Foreign Threat and Economic Growth*

Transcription

1 Foreign Threa and Economic Growh* Hideaki Murase**, Hideki Toya***, and Di Zhao**** Sepember 2010 Absrac This paper offers a heoreical model and empirical evidence on he effecs of hreas of foreign invasion on he hreaened counry s economic growh. In he model, he governmen conrolled by he elie who hold an encompassing power has an incenive o be developmenal. However, developmen increases he masses abiliy o cones power and hence he incenive is consrained by he elie s fear of losing power. Under his circumsance, hreas of foreign invasion may relax he consrain and allow he governmen o be more developmenal since hey may decrease he masses willingness o cones power, i.e., foreign counries can play a role of he common enemy, which enhances he cooperaion beween he elie and he masses. Our empirical evidence validaes his predicion of he model: an increase in hreas of foreign invasion significanly increases he growh rae of he hreaened counry hrough an increase in is human capial and TFP. Specifically, i significanly affecs he composiion of governmen expendiures, i.e., i increases educaion and invesmen expendiures bu decreases consumpion expendiure, wih he oal size of governmen expendiures unchanged. Keywords: Economic growh; Invasion; Revoluion; Endogenous policy JEL Classificaion: D72; D74; D78; O10; O57 * We hank Kazumi Asako, Juro Teranishi, Masaya Sakuragawa, Moosugu Fukushige, Masayuski Oaki, Moriki Hosoe, seminar paricipans a he Insiue of Economic Research of Kyoo Universiy, Keio Universiy, Nihon Universiy and Nagoya Gakuin Universiy, and paricipans a he 2010 Spring Meeing of Japan Associaion for Applied Economics for heir helpful commens. ** Faculy of Economics, Nagoya Ciy Universiy, Yamanohaa Mizuho-cho Mizuho-ku Nagoya, , Japan. Tel&Fax: ; murase@econ.nagoya-cu.ac.jp *** Faculy of Economics, Nagoya Ciy Universiy **** Graduae School of Economics, Nagoya Ciy Universiy

2 1. Inroducion To explain he cross-counry difference in economic growh, he recen lieraure has emphasized he role of governmen policies played in fosering or hindering developmen. However, if policies play a crucial role in deermining each counry s growh performance, a deeper quesion is: why all counries do no adop policies good for developmen. The simples answer is ha, in poor counries, non-democraic governmens pursue inefficien policies o maximize he welfare of he dicaor, auocras, or more broadly ruling elies, which differ from he efficien policies o maximize naional produc or social welfare. While his answer seems plausible, he raional choice heory predics he opposie, i.e., since non-democraic regimes conrolled by he poliically powerful elie would be a self-serving sysem, he elie could have a srong incenive o promoe developmen and o exrac sufficien fruis from he resuling prosperiy. Specifically, if he elie have a large enough encompassing ineres in he economy, hey may be srongly moivaed o adop he efficien policies (see Olson, 1982; 1993 and McGuire and Olson, 1996). 1 The recen research has aemped o fill his gap beween heory and realiy. Acemoglu and Robinson (2000, 2006) and Robinson (2001) develop a heory o explain why he elie canno exrac enough resources from he res of he economy and, as a resul, fail o adop he efficien policies (see also Robinson, 1998). They argue ha developmen ends o be inconsisen wih he mainenance of he poliical saus quo, and hereby he elie s incenive o promoe developmen is consrained by heir fear of losing poliical power. This poliical consrain can be hough of as a paricular form of conracual incompleeness in poliics, which may deer he adopion of he efficien policies by he governmen (see Norh, 1981). Specifically, due o his poliical consrain, he elie may confron he rade-off beween an increase in economic ren and a decrease in poliical power and, if he elie face a igher consrain, hey are more prone o adop inefficien policies in 1 This incenive of non-democraic governmens is ofen hough of as analogous o ha of firm owners. Tha is, as he residual claimans on naional produc, hey do no wish o kill he goose ha laid he golden egg. 1

3 order o mainain heir curren poliical power. 2 This paper focuses on hreas of foreign invasion as a criical poliical facor which may affec he ighness of he poliical consrain in non-democraic counries. Specifically, we offer a simple model o discuss how his geopoliical facor can influence he governmen s incenive for developmen, policies, and economic growh. The crucial poin is ha, in he conex of he conflic beween he elie and he masses, an addiional conflic of foreign invasion may decrease he masses willingness o cones power and hence relax he poliical consrain on he elie. This may occur because foreign counries play a role of he common enemy leading o a more congruence of ineres beween he elie and he masses in he domesic counry. Specifically, domesic poliical sabiliy may help he vicory in an inernaional conflic and hence benefi no only he elie bu he masses as well. As a resul, under an increase in hreas of foreign invasion, he masses more resrain hemselves from conesing power and he elie can pursue more developmen, alhough developmen may increase he masses abiliy o cones power. We also perform empirical research ino his relaion beween foreign hrea and economic growh, using a newly consruced index of each counry s poliical insabiliy. Our index of poliical insabiliy is he weighed sum of number of inernaional conflics ha foreign counries have engaged in (he weigh is given by he inverse of he domesic counry s disance from each inernaional conflic). We use his index as a proxy for he amoun of hreas of foreign invasion perceived by he domesic counry, assuming ha he domesic counry is more likely o be hreaened by 2 This poliical consrain can also be hough of as one of he causes for he failure of he poliical Coase heorem ha assers he irrelevance of poliical regimes (i.e., who has poliical power) o economic efficiency. For insance, if non-democraic governmens can gain less poliical legiimacy due o he limied people s paricipaion in poliical process han democraic ones, he consrain may be more significan for he former. Hence, according o he degree of consrain ighness, he adoped policies (and hence growh performance) may be more diverse among non-democraic counries han among democraic ones; see, for furher discussion on (he failure of) he poliical Coase heorem, Acemoglu (2003). 2

4 foreign invasion as neighboring counries have more ofen engaged in inernaional conflics. One imporan advanage of our index is is relaive exogeneiy in growh regressions, i.e., he number of inernaional conflics ha foreign counries have engaged in is admiedly exernal o he domesic counry s economic and poliical condiions. Thus, our findings are less subjec o he usual causaliy problem ha poliical insabiliy, policies, and economic growh are joinly deermined variables. Our esimaion resuls srongly suppor he model predicions. Tha is, an increase in hreas of foreign invasion significanly increases he growh rae of he hreaened counry. Our resuls also show ha his occurs hrough an increase in is human capial and TFP. Specifically, an increase in foreign hrea significanly affecs he composiion of governmen expendiures, i.e., i increases he educaion and invesmen expendiures bu decreases he consumpion expendiure, wih he oal size of governmen expendiures unchanged. Furhermore, we find ha hese resuls are paricularly rue for non-democraic counries raher han for democraic ones, alhough he poliical regime iself has no direc relaion o he counry s economic growh or he composiion of governmen expendiures. Hence, our evidence suggess ha wha maers in deermining each counry s growh performance is no he poliical regime iself, bu he poliical consrain which may be more significan for non-democraic regimes facing more subsanial domesic conflics beween he elie and he masses. From he hisorical perspecive, here are several examples of developmen moivaed by hreas of foreign invasion. For example, afer he defea in he Crimean War, Russia iniiaed a large-scale invesmen in infrasrucure in order o modernize is economy, recognizing is vulnerabiliy o foreign hrea. Japan in he 19h cenury, facing he hrea of being colonized by he European powers, spurred rapid indusrializaion. Similarly, Turkey hreaened by foreign invasion afer he decline of he Ooman Empire aemped o indusrialize is economy and moreover o democraize is poliical sysem. The mos prominen examples afer World War II may be Asian developmenal dicaorships such as Taiwan under Chiang Kai Shek and Souh Korea under Park Chung Hee, which faced he hreas of communis invasion. Our empirical evidence may demonsrae ha his phenomenon of developmen moivaed by foreign hrea is no 3

5 limied o ha of hisorical episodes, bu is a general endency prevailing in he modern world. There are several heoreical and empirical sudies closely relaed o ours. On he heoreical side, Acemoglu and Robinson (2006) and Chaudhry and Garner (2006) presen models which relae an increase in hreas of inernaional conflic o a higher economic growh of he hreaened counry. Their model formulaion is based on he noion ha a more producive counry is more advanageous in an inernaional conflic. Thus, in heir models, facing foreign hrea, he elie pursue developmen in order o hold a dominan posiion in an inernaional conflic, while aking a risk of an escalaing domesic conflic. Our model formulaion is based on a slighly differen noion. Facing foreign hrea, he masses adop a more cooperaive aiude oward he regime since hey share common ineress wih he elie in an inernaional conflic. Hence, he elie are allowed o pursue developmen, facing a relaively low risk of an escalaing domesic conflic. Our formulaion parially reflecs he experience of Japan in he 19h cenury. Tha is, afer he shor-erm civil war beween compeing miliary oligarchies, he Japanese auocracy had been quie sable unil is defea in World War II. This sabiliy migh be produced parly because he Japanese auocracy successfully esablished a srong cenralized governmen in he early phase of indusrializaion, bu a he same ime he sabiliy migh be mainained because foreign hrea harmonized he poenially conflicing ineress of he elie and he masses in he Japanese auocracy. In conras o his Japanese case, in Russia, rapid indusrializaion acually desabilized he elie s poliical power and he regime was finally overhrown by he socialis revoluion. Thus, he Russian case seems o be beer capured by he noion of Acemoglu and Robinson (2006) and Chaudhry and Garner (2006), while he Japanese case seems o be beer described by our noion. In realiy, boh mechanisms may work, perhaps simulaneously. Hence, we believe ha heir models and ours are no muually exclusive bu complemenary. On he empirical side, using he number of poliical assassinaions, revoluions and coups as he index of poliical insabiliy, Barro (1991) demonsraes ha each counry s growh performance negaively depends on hese poliical insabiliy variables. Alesina, Ozler, Roubini, and Swagel (1996), using oher variables of poliical insabiliy, reach he same 4

6 conclusion. Benhabib and Spiegel (1992) and Alesina and Peroi (1993) find evidence ha poliical insabiliy impedes invesmen. Among ohers, Ades and Chua (1997) is closes o our paper in ha hey focus on he inernaional spillovers of effecs of poliical insabiliy. They provide evidence ha poliical insabiliy in neighboring counries (measured by he number of revoluions and coups) has a negaive impac on he growh performance of a domesic counry. A firs glance, he resul of Ades and Chua (1997) which discovers negaive spillovers of regional insabiliy conrass wih our resul. However, i is consisen wih our model predicion, i.e., poliical insabiliy in neighboring counries reduces heir naional power and miigaes foreign hrea upon he domesic counry, which in urn hinders he developmen incenive of he domesic governmen. The resul of Easerly and Levine (1998) ha he poor economic performance of neighboring counries negaively affecs he income level of a domesic counry could be reinerpreed in a similar vein. The res of he paper is organized as follows. The nex secion offers he heoreical model which discusses a possible mechanism in which an increase in hreas of foreign invasion can lead o a higher economic growh of he hreaened counry. Secion 3 explains he daa used in he empirical analysis. Secion 4 presens empirical evidence ha validaes he predicion of he model. Secion 5 concludes he paper. 2. The model This secion presens a simple model o illusrae a possible mechanism relaing an increase in hreas of foreign invasion o a higher growh of he hreaened counry. To explain he basic srucure of he model, we firs consider he counry under no hreas of foreign invasion. Then, we add he possibiliy of foreign invasion o he model and discuss how i affecs he hreaened counry s growh. 5

7 2.1. No hreas of foreign invasion Populaion Consider an infinie horizon economy in discree ime. The economy consiss of he masses and he ruling elie. The poliical regime is non-democraic, in which he governmen is conrolled by he ruling elie. Each agen lives for one period and bears one child a he ime of his/her deah. Hence, generaions do no overlap and he size of populaion remains consan over ime. The membership of wo groups, he masses and he elie, is assumed o be exogenous, i.e., he children of he masses (he elie) become he masses (he elie). Furher, wihou loss of generaliy, he size of agens in each group is normalized o 1. Preference The elie have preference over consumpion and beques. The preference of each agen belonging o he elie group is given by: 1 U U c w c w (1 ) e e e e e 1 e (, ) ( ) ( ) 1, (1) where e c is her consumpion, e w is her beques, and is a consan wih 0 1 (he subscrip denoes ime hroughou he paper). Meanwhile, he masses have preference over consumpion. The preference of each agen belonging o he mass group is given by: m m m m U U ( c ) c, (2) where m c is his consumpion. 3 Noe ha he preferences given by (1) and 3 In our model, wha maers is ha he beques moive of he elie is sronger han ha of he masses. Hence, we could alernaively formulae m m m m 1 m 1 m he masses preference as U U ( c, w ) ( c ) ( ) 1 w, where (1 ) m w is he masses beques and is a consan wih 0. However, 6

8 (2) are linear (risk-neural) in erms of income; specifically, he expeced uiliy of each agen coincides wih his/her expeced income. Producion The economy produces a single good using he masses labor. Specifically, he good is produced by he following linear echnology: Y A( G ) L m, (3) where Y is he produced good, m L is he masses labor devoed o he good producion, and AG ( ) is he produciviy of he masses labor. G is he governmen invesmen in he public infrasrucure and/or public educaion, which increase he produciviy of he masses labor. In he below, we simply call G as he invesmen and assume ha AG ( ) AG, where A is a posiive consan. We also assume ha he elie do no engage in he good producion, bu only engage in he governmen aciviy, especially he mainenance of he exising order such as policing. Governmen policies In he exising order, poliical power o decide public policies is monopolized by he elie. However, he masses have a poenial o aemp a revoluion, he success of which allows he masses o ake poliical power from he elie. There are wo public policies: invesmen policy and axaion policy. The invesmen policy deermines he produciviy of he masses labor. Specifically, a he beginning of each period, he governmen implemens he invesmen policy ha allocaes he fracion of he elie s inheried beques o he invesmen. We denoe his fracion by (0 1). The remaining fracion of he elie s inheried beques is used for he governmen aciviy ha he elie engage in, and deermines is produciviy. In he below, we simply call he produciviy of he governmen aciviy as he produciviy of he elie and denoe i by D. Given his invesmen policy, he produciviy his formulaion would complicae he analysis bu add lile insigh. 7

9 e of he masses and ha of he elie are assumed o be given by G w 1 and D (1 ) w respecively. 4 The axaion policy, on he oher hand, e 1 deermines he allocaion of he produced good beween he masses and he elie. Specifically, afer implemening he invesmen policy, he governmen decides he ax rae imposed on he produced good. We assume ha his ax rae denoed by * is applied only in he case where no revoluion is aemped; he ax rae in he case where a revoluion occurs will be discussed below. Leing (0 1) be he ax rae, we obain he income of he masses and ha of he elie as (1 ) Y and Y respecively. This formulaion of he axaion policy implies ha he income of he elie is a pure ren and parameerizes he elie s encompassing power. Revoluion (Domesic conflic) Now consider a revoluion, which is he masses aciviy wih he aim of aking poliical power from he elie. If a revoluion is iniiaed, i succeeds wih probabiliy p and fails wih probabiliy 1 p ( 0 p 1 ). The governmen invesmen in public infrasrucure and/or public educaion increases he masses abiliy o cones power and desabilizes he elie s poliical power. To capure his noion, we assume ha he larger he produciviy of he masses relaive o ha of he elie, he larger he success probabiliy of revoluion. Specifically, he success probabiliy akes a form of he cones success funcion inroduced by Tullock (1967, 1980) for he analysis of ren-seeking coness: 5 p G G D. (4) If a revoluion is iniiaed, i absorbs human resources. Specifically, i absorbs he fracion (1 ) of labor of he masses and ha of elie. Hence, 4 For simpliciy, G and a he end of each period. D las for one period and compleely depreciae 5 See also Grossman (1991, 1994) and Skaperdas (1992) for he cones success funcions used in conflic lieraure. 8

10 he masses labor devoed o good producion decreases from L 1 o m m L when a revoluion is iniiaed. For he convenience of he analysis below, we assume ha Furher, we assume ha he iniial axaion policy decided by he governmen is abolished if a revoluion is aemped. Wihou loss of generaliy, we assume a-winner-akes-all siuaion: when a revoluion succeeds, he masses remove he elie from poliical power and se he lowes possible ax rae, i.e., 0 ; when i fails, he elie se he highes possible ax rae, i.e., 1. Given hese ax raes, he income (he uiliy) of each group is summarized in he game-ree depiced in Figure 1. Economic growh We firs examine he wihin-a-period equilibrium by working backward and hen compue he equilibrium growh rae. Recalling ha he preference of each agen is linear in erms of his /her income, we obain he masses expeced uiliy as: U m (1 * ) AG if 0, (5) pag if 1 where is he index variable ha akes on 1 if he masses aemp a revoluion; 0 oherwise. The masses aemp a revoluion when hey find i profiable. Thus, from (5), he index variable is deermined as: 0 1 if if 1 p * 1 p *. (6) Eq. (6) can be hough of as a reacion funcion ha describes he masses poliical response o he ax rae ha he governmen decides. Nex, he expeced uiliy of he elie is given by: 9

11 U e * AG if 0 (1 p) AG if 1. (7) The governmen decides he ax rae in order o maximize he elie s expeced uiliy, aking ino accoun he masses reacion funcion. From (6) and (7), we can obain he soluion as: * 1 p and 0, i.e., he governmen finds i opimal o se he highes possible ax rae ha does no rigger a revoluion. Noe ha his soluion * 1 p exacly corresponds o he poliical consrain on he elie, which means ha he higher he masses abiliy o cones power, he lower he ax rae. Hence, subjec o his poliical consrain, he governmen decides he invesmen policy o maximize he elie s expeced uiliy: e e max U * AG subjec o * 1 and G w1. (8) {, * } 1 1 This maximizaion problem yields and * (Noe ha since ). Hence, noing ha e w 1 * 1 AG 1 and ha * is ime-invarian, we obain he equilibrium growh rae of he counry as: A g 1. 4 I is noable ha he efficien invesmen policy under no hreas of revoluion is obviously given by 1 and hence he conflic leads o he underinvesmen problem. Specifically, he maximizaion problem of (8) uncovers he rade-off ha he governmen faces, which sems from he poliical consrain. Tha is, he governmen, on he one hand, wishes o promoe developmen since i enlarges he ax base, i.e., he elie s encompassing ineres. However, he governmen, on he oher hand, wishes o impede developmen since i diminishes he ax rae, i.e., he elie s encompassing power. The equilibrium growh rae is deermined by his rade-off beween he encompassing ineres and he encompassing power Threas of foreign invasion 10

12 Foreign invasion (Inernaional conflic) Now we consider he counry under hreas of foreign invasion. In his case, he counry may experience one of he following four possible saes associaed wih conflics: (No revoluion, No invasion), (No revoluion, Invasion), (Revoluion, No invasion), and (Revoluion, Invasion); see he game-ree in Figure 2. We assume ha a foreign invasion occurs wih an exogenous probabiliy (0 1). If an inernaional conflic is iniiaed, he domesic counry defeas he foreign counry wih probabiliy Q in he case of (No revoluion, Invasion) and wih probabiliy q in he case of (Revoluion, Invasion), where 0 Q 1 and 0 q 1. We assume ha Q q, which capures he noion ha domesic poliical sabiliy helps he vicory in an inernaional conflic. Specifically, since domesic conflic decays he human resources of he domesic counry devoed o an inernaional conflic, is naional power relaive o he foreign counerpar is reduced. 6 We assume ha an inernaional conflic absorbs human resources as a revoluion does. Specifically, i absorbs he fracion (1 ) of labor of he domesic counry ( 0 1). Hence, he masses labor devoed o he good producion decreases from L 1 o L m m in he case of no revoluion and from m L o L m in he case of revoluion. Furher, he iniial axaion policy decided by he domesic governmen is abolished if he domesic counry loses an inernaional conflic. We again assume a-winner-akes-all siuaion: when he domesic counry loses, he foreign counry removes he domesic elie from poliical power and ses he highes possible ax rae, i.e., 1. By conras, when he domesic counry defeas he foreign counry, he iniial ax rae is preserved in he case of no 6 For example, given he fracion of labor of he domesic elie and masses devoed o he revoluion (1 ), he winning probabiliies may be given by (1 )( G D) 1 he following cones success funcions: Q (1 ) V (1 )( G D) 1 and (1 )( GD) q (1 ) V(1 )( GD), where V is he naional power of he foreign counry, G D is he domesic counerpar, is heir relaive size, and (1 ) is he fracion of he naional power devoed o an inernaional conflic. Thus, Q q. 11

13 revoluion, i.e., *, while 0 in he case of successful revoluion and 1 in he case of failed revoluion. Hence, given hese ax raes, he uiliy of each group is summarized as in he game-ree in Figure 2. Economic growh Le us urn o he equilibrium growh rae in he presence of foreign hrea. Firs, he masses expeced uiliy is given by: U m ( Q1 )(1 * ) AG if 0 ( q1 ) pag if 1, (9) Then, he index variable ha indicaes wheher a revoluion is aemped or no can be obained as: 0 1 if if 1 (, ; Qq, ) p* 1 (, ; Qq, ) p*, (10) q 1 where (, ; Qq, ) ; noe ha 1 Q 1 and 0. Meanwhile, he expeced uiliy of he elie is given by: U g ( Q1 ) * AG if 0 ( q1 )(1 p) AG if 1. (11) As in he case of no hreas of foreign invasion, he governmen decides he ax rae in order o maximize he elie s expeced uiliy, aking ino accoun he masses reacion funcion (10). Again, he opimal ax rae is obained as he highes possible one ha does no rigger a revoluion, i.e., * 1 p (, ; Qq, ) and 0. Noe ha, compared wih he case of no hreas of foreign invasion, he poliical consrain may be more relaxed in he presence of foreign hrea since 1. This is because he defea in an inernaional conflic harms no only he elie bu he masses as well and he masses more resrain hemselves from aemping a revoluion. Hence, in he presence of foreign hrea, he governmen can decide he invesmen 12

14 policy subjec o a more relaxed poliical consrain, i.e., g max U ( Q1 ) * AG {, * } e subjec o * 1 (, ;, ) and G w1. (12) Qq This maximizaion problem yields 1 min[,1] 2 (, ; Qq, ) 1 * max[,1 (, ; Qq, )]. Since 0, as foreign hrea increases, he 2 invesmen by he governmen increases and approaches he socially efficien level, 1 (and, afer reaching ha level, remains here). 7 Noing e ha w 1 * 1 AG 1 and ha * is ime-invarian, we obain he economic growh rae of he hreaened counry as: and g A 1 if (, ; Q, q) 4 (, ; Qq, ) 2 1 A{1 (, ; Q, q)} if (, ; Q, q) 2. (13) Hence, he heory predics ha an increase in hreas of foreign invasion increases he rae of economic growh of he hreaened counry hrough he changes in he governmen incenive for developmen. Specifically, i saes ha foreign hrea changes he composiion of governmen expendiures of he hreaened counry in he direcion of more invesmen in public infrasrucure and /or public educaion. The inuiion behind he resuls is sraighforward. On he one hand, as foreign hrea increases, he masses willingness o cones power decreases since domesic insabiliy makes he vicory in an inernaional conflic more unlikely and hereby harms he masses hemselves. On he 7 I is also noeworhy ha, since 0, an inernaional conflic is more damaging o he domesic economy, he counry makes a more efficien level of invesmen. 13

15 oher hand, his decrease in he willingness o cones power may relax he poliical consrain on he elie, and he governmen can make more invesmen, alhough i increases he masses abiliy o cones power. In oher words, foreign counries can play a role of he common enemy o alleviae he poenial conflic of ineress beween he elie and he masses of he hreaened counry and enhance heir cooperaion in favor of economic growh. 3. The daa To examine wheher or no our model predicions are consisen wih he acual daa, we perform growh regressions using he cross-counry daa for he period Appendix A provides he lis of he counries used as he sample. We perform he cross-counry regressions wih special focus on he geopoliical condiion ha each counry faces. For his purpose, we consruc wo new variables represening poliical insabiliy of each counry. The firs variable is he index of hreas of foreign invasion (hereafer called as TFI) which is given by: TFI i Number of inernaional conflics Counry j has engaged in log ji Disance from Counry i o Counry j We use his index as a proxy for he poenial foreign hrea for Counry i. Specifically, we assume ha Counry i is more likely o be hreaened by foreign invasion as neighboring counries have more ofen engaged in inernaional conflics. Noe also ha he index is consruced as he sum of number of oher counries inernaional conflics weighed by each counry s disance from hose conflics. Thus, we are here focusing on he relaive geography variable which specifies each counry s siuaion on he globe 14

16 vis-a-vis oher counries. 8 The second variable is he index of domesic and inernaional conflics which each counry has acually engaged in (hereafer called as WAR). This index is given by: WARi log(number of domesic and inernaional conflics Counry i has engaged in) We use his index o conrol for he effecs of domesic and inernaional conflics which Counry i has engaged in on he counry iself. Hence, wha he wo insabiliy variables indicae are clearly separaed, i.e., TFI i capures he poenial hreas of conflics o Counry i, while WAR i capures he oubreaks of conflics and he resulan physical and human damages o Counry i. In our growh regressions, we also conrol for a lo of oher variables considered as imporan deerminans of economic growh in previous research. These variables are consruced from several differen daa sources. Appendix B summarizes he lis of variables and he daa source associaed wih each variable. Appendix C provides he summary saisics of all variables. 4. Empirical evidence This secion presens our empirical resuls concerning he effecs of hreas of foreign invasion and oher variables on per capia GDP growh and he composiion of governmen expendiures. We esimae all he regressions using an ordinary leas squares procedure wih he Whie s 8 By conras, he absolue geography, which is independen of he locaion and siuaion of oher counries, refers o purely physical geography such as each counry s own climae and is own access o he sea; see Krugman (1993), Redding and Venables (2004) and Bosker and Garresen (2008) for he disincion beween he absolue geography and he relaive geography. 15

17 heeroskedasiciy-consisen covariance marix. Table 1 shows he basic regression resuls concerning he effecs of hreas of foreign invasion and oher variables on per capia GDP growh. Column 1 repors he benchmark regression including TFI (he index of hreas of foreign invasion) and conrolling for iniial GDP per capia and iniial oal schooling years. The regression resul indicaes ha he coefficien of TFI is posiive and highly significan, while ha of WAR is negaive and is magniude is relaively small (Figure 3 depics he parial relaionship beween TFI and per capia GDP growh obained in Column 1). In Column 2, we conrol for oher variables which previous research has recognized as imporan deerminans of economic growh, i.e., populaion growh, invesmen, and openness. The resul is mosly unchanged from Column 1, i.e., he coefficien of TFI remains significanly posiive, while ha of WAR is negaive and becomes insignifican. Furher, i migh be suspeced ha FTI and WAR merely capure he effecs of poliical regimes and hereby affec economic growh. To examine his poin, we include he index of poliical regimes, Auocracy, of Glaezer e al. (2004) in Column 3. We again obain a similar regression resul associaed wih TFI and WAR. In addiion, we find ha he coefficien of Auocracy is negaive bu insignifican, which implies ha here is no sysemaic difference in growh performance beween non-democraic counries and democraic ones. I is widely recognized ha he economic growh of each counry is driven by is physical capial accumulaion, human capial accumulaion, and produciviy growh. Hence, we nex include he variables associaed wih hese facor accumulaions ino our regressions in order o examine he channels hrough which TFI has a posiive impac on economic growh. Table 2 repors he resuls of he regressions conrolling for he facor accumulaions. Columns 1-3 show ha, when we include physical capial accumulaion, human capial accumulaion, and produciviy growh ino he regressions, he impac of TFI on economic growh capured by is significance and magniude of he coefficiens is weakened. Then, as shown in Column 4, when all hese facor accumulaion variables are included, he impac of FTI becomes insignifican. These resuls srongly indicae ha TFI has a posiive impac on economic growh hrough he channel of enhancing hese facor accumulaions. Given he resuls obained in Table 2, we proceed o examine wheher 16

18 or no TFI acually simulaes facor accumulaions. Table 3 repors he regression resuls which indicae ha TFI is posiively relaed o all he facor accumulaion variables, while WAR is no. These resuls are consisen wih our model predicion ha foreign hrea has a posiive impac on facor accumulaions by simulaing he governmen invesmen in public infrasrucure and/or public educaion. In order o more closely inspec his poin, we perform he regressions concerning he effecs of foreign herea on governmen expendiures. Table 4 shows ha TFI has a posiive and significan impac on he governmen educaion and invesmen expendiures, while a negaive and significan impac on he governmen consumpion expendiure. Furhermore, i shows ha TFI has no significan impac on he governmen size. These resuls are consisen wih our model predicion, i.e., confroning wih an increase in foreign hrea, he composiion of he governmen expendiures is changed in he direcion of more invesmen and less consumpion. Furher, in all columns, we find ha he coefficiens on Auocracy are insignifican, which implies ha poliical regime has no sysemaic effecs on he composiion of governmen expendiures. Tables 1 and 4 demonsrae ha poliical regime iself has no direc impac on economic growh and he composiion of governmen expendiures. However, he poliical consrain may be more significan for non-democraic counries han for democraic ones since he former can be hough of as facing more subsanial domesic conflics beween he elie and he masses. Hence, he effecs of foreign hrea on economic growh may larger for non-democraic counries han for democraic ones. In order o examine wheher or no such a difference across poliical regimes exiss, we divide he sample counries ino wo sub-samples, i.e., he auocracy sample and he democracy sample. Table 5 repors he regression resuls for boh sub-samples. They show ha he effecs of FTI on economic growh, facor accumulaions, and governmen expendiures are all larger for he auocracy sample han for he democracy sample, boh in heir significance and magniude. This resul can be inerpreed as evidence ha wha maers in deermining he governmen incenive for developmen is no a poliical regime iself, bu he poliical consrain ha each governmen faces. Tha is, foreign hrea may more relax he poliical consrain in non-democraic counries which face more subsanial domesic poliical conflics han 17

19 democraic ones. 5. Conclusion This paper has offered a heoreical model and empirical evidence on he effecs of hreas of foreign invasion on he hreaened counry s economic growh. In he model, we show ha in non-democraic counries foreign hrea may decrease he masses willingness o cones power, relax he poliical consrain on he poliical elie, and allows he governmen o pursue more developmen alhough i increases he masses abiliy o cones power. Our empirical resuls validae his predicion of he model. We find a significanly posiive influence of hreas of foreign invasion on he growh rae of he hreaened counry. We also find ha his increase in he growh rae occurs hrough he increase in he hreaened counry s human capial and TFP. Specifically, foreign hrea significanly increases he governmen educaion and invesmen expendiures bu decreases he governmen consumpion expendiure, while he oal size of expendiures remains unchanged. Furhermore, hese resuls are paricularly rue for non-democraic counries raher han for democraic ones. Our heoreical and empirical resuls sugges ha each counry s economic growh is no an economic even isolaed from he res of he world, bu i is grealy influenced by poliical evens aking place in he res of he world. Specifically, our resuls indicae ha he geopoliics surrounding each counry, ogeher wih is own domesic poliics, may be one of he deep deerminans of he counry s policies and growh performance. 18

20 References Acemoglu, D. (2003) Why no a Poliical Coase Theorem? Social Conflic, Commimen, and Poliics, Journal of Comparaive Economics 31: Acemoglu, D., Robinson, J. A. (2000) Poliical Losers as a Barrier o Economic Developmen, American Economic Review 90: Acemoglu, D., Robinson, J. A. (2006) Economic Backwardness in Poliical Perspecive, American Poliical Science Review 100: Ades, A., Chua, H. B. (1997) Thy Neighbor s Curse: Regional Insabiliy and Economic Growh, Journal of Economic Growh 2: Barro, R. J. (1991) Economic Growh in a Cross Secion of Counries Quarerly Journal of Economics 106: Barro, R. J., Lee, J. W. (1994) Daa Se for a Panel of 138 Counries. NBER. Available online a: hp:// Barro, R. J. Lee. J. W. (2000) Inernaional Daa on Educaional Aainmen: Updaes Implicaions, CID Working Paper, No. 42. Benhabib, J. Spiegel, M. M. (2002) Human Capial and Technology Diffusion, FRBSF Working Paper, 2. Bosker, M., Garresen, H. (2009) Economic Developmen and he Geography of Insiuions, Journal of Economic Geography 9: Chaudhry, A., Phillip, G. (2006) Poliical Compeiion beween Counries and Economic Growh, Review of Developmen Economics 10: CEPII Geographical Disances Daa Se. Available online a: hp:// Easerly, W., Levine, R. (1998) Troubles wih he Neighbors: Africa s Problem, Africa s Opporuniy, Journal of African Economics 7: Easerly, W., Levine, R. (2001) I s no Facor Accumulaion: Sylized Facs and Growh Models, World Bank Economic Review 15: Lapora G. E., Lopes-de-Silanes R., Shleifer, A. (2004) Do Insiuions Cause Growh?, Journal of Economic Growh 9: Grossman, H. I. (1991) A General Equilibrium Theory of Insurrecions, American Economic Review 81: Grossman, H. I. (1993) Producion, Appropriaion and Land Reform, American Economic Review 84: Krugman, P. (1993) Firs Naure, Second Naure, and Meropolian Locaion, Journal of Regional Economics 33:

21 McGuire, M. C., Olson, M. (1996) The Economics of Auocracy and Majoriy Rule, Journal of Economic Lieraure 34: Norh, D. (1981) Srucure and Change in Economic Hisory. Noron: New York. Olson, M. (1982) The Rise and Decline of Naions. Yale Universiy Press: New Haven, CT. Olson, M. (1993) Dicaorship, Democracy and Developmen, American Poliical Science Review 87: Redding, S., Venables, A. J. (2004) Economic Geography and Inernaional Inequaliy, Journal of Inernaional Economics 62: Robinson, J. A. (2001) When is a Sae Predaory?, Unpublished, Deparmen of Governmen, Harvard Universiy Robinson, J. A. (1998) Theories of Bad Policy, Journal of Policy Reform 2:1-46. Skaperdas, S. (1992) Cooperaion, Conflic, and Power in he Absence of Propery Righs, American Economic Review 82: Tullock, G. (1967) The Welfare Coss of Tariffs, Monopolies and Thef, Wesern Economic Journal 5: Tullock, G. (1980) Efficien Ren Seeking, In: Buchanan, J.M., Tollison, R.D., Tullock, G. (eds.) Toward a Theory of he Ren-seeking Sociey. Texas A &M Universiy Press: College Saion, pp UCDP/PRIO Armed Conflic daa se (Version4-2008). Available online a: hp:// 20

22 Figure 1. Payoffs under No Threas of Foreign Invasion * AG,(1 * ) AG No revoluion Invesmen policy * Taxaion policy Fail AG,0 Revoluion Succeed 0, AG Noe : The firs and second variables in each parenhesis denoe he elie s uiliy and he masses uiliy respecively. 21

23 Figure 2. Payoffs under Threas of Foreign Invasion No revoluion No invasion * AG,(1 * ) AG Win * AG,(1 * ) AG No revoluion Invasion Lose 0, 0 Invesmen policy * Taxaion policy Fail AG,0 Revoluion No invasion Succeed Fail Win 0, AG AG,0 Revoluion Invasion Fail Lose Succeed Win 0, 0 0, AG Succeed Lose 0, 0 Noe: The firs and second variables in each parenhesis denoe he elie s uiliy and he masses uiliy respecively. 22

24 Figure 3. The Parial Relaion of Threas of Foreign Invasion and Economic Growh 23

25 Table 1: Growh and Threas of Foreign Invasion Dependen Variable: Per capia GDP Growh Iniial GDP per capia (-2.953) (-3.517) (-3.509) Iniial schooling years (6.476) (3.535) (3.393) Populaion growh (-2.300) (-1.935) Invesmen (4.057) Openness (-1.417) (3.839) (-1.378) Auocracy (-1.378) TFI (6.331) (4.853) (4.860) WAR (-2.450) (-1.566) (-1.623) Adjused R-squared Numbers of Observaions Noe: All regressions use OLS mehod wih heeroskedasiciy-consisen sandard errors. Numbers in parenheses are values. A consan erm is included bu no repored here. 24

26 Table 2: Growh, Facor Accumulaion, and Threas of Foreign Invasion Dependen Variable: Per capia GDP Growh Iniial GDP per capia (-1.057) (-3.381) (-1.501) (-0.419) Iniial schooling years (3.421) (6.384) (1.353) (0.821) Growh in physical capial (6.660) (7.068) Change in schooling years (2.934) Growh in TFP (11.56) TFI (3.012) (5.312) (4.166) WAR (-2.520) (-2.224) (-3.837) (-0.926) (10.57) (1.316) (-0.842) Adjused R-squared Numbers of Observaions Noe: All regressions use OLS mehod wih heeroskedasiciy-consisen sandard errors. Numbers in parenheses are values. A consan erm is included bu no repored here. 25

27 Table 3 : Facor Accumulaion and Threas of Foreign Invasion Dependen Variable Growh in physical capial Iniial GDP per capia (-3.126) Iniial schooling years (4.249) TFI (4.354) WAR (-1.162) Change in schooling years (1.565) (1.226) (2.767) (-1.431) Growh in TFP (-2.077) (5.160) (3.860) (-0.388) Adjused R-squared Numbers of Observaions Noe: All regressions use OLS mehod wih heeroskedasiciy-consisen sandard errors. Numbers in parenheses are values. A consan erm is included bu no repored here. 26

28 Table 4: Governmen Expendiures and Threas of Foreign Invasion Dependen Variable Governmen spending on educaion Iiniial GDP per capia (1.639) Iniial schooling years (0.560) Governmen spending on invesmen (0.234) (1.752) Governmen spending on consumpion (-4.539) (0.485) Toal Governmen spending (-0.303) (0.320) Auocracy (-0.516) (0.373) (-0.647) (1.046) TFI (2.649) WAR (-3.078) (2.204) (-1.407) (-2.214) (-1.124) (0.873) (0.440) Adjused R-squared Numbers of Observaions Noe: All regressions use OLS mehod wih heeroskedasiciy-consisen sandard errors. Numbers in parenheses are values. A consan erm is included bu no repored here. 27

29 Table 5: Poliical Regimes and Threas of Foreign Invasion Auocracy sample Dependen Variable Per capia GDP growh (5.069) Growh in physical capial (4.026) TFI WAR Adjused R-squared (-1.776) (-0.896) Numbers of Observaions Change in schooling years (3.824) Growh in TFP (3.595) Governmen spending on educaion Governmen spending on invesmen Governmen spending on consumpion (2.462) (3.413) (-2.976) Toal Governmen spending (0.265) (-1.447) (-0.432) (-2.837) (-1.946) (-0.403) (-0.266) Democracy sample Dependen Variable Per capia GDP growh (2.432) Growh in physical capial (2.696) TFI WAR Adjused R-squared (-2.735) (-1.652) Numbers of Observaions Change in schooling years (0.609) Growh in TFP (1.985) Governmen spending on educaion Governmen spending on invesmen Governmen spending on consumpion (0.977) (1.580) (0.751) Toal Governmen spending (1.449) (-0.612) (-1.598) (-0.983) (-0.907) (-1.218) (1.152) Noe: All regressions use OLS mehod wih heeroskedasiciy-consisen sandard errors. Numbers in parenheses are values. A consan erm, iniial GDP per capia and iniial schooling years are included bu no repored here. 28

30 Appendix A. Lis of Counries Algeria A Honduras A Philippines A Argenina A Iceland B Porugal A Ausralia B India B Senegal A Ausria B Indonesia A Singapore A Bangladesh A Iran A Souh Africa A Barbados B Ireland B Spain A Belgium B Israel B Sri Lanka A Bolivia A Ialy B Swaziland A Boswana A Jamaica B Sweden B Brazil A Japan B Swizerland B Cameroon A Jordan A Syria A Canada B Kenya A Taiwan A Cenral African Rep. A Korea, Republic of A Thailand A Chile A Lesoho A Togo A Colombia B Malawi A Trinidad &Tobago B Congo, Dem. Rep. A Malaysia A Tunisia A Cosa Rica B Mali A Turkey A Denmark B Mala B Uganda A Dominican Republic A Mauriius B Unied Kingdom B Ecuador A Mexico A Unied Saes B El Salvador A Mozambique A Uruguay A Fiji A Neherlands B Venezuela B Finland B New Zealand B Zambia A France B Niger A Zimbabwe A Germany B Norway B Ghana A Pakisan A Greece A Panama A Guaemala A Papua New Guinea B Guyana A Paraguay A Haii A Peru A Noe: A indicaes ha he counry s auocracy index is higher han zero. B indicaes ha he counry s auocracy index is equal o zero. 29

31 Appendix B. Definiion of Variables TFI WAR Variable Variable Definiion Source Per capia GDP growh Growh in physical capial Change in schooling years Growh in TFP Iniial GDP per capia Iniial schooling years Populaion growh Invesmen Openness Auocracy Governmen spending on educaion Governmen spending on invesmen Governmen spending on consumpion Toal governmen spending Logarihm of number of exernal wars in all foreign counries for he period , weighed by disance from he domesic counry o each counry Logarihm of 1+numbers of inernal and exernal wars in he domesic counry for he period Average growh rae of real per capia GDP for he period Average growh rae of physical capial per capia for he period Difference beween average years of schooling in 1990 and average years of schooling in 1960 Average growh rae of oal facor produciviy for he period Logarihm of iniial real per capia GDP in1960 Logarihm of iniial average schooling years in oal populaion over age 25 in1960 Average growh rae of populaion for he period Logarihm of average raio of real domesic invesmen o real GDP for he period Logarihm of average raio of oal rade value o GDP for he period Average index of auocracy for he period Logarihm of average raio of nominal governmen educaion expendiure o nominal GDP for he period Logarihm of average raio of nominal governmen invesmen expendiure o nominal GDP for he period Logarihm of average raio of nominal governmen consumpion expendiure o nominal GDP for he period Logarihm of average governmen share of real GDP per capia for he period UCDP/PRIO & CEPII UCDP/PRIO & CEPII Barro & Lee(1994) Easerly & Levine (2001) Barro & Lee(2000) Benhabib & Spiegel(2003) Barro & Lee(1994) Barro & Lee(2000) Penn World Table 6.3 Barro & Lee(1994) Barro & Lee(1994) Shleifer e al.(2004) Barro & Lee(1994) Barro & Lee(1994) Barro & Lee(1994) Penn World Table

32 Appendix C. Summary of Saisics Variable Mean Sandard Dev. Observaions TFI WAR Per capia GDP growh Growh in physical capial Change in schooling years Growh in TFP Iniial GDP per capia Iniial schooling years Populaion growh Invesmen Openness Auocracy Governmen spending on educaion Governmen spending on invesmen Governmen spending on consumpion Toal governmen spending