher wages, have hgher factor productvty, are more ntensve n captal, sklled labour, and ntellectual property are more proftable and are more lkely to export (Haddad and Harrson, 1993; Atken et al., 1997; Atken et al., 1997; Atken and Harrson, 1999; Blomstrom and Sjoholm, 1999). 24 That multnatonals possess these attrbutes s not surprsng, gven that to become a vable multnatonal a frm must have outperformed domestc and foregn rvals n some dmenson. The relatve technologcal superorty of multnatonals also makes t possble that they would be a drect or ndrect source of technologcal advancement for domestc frms n host countres, especally where these countres are relatvely far from the technologcal fronter. Theory dentfes several channels through whch multnatonals generate externaltes that rase the productvty of host-country factors of producton. It s entrely possble, however, for the net effect of such lnkages on host-country welfare to be negatve, once we take nto account the mpact of FDI on the proftablty of domestc frms. Whether spllovers from multnatonals rase host-country welfare s an emprcal queston. Early lterature s optmstc about the mpact of multnatonals on host-country productvty. 25 Caves (1974) fnds a postve correlaton between ndustry average value-added per worker and the share of ndustry employment n foregn frms for Australan manufacturng n More recent work confrms ths basc fndng n a wde array of envronments. A partal lst of studes whch fnd a postve correlaton between average ndustry productvty and the presence of foregn frms n the ndustry nclude Globerman (1979) for Canada n 1972; Blomstrom and Persson (1983), Blomstrom (1986), and Kokko (1994) for Mexco n the 1970s; and Blomstrom and Sjoholm (1999) for Indonesa n In related work, Borenszten et al. (1998) fnd a weak postve correlaton between FDI nflows and per capta GDP growth for a panel of countres n the 1970s and 1980s (although when they nteract FDI and the level of schoolng, FDI has a negatve drect effect on growth and a postve ndrect effect through schoolng). Takng a slghtly dfferent approach, Mansfeld and Romeo (1980) use a sample of US-based multnatonals to examne whch sorts of technology these frms transfer abroad and whether there s leakage of these technologes to non-us frms n host countres. They report that n 20 out of 26 cases transferred technologes became known to foregn rvals wthn sx years. In nne of the cases, access to US technology appeared to accelerate foregn frms ntroducton of competng products or processes by two years or more. 27 What does t mean for ndustry productvty or ndustry technology adopton to be postvely correlated wth the presence of multnatonal frms n an ndustry? A common nterpretaton of ths fndng s that FDI creates postve productvty spllovers for domestc frms n host countres. Ths nterpretaton, however, s subject to the same concerns about omtted varables and endogenety bas that we encountered n emprcal work on the mpact of taxes and agglomeraton effects on frm locaton decsons. Most emprcal studes n the lterature use crosssecton data on average ndustry characterstcs. A postve smple correlaton between ndustry productvty and the presence of multnatonals s, n prncple, just as lkely to mean that multnatonals are attracted to hgh-productvty ndustres as t s to mean that multnatonals rase host-country productvty. Though most emprcal studes ntroduce addtonal controls n estmatng the correlaton between ndustry productvty and multnatonal presence, the ncluded varables surely do not exhaust the set of factors whch are lkely to nfluence ndustry productvty and multnatonalty. 28