INTRODUCTION
Practical Motivation andResearchProblems
When a country receives increased aid from a donor, it tends to recognize the donor's generosity, foster cooperation, and create a favorable legal environment for the donor's enterprises (Kimura & Todo, 2010; Rodrik, 1995) This aid can enhance the recipient's social and economic infrastructures, leading to improvements in human capital and total factor productivity (Harms & Lutz, 2006) Consequently, the recipient country may attract more foreign direct investment (FDI) due to heightened competitiveness Thus, foreign aid is often viewed as a crucial factor in driving private capital inflows, which are believed to significantly promote growth, technology, and employment in the host country However, research has not consistently demonstrated a strong relationship between foreign aid and FDI (Alesina & Dollar, 2000; Harms & Lutz, 2006).
HarmsandLutz(2006)suggestthatoneshouldconsidertheroleofpoliticalandinstitutionalcha racteristicswhenquantifyingthisrelationship.Indeed,institutionalqualityoft h e h o s t countryit selfi s a n importantdirectmagneto f privatecapitalinflows.Abundantempiricalstudieshavepointe doutthenegativecausalityofabadinstitutiontotheinflowsofFDI.Ironically,severalcountrieswhich areperceivedash a v i n g highcorruptiona n d l o w political,institutionalprofilesstillhavelargeinfl owsofFDI(Habib&Leon,2002).
Intheaspectofmodeling,theinfluenceofforeignaidonFDIisdifficulttoestimateduet o t h e pr oblemso f simultaneitya n d reversecausality.Byu s i n g laggedvariablesa s instruments,2SLSa n d GMMmethods,t o somee x t e n t , couldalleviates u c h e n d o g e n e i t y However,thetreatment ispurelytechnicalanddoesnot reflectthenatureoftheproblems.Asiedu,J i n , a n d Nandwa( 2 0 0
9 ) proposea simultaneousequationsmodelthatcoulds o l v e t h e s e problems.Inthisapproac h,foreignaidandFDIaredeterminedatthesametime,ande a c h of them isthedeterminantof theother.Whilethedualapproachis undoubtedly asuperbi d e a , theappliedmodelandtheresultsofthisresearchnonethelesscontainsomeflawsandc ontradictions.First,thereisnoinstitutionaldeterminantintheaidequation.Second,inthea i d eq uation,t h e positivecoefficiento f FDIc o u l d b e interpretedthatwhileforeigna i d r e d u c e s FDI,FDIcould,however,increaseforeignaid.
Research Objectives
(2009)andvisualizetheintricaterelationshipbetweenforeignaidandFDI,whichmightcomprisesi multaneityandreversecausality.Withregardtopurposes,whileAsieduetal.
(2009)focuso n thealleviatingroleofforeignaidontheadverseeffectofexpropriationriskonF DI,weconcentrateontheeffectofforeignaidon finalFDI.Withr eg ar d tosamples,weuseboth l ow-incomea n d middle-incomesubsamplest o s u p p o r t ouranalysis,whereasitisonlythelow- incomecountriesinAsieduetal.
(2009).IncomparisonwithHarmsandLutz(2006),wea p p l y adifferentmodelwithdifferentprox iesofvariablesandamorerecentperiodtoassesst h e effectofforeignaidonFDI.
This paper explores the intricate relationship between foreign aid and foreign direct investment (FDI) by analyzing data within a specified framework We aim to determine if multilateral aid can effectively lead to increased FDI Additionally, our research contributes empirical evidence regarding the significance of institutions in attracting FDI, aligning with theoretical expectations and previous studies We evaluate the impact of various institutional measures on FDI, while also investigating whether factors such as democracy, corruption control, and political stability enhance a country's ability to receive foreign aid.
Structure
InChapter2,thispaperbrieflyreviewsatradetheorywhichiswidelyusedtoexplaintheinvestmen tdecisionofforeigninvestorsandsomeempiricalresultsbasedonthistheory.Wemainlyconcentrate onthepapersthathaveinstitutionsandforeignaidasthedeterminantsofprivatecapitalinflows.InCh apter3,weexplainthedual- approachframeworkandtheregressionmodel.Thevariablesandd a t a sourcesarealsodescribedi n t h i s chapter.Theempiricalfindingsa n d associatedexplanationsa r e locatedi n Chapter4 Chapte r5 recapitulatestheresultsformakingpolicy andresearch.
LITERATUREREVIEW
WefirstreviewtheOLItheoryontheinvestmentdecisionofforeigninvestors;thenwecomeinto thepapers mentioninginstitutionsandforeignaidasseparateexplanatoryvariableso f FDI;n e x t , weh a v e a l o o k o n theresearchwhichembedst h e politicala n d institutionalfactorsintot h e influenceo f foreigna i do n FDI.Lastly,wesummarizetheinstitutional measureswhichmightaffectforeignaid.
The OLI paradigm, developed by Dunning in 1988, 1998, and 2001, serves as a general framework to explain the activities of foreign investors The ownership advantages, classified as the O component, emphasize the comparative advantages of firms that can expand their business abroad, providing insight into the nature of products and the ability of firms Location advantages, defining the L component, relate to human and natural resources, favorable conditions for production and business, and market size in the host country Internalization advantages, belonging to the I component, focus on lowering transaction costs, enabling firms to decide whether to import intermediate products from markets or internalize foreign suppliers into their production chain, a crucial consideration when choosing a destination.
Politicalandinstitutionalfactorsofthehostcountryareconsideredasthelocationadvantagesi n t heOLIframework.Theinfluencemechanismo f t h e s e factorso n foreigninvestorsarementionedi n t h e paperss u c h a s Habiba n d Leon(2002)a n d Dunninga n d Lundan(2008).Ontheempirica lside,HabibandLeon(2002)findout a negativerelationshipbetweencorruptionlevelsinthehostcountriesandtheirinflowsofFDI.Accordi ngtoHabib andLeon,foreign investorsmightseecorruptionasviolatingsocialand professionalethicsa n d increasingunnecessarycosts.1M o r e o v e r , payingbribesisstrictlyprohibitedinthehomecountriesofs omeforeigninvestorssuchastheUnitedStates (Hines,1995).
Bussea nd Hefeker(2007)examinet he impactsofgovernmentstability,law a nd order,a b s e n c e ofinternala n d externalconflicts,lacko f e t h n i c t e n s i o n s , controlo f corruption,de mocracy,a n d bureaucraticperformanceo n FDIinflowst odevelopingcountriesi n thep e r i o d 1984-
2003.Thep a p e r doess h o w positiverelationshipsbetweens u c h measuresa n d privatecapitalinfl ows.Withthesameperiodofresearch,Bénassy-Quéré,Coupet,andMayer
(2007)usethegravitymodeltotesttheinfluencesofvariousinstitutionaldataonFDIstocks.I n gener al,acountryowninghigherinstitutionalqualitywouldhavealargeraggregatestocko f FDI,a n d h i g h e r institutionaldistancebetweent h e homec o u n t r y andt h e h o s t c o u n t r y lowerstheFD Istockof theformerinthe latter.Inotherwords,bad institutionalmeasureswouldhinderFDItoa country.Forexample,oneoftheresultsinAsieduetal.
(2009)isthatexpropriationriskrestrainstheinvestmentdecisionofforeignenterprises.Thesampleso f thisresearch are low-incomeandSub-SaharanAfricancountries.
Ont h e otherh a n d , t h e numbero f researchstudiesthatfindo u t b a d institutionsa s a n incenti veofFDIisquitelimited.EggerandWinner(2005)showtheempiricalevidenceont h e positivere lationshipbetweencorruptionandinwardFDI.Theresearchusesasampleofb o t h developedan ddevelopingcountriesintheperiod1995-
The relationship between foreign aid and foreign direct investment (FDI) is complex and often contradictory While some studies suggest that foreign aid can positively influence FDI, others highlight adverse effects, such as infrastructure and rent-seeking issues (Harms & Lutz, 2006; Selaya & Sunesen, 2012) Many research efforts have failed to identify a significant overall impact of aggregate aid on FDI (Bird & Rowlands, 1997; Harms & Lutz, 2006; Kimura & Todo, 2010) However, Selaya and Sunesen (2012) found evidence supporting the positive influence of bilateral aid on attracting FDI, a finding echoed by Rodrik (1995) and Kimura and Todo (2010) In contrast, there is a lack of empirical support for the immediate enhancement of FDI through multilateral aid (Rodrik, 1995) Additionally, Asiedu et al (2009) reported negative coefficients for aggregate, bilateral, and multilateral aid in FDI regressions, particularly in low-income and Sub-Saharan African nations.
Harms and Lutz (2006) investigate the unclear relationship between foreign aid and foreign direct investment (FDI) by considering institutional variables and employing various estimation techniques across different time periods and country groups, including the disaggregation of foreign aid types and private foreign investment Despite their efforts, they find that, on average, foreign aid does not impact FDI However, they identify a significant finding: in countries with high regulatory hindrances and low institutional quality, foreign aid positively influences FDI This effect arises because foreign aid helps reduce expropriation risk, thereby encouraging foreign investment, a phenomenon that is particularly evident in these nations Furthermore, the positive interaction between expropriation risk and foreign aid is also noted in the findings by Asiedu et al.
( 2 0 0 9 ) meanst h a t expropriationr i s k ispositivelyaddedtotheeffectofforeignaidonFDI.Given thenegativecoefficientofforeigna i d a t theb e g i n n i n g , theincreaseofexpropriationr i s k leve l,t o somee x t e n t , couldmakeforeignaidmovinginthesamedirectionwithFDI,althoughexpro priationisperceivedasab a d practice.
Inregardtoinstitutionaldeterminantsofforeignaid, AlesinaandDollar(2000),Dollara n d L evin(2006)h a v e c o n s e n s u s ont h e positiveimpactofdemocracy.Corruptionlevel,however,i s no t foundtoultimatelyaffectthedecisiono f a majorityo f d o n o r s (Alesina& Weder,2002).
MODEL AND DATA
Dual-ApproachFramework
Figure1graphicallyillustratestheinfluenceofinstitutionsonforeignaidandFDI,andt h e in teractionbetweenthesetwoforeigninflows.Theformerinducessimultaneityandthelattercause sreversecausality.Solidarrowsarethefociofthispaper.Inthe bottomofthefigure,institutionsaret hemajorfactorsof bothFDIandforeignaid.TheOLItheorysuggestst h a t abetterinstitutioncouldhavemoreFDIinfl ows.Similarly,foreignaidmightbe disbursedmoretoacountrywithbettergovernance.Hence, 2 positive. and 2 areexpectedtobe
1 Inc a s e o f multilaterala i d , i t aimst o enhancet h e sociala n d economicinfrastructuresa n di s typicallythoughta s creatingt h e “infrastructureeffect.
The impact of multilateral aid on infrastructure development is not immediate and varies between countries Even if country A receives more aid than country B under similar conditions, this does not guarantee that country A will quickly enhance its infrastructure or attract more foreign investors This is primarily because multilateral aid is generally not aimed at supporting specific industries Instead, it is the informational and conditional policy functions of such aid that can provide a protective framework for foreign investors in recipient countries This leads to the proposal of the "institutional effect" of multilateral aid, which suggests that while it can safeguard foreign investors, it does not alter the institutional landscape in the short term.
Incaseofbilateralaid,besidesproductivesectors,anumberofitsprojectsarealsoforinfrastruct uraldevelopment,anditshouldalsohavethesameinfrastructureeffectasthatofmultilateralaid.In addition,bilateralaidhasitsownspecialeffectwhichisknowntogreasebilateralprivatecapitalinflo wsandisreferredtoasthe“vanguardeffect.”Itisnoticedthatt h e vanguardeffecto f bilaterala i d i sq ui te similart o th e institutionaleffecto f multilateral aid, but onlyappliedforbilateralinvestors.Thecoefficient 1 isconjecturedtobepositivein bothcasesofforeignaid.Theimpactlevelisexpectedtovarysystematicallyamong
Rodrik(1995)fordetaileddiscussionontherolesofmultilateralaid. governmentsh a v i n g differentperformances.Therefore,themarginaleffecto f f o r e i g n aid couldb e expresseda s a linearfunctiono f institutions,i.e.
Thei d e a o f puttingpoliticala n d institutionalvariablesi n t o t h e effecto f foreignaido n FDIw a s firstemploye dbyHarmsandLutz(2006).
Foreign Direct Investment (FDI) can influence foreign aid through both direct and indirect mechanisms The direct mechanism is particularly relevant for understanding short-term bilateral aid flows, as increased FDI in a host country may motivate the home country's government to allocate more aid to that location While it is not mandatory for governments to do so, the relationship between economic activities and diplomatic ties makes the use of FDI as a factor in bilateral aid decisions reasonable Conversely, multilateral donors typically do not consider FDI levels in recipient countries when distributing aid In summary, in the short term, FDI inflows may have a positive effect on bilateral aid equations while remaining neutral in multilateral aid equations.
In the long run, foreign direct investment (FDI) can lead to significant changes in both bilateral and multilateral aid By enhancing GDP growth and GDP per capita, FDI ultimately influences the level of foreign aid a country receives If FDI inflows successfully contribute to the wealth of a host nation, that country may eventually become less dependent on concessional loans or grants from donors Consequently, the coefficients of FDI inflows in the equations for bilateral and multilateral aid are expected to turn negative over time.
The impact of Foreign Direct Investment (FDI) on foreign aid can be influenced by the quality of institutions in a country Research has shown that various institutional factors, such as democracy and corruption, can significantly affect a nation's GDP growth and GDP per capita By incorporating these institutional factors into the coefficient of FDI inflows, it is possible to assess how effectively an economy absorbs FDI and reduces its dependence on foreign donors.
3 Some conditionsthatFDIcauseseconomicgrowtharehumancapital(Borensztein,DeGregorio,&Lee,1998) ,absoluteandrelativenature ofFDI(Alfaro,2003;Mello,1999),andtradepolicy(Balasubramanyam,Salisu,&Sapsford,1996).
Dual-ApproachDynamics-BalancedModel
Fromthedual-approachframeworkabove,wehave thegeneraldual-approachmodel: fdi aid f 1 ( 1 aid, 2 insf, 1 others), f 2 ( 1 fdi , 2 insa, 2 others).
ThismodelaugmentsthesimultaneousequationsmodelofAsieduetal.(2009)byaddingthe politicalandinstitutionaldeterminant,reg ressionis insa, int h e a i d equations.Thedetailedmodelfor infdigdp it 1 aidgdp it 2 insf it 1 infdigdp i,t1 11 gdpgrow i,t2 12 utilcom i, t2
lngd ppc lngdppc 2 it 1 it 2 it 2 i,t 1 21 i,t 2 22 i,t 2 (4) where infdigdp i t
23 debtgdp i,t 2 24 lnpop it u 2it , ist h e n e t inflowso f foreignd i r e c t investment,a n d aidgdp it ist h e net inflowsofofficialdevelopmentassistance(ODA)tocountryiinyeart.Bothfiguresarein percentageofGDP.Thefirstlagofaidinflowsintheaidequations, aidgdp i,t 1, isamajor differenceincomparisonwiththepreviousresearchofAsieduetal.
(2009).Itcouldbeaddedb e c a u s e foreignaidisdisbursedgraduallywiththecompletenessofpr ojects.Togetherwith thefirstl a g o f FDIinflows, infdigdp i,t 1 , thisadditionmakest h e systemo f twoequations balanced,thusthemodelisreferredtoasdual-approachdynamics-balancedmodel 4
(2009),b u t someadjustmentsaremade.IntheFDIequation,theproxyofeconomicinfrastructurei st h e totaldensityofcommunicationutilities(utilcom),whichiscomprisedoftelephonelines,
4 Some papers usingthefirstlagofFDIasan explanatoryvariableareAsiedu etal.
(2009),BusseandHefeker(2007),Jensen(2003),andGastanaga,Nugent,andPashamova(1998). mobilecellular,andinternetsubscribers.However,theincreaseofcurrentFDIinflowsmightp r o v o k e theusageofcommunicationsystem.Thus,welagthisvariabletoavoidtheinteractionwithth edependentvariableanditsfirstlagintheright-handsideoftheequation.
Thetrade openness(tradeopen) asmeasuredbythetotalexportsandimportsoverGDPis alsolaggedtwoperiodsbecausethereisadirectcontributionofFDItothisvariableinthesameye ar.Asforeigninvestorsexpandtheirbusinessactivitiestoothercountries,eitherint h e formofgre enfieldormergera n d acquisition,t h e y a r e morelikelyt o e x p o r t machines,equipment,materials, andexpertsto,andimportfinalproductsfromthehostcountry.Alsot h e consumptiono f expatria tesraisesthein-border-exportscomponento f t h e destinationc o u n t r y
Asdepicted inFigure1,FDIinflowscould affecttheGDPgrowthofacountry withint h a t year,sogdpgrowi salsolaggedtwotimestoeliminatethereversecontributionofFDIinflowsintot hisexplanatoryvariable.Finally,theinstitutionsthatmightaffectFDIinflows
(insf) areregulatoryquality,controlofcorruption,governmentefficiency,ruleoflaw,and politicalstability.Theimpactoftheseinstitutionalmeasureswillbeexaminedseparatelyint h e regr ession.
Intheaidequations,toavoidthe componentrelationshipofforeignaidand FDIinthe currentGDPp e r capita,wel a g thevariable lngdppc Apartfromt h e projectednegative relationshipbetweentheincomelevelofacountryandtheamountofaidthatitreceives,wea l s o e xpectthatthedecreasingratedoesnotholdconstant,inparticulardiminishes.Hence,t h e quadr aticcomponentofl n g d p p c i sadded.Thesecondmacrovariablewhichisdebtover
Moreover,thedebtinthepastismoreappropriatetobeadeterminantofaiddisbursementsint h e future. Thepoliticalandinstitutionalvariablesofforeignaid(insa) aresupposedtobe democracy,controlofcorruption,andgovernmentefficiency.Asademographicfactor, population(lnpop) isusedbecauseitcouldcreatethesmall-countryeffectontheinflowsof foreignaid.Withthesameinstitutionsandotherthingsequal,acountrywithlowerpopulationism orehomogenousandmorelikely toreceivethesupportofforeigncommunity.
NetFDIinflows(%GDP) 3.693 7.374 3,711 3.628 9.731 3.733 5.498 Aggregateaid(%GDP) 9.129 13.28 3,827 16.14 14.77 4.787 10.06 Bilateralaid(%GDP) 6.119 10.42 3,827 10.27 10.96 3.546 9.178 Multilateralaid(%GDP) 3.020 4.596 3,815 5.871 5.825 1.247 2.227 Regulatoryquality -0.494 0.714 1,922 -0.866 0.614 -0.264 0.674 Controlofcorruption -0.480 0.642 1,928 -0.737 0.557 -0.321 0.639 Governmentefficiency -0.499 0.656 1,922 -0.865 0.582 -0.272 0.595
Note:Ful l sample(144countries)includeslow-income(54 countries)andmiddle- income(90 countries)subsamples.
Thetargetsampleo f t h i s s t u d y i s low-a n d middle- incomecountriesbecausemosto f t h e s e countrieshavereceivedforeignaid.Thecountryclassific ationsbasedonGNIpercapitaa r e oftheWorldBankandcapturedin2013 6The macrodataaremost lytakenfromWorldDevelopmentIndicators(WorldBank,2014)andWorldEconomicOutlook(IM F,2014).Theinstitutionald a t a whichi s a d o p t e d fromWorldwideGovernanceIndicators(Worl dBank,2 0 1 2 )includesregulatoryquality,controlofcorruption,governmentefficiency,ruleofla w, andpoliticalstability.Duetothetemporallimitationofdata,theresearchperiodisnarroweddownto1996-2012.Inaddition,weusedemocracyscorefromFreedomHousetodiversify
5 Dudley andMontmarquette( 1 9 7 6 ) discussi n detailt h e relationshipbetweensmallcountryandforeignaid.
2013incomeclassificationsoftheWorldBankarelowincome(nomorethan1,035US$),middleincome(1,036-12,615US$),andhighincome(nolessthan12,616US$). thesourcesofthesesubjectivevariables.Table1summarizesthedescriptivestatistics.Moredetailsof thevariables andthedatasourcesareshowninTableA1(AppendixA).
RESULTS
Independent MarginalEffectsbetweenInstitutions,ForeignAid,andFDI
Table 2 presents the regression results based on the DADB model, with Panel A detailing the base regression for different types of foreign aid: aggregate (AA), bilateral (BA), and multilateral (MA) The analysis focuses on three key variables: FDI inflows, foreign aid, and institutional quality In the FDI equation, five institutional measures—rule of law, regulatory quality, control of corruption, government efficiency, and political stability—are evaluated individually, revealing a positive impact on FDI inflows Panel A of Table 3 reports the coefficients and significance levels for the full sample, indicating that, except for government performance, all other institutional and political measures effectively attract more FDI, with significance levels of at least 5 percent Notably, the quality of regulation in the private sector has the most substantial effect on the investment decisions of multinational enterprises (MNEs), where a one-point increase in this measure correlates with a 0.66 percentage point rise in FDI inflows.
The institutional measures significantly influence the dependent variable, leading us to utilize principal component analysis (PCA) to create a composite institutional index, denoted as INSF, which reflects the impact of government policies on promoting foreign direct investment (FDI) This index will be instrumental in subsequent FDI-related regressions In addition to streamlining our workload, this composite approach is justified as these measures typically exhibit strong correlations Generally, sound governance is associated with high scores in areas such as the rule of law, regulatory quality, control of corruption, and government efficiency, while minimizing the risks of political instability, civil unrest, and external conflicts.
Bilateral aid has a positive coefficient in the FDI equation, indicating that a 1% increase in bilateral aid results in approximately a 0.18% rise in FDI, consistent with Rodrik (1995) Conversely, the impact of multilateral aid is significantly greater, with a coefficient of 0.39, as it attracts foreign investors from all countries, unlike bilateral aid, which only influences FDI from donor nations An analysis of low-income and middle-income countries shows that the effect of foreign aid on FDI is more than double in low-income countries, although institutional effects are negligible in this group This discrepancy may arise from an institutional threshold perceived by foreign investors; when host country institutions fall below this threshold, differences are overlooked Additionally, low-income countries, which often have lower institutional quality on average, may benefit more from the extended vanguard effect of bilateral aid and the institutional impact of multilateral aid Compliance with international regulations enhances investor confidence in these countries, where foreign aid levels are significantly higher than in middle-income nations.
Othercontrolvariablesi n t h e FDIequationarestatisticallysignificanta n d havet h e e x p e c t e d signs.WhenGDPgrowthortradeopennessincreasesby1percentageofGDP,then e t inflo wso f FDIa f t e r twoy e a r s c o u l d b e higherb y 0.06o r 0 0 1 percentageo f GDP,respectively Whenthetotalnumberoftelephonelines,mobilesubscribers,andinternetusersi s higherby10unit sper100people,privatecapitalinflowsinthenexttwoyearscouldbelargerby
Int h e a i d equations,theinsignificantcoefficientso f FDIinflowsi n t h e s e c o n d parto f colum ns(1),(2),and(3)ofTable2illustratethenon- impactofFDIonforeignaid.Itmeanst h a t acountrywithmoreFDIdoesnotnecessarilyhavemor eorlessofbilateralormultilateralaid Inotherwords,althoughFDIcould motivatemoreaidfrom somespecificd o n o r s , t h e totalso f bilaterala i d inflowsbetweenrecipientcountriesa r e n o t ultimatelydifferentduetotheirtotalreceivedFDI.ThisresultisinlinewiththestatementofAlesi naa n d Dollar (2000):thereis norelationshipbetweenFDIandforeignaid.
The Democracy Index, which evaluates political rights and civil liberties, plays a crucial role in the allocation of both bilateral and multilateral foreign aid Negative coefficients indicate that countries perceived to have greater freedom receive more foreign aid Additionally, factors such as control of corruption and political stability significantly impact aid distributions from donors Notably, empirical findings reveal that higher levels of corruption lead to reduced foreign aid, contrasting with previous research by Alesina and Weder (2002) Furthermore, the coefficients for political stability are notably larger than those for corruption in this context To better assess governance indicators that may enhance a country's ability to secure foreign aid, we have developed a composite institutional index known as INSA The effects of Foreign Direct Investment (FDI) and institutions on foreign aid within low-income and middle-income groups are summarized in the latter sections of Table 3.
Othercontrolvariablesi n t h e aidequationshavestatisticallysignificantimpacto n t h e distributi ono f foreigna i d Thesquaredcomponento f t h e incomelevelimpliesn o t o n l y a negativemargina leffectbutalsoadiminishingrateofthiseffect.Additionally,theincomel e v e l i s muchmorese nsitivea n d significantint h e multilateralaidequationthani n t h e bilateralaidversion.Thisfigur eprovesthetargetofreducingpovertyofmultilateralorganizations.Finally,acountrywithmorede btorlowerpopulationismoreopenandlikelyt o l o o k forthehelp offoreigncounterpartsandinternationalorganizations.
ReliabilityandRobustnessChecks
Inthispart,wec h e c k t h e reliabilitya n d t h e robustnessoft h e specifiedmodel.Thereliabili tycheckisdonebynotlaggingthecontrolvariablesofthetwoequations.Besidesthesmaller- sampleapproachmentionedabove,therobustnesscheckisundertakenbythreemoreexperiments.Fi rst,acontrolvariableisremovedfrom thespecification.Secondexperimentist h e insertionofsomeirrelevantvariables.Third,different methodsofregression— whicharePOLS,differenceGMM(Arellano& Bond,1 9 9 1 ),a n d s y s t e m GMM(Blundell& B o n d , 1 9 9 8 )—areappliedseparatelytoeachofthe equations.
(9)]ofTable2reportstheresultswhenwedonotlagthecontrolvariables.Therea r e severalreasonst h a t t h e lagv e r s i o n i s morereliablet h a n t h e n o n - l a g v e r s i o n o f thespecification.First,theeffecto f GDPg r ow t h i n t h e non- lagspecificationisattenuatedto0.05,incomparisonwith0.06inthelagspecification.Thisattenuation iscaused
real bythefeedbackofthedependentvariable.Theprocedureofreversecausalityisexplainedasfollowi ng.GDPgrowth,whichrepresentstheproductionandconsumptioncapabilitiesofaneconomy,ison eofthemostimportantcriteriawhenMNEschooseacountrytoinvest.Theh i g h e r GDPgrowth o f a c o u n t r y , themoreopportunitiesa r e thattheinvestmentc o u l d b e a b s o r b e d andt u r n e d intoprofits.Int u r n , whenforeigni n v e s t o r s decidet o expandt h e i r b u s i n e s s t o a hi gher- growthcountry,t h e y a l s o contributetot h e productiono f g o o d s a n d serviceswithinthatcount ry.Asaresult,GDP growthisspirallyaffectedbyFDIinflows.
Tomathematicallyillustratet h e attenuatedcoefficiento f a positivelyfeedback-affected explanatoryvariable x,let denoteaself-changeofthisassumedexogenousvariableas
real y/x.Next,thechange yofyenlargestheself-changeofxanadditionalamount
x nominal xx feed Thecoefficientofx whichweobtainin themultivariate regressionofyis
Int h i s case,wehave nominal 0.05, real 0.06,and h0.05/0.060.83
Itmeansthat, whentaking into accountthe reversecausalityofFDIinflows,themarginaleffectofGDPgrowthis downward-biasedandequalto 83percentofitsrealmagnitude.
Second,incontrasttoGDPgrowth,thecoefficientoftradeopennessinthenon- adjustedspecificationismuchlargerandmoresignificantthanthatintheadjustedspecification.This phenomenonoccurswhenxoryisthecomponentofeachother.Supposethattheeffectofx onyis ,t h u s theeffectrelationshipisyx
Inaddition,duetothecomponentialnature betweenyandx,thecomponentrelationshipis yb x Themultivariateregressionofyonx wouldgenerateacombinationas y(b)x.
Therefore,thecoefficientoftheend ogenous variablexincreasesinthevalueifthecomponentrelationshipispositive(b0) Moreover, thesignificancelevelsalwaysincreasea s a resulto f directcomponents,eitherpositiveo r negative,in theregression.Inour case, wehave 0.01and b0.03.Similarly,asmallcomponentrelationshipcouldalsobefou ndbetweentheamountofdebtandtheinflowsof aidi n t h e aidequationso f t h e n o n - a d j u s t e d specification.Foreigna i d i s a c o m p o n e n t o f s o v e r e i g n debtwithinthesameye ar.
Third,amodeloraspecificationisconsideredmorereliablewhenitsregressioncoefficientsares tabledespitechangesintheproxiesofothervariables.Onthiscriterion,thetwo- lagspecificationisalittlebetterthanthenon- lagcounterpart.ThecoefficientoflagofFDIvariablei n t h e formeri s stablea t t h e value0 5 6 , i n s p i t e o f t h e foreigna i d proxies,whereasthati n t h e latterfluctuatesbetween0.54a n d 0 5 5 Fou rth,wed o n o t empiricallypreferthenon-lagspecificationbecauseitislessrobustthanthetwo- lagversion.AsseeninPanelD (Table2 ) , t h e coefficiento f INSFi n d e x becomesinsignifican tafterutilcomi sremovedfromthemodel.
Next,wet u r n o u r attentiont o t h e robustnessc h e c k s First,weremovesomec o n t r o l v ariablesfromt h e model,whicha r eutilcomi nt h e FDIe q u a t i o n andl n p o p i nt h e a i d equations.TheresultsarereportedinPanelB,G,andHofTable2.Inthecaseofutilcom,allo f thecont rolvariableswhichmightshareitsroleintheFDIequation—
GDPgrowth,tradeo p e n n e s s , andthelagofFDIinflows— increasesignificantlyintheirimpacts.Incontrast,thecoefficientsofourvariablesofinterest— INSFindexandforeignaid— aredownwardbiased.ThecoefficientofINSFindexbecomesunstable.Nevertheless,theirimpacts onFDIinflowsremainsignificant,andtheimpactofmultilateralaidstilldoublesthatofbilateralaid. Therei s notmuchchange inthethree aidequations withthis omissionofutilcom.Inthecas eofl n p o p,theabsenceofthisdemographicvariablejustinducestheroleofcontrolofcorruptiona n d political stabilityinhavingmoreforeignaid.
In this analysis, we incorporate additional variables into the base specification of the ADB model, specifically adding the money stock M2 to the FDI equation and including the inflation rate in the aid equations We also consider the young and old dependency ratios of the working-age population The results, presented in Panels E and F, indicate that while domestic financial development, inflation, and the younger dependents are largely irrelevant, the old dependency ratio shows some indirect relevance This ratio can reflect human capital, living conditions, and future debt repayment capacity On one hand, favorable human capital and living conditions attract foreign investment, while on the other hand, the limited ability of the older population to work and repay debt may raise concerns for bilateral donors The regression analysis reveals significant coefficients for the old dependency ratio in both the FDI and bilateral aid equations.
Incolumn( 1 4 ) , e v e n thought h e coefficientoff r e e d o m i n t h e bilateralaidequationr e d u c e s inthesize,from-0.15to-
0.08,comparedtothebaseregressionincolumn(2), itssignificancelevelalsodecreases.Thus,the roleoffreedomcouldbeplausibleinbilateralaiddisbursements.SuchsituationsalsooccurwithINSFi ndexandcommunicationutilitiesintheFDIequation[PanelF,column(16)] 7For othercasesinPanelEandF,thecoefficientsandtheirsignificancelevelsarealmostunchanged.
7 The coefficients(t-statistics)ofcommunicationutilitiesare.009(2.67)incolumn(1)and.005(1.38) incolumn(16).
Note:3SLSestimation,t- statisticsinparentheses,***p