POLITICAL STABILITY AND ECONOMIC GROWTH. A TWO WAY RELATION. EDGARDO E. ZABLOTSKY

POLITICAL STABILITY AND ECONOMIC GROWTH. A TWO WAY RELATION. EDGARDO E. ZABLOTSKY FEBRUARY 1996

POLITICAL STABILITY AND ECONOMIC GROWTH. A TWO WAY RELATION. EDGARDO E. ZABLOTSKY 1 FEBRUARY 1996 I. INTRODUCTION. It s usually asserted that poltcal nstablty sgnfcantly lowers prvate nvestment, as well as economc growth, snce t has adverse nfluence on property rghts, and by that on nvestment and growth (.e, Robert Barro, 1991; Ross Levne & Davd Renelt, 1992; Paolo Mauro, 1994; Edgardo Zablotsky, 1994). Poltcal nstablty may lead entrepreneurs to wat untl the uncertanty s resolved, before undertakng rreversble nvestment projects, t also may lead to captal flght; by the same token, multnatonal companes may be less lkely to locate ther subsdares n countres that face the possblty of coups, revolutons, terrorsm, or expropraton. For example, Robert Barro (1991) reports, for a sample 98 countres n the perod 1960-1985, that growth rates are negatvely related to measures of poltcal nstablty. He makes use of varables lke fgures on revolutons, coups, and poltcal assassnatons, snce these relatons could nvolve the adverse effects of poltcal nstablty on property rghts, and therefore on prvate nvestment. Smlarly, Ross Levne and Davd Renelt (1992) conclude that the fgure on revolutons and coups per year s robustly negatve correlated wth the nvestment share of gross domestc 1 Senor Researcher and Charman of the Master n Fnance, CEMA, Córdoba Ave. 637, Buenos Ares, Argentna.

3 product. Thus, not surprsngly, countres that experence a hgh number of revolutons and coups tend to be countres that nvest less of ther resources domestcally than countres wth stable poltcal envronments. By the same token, Paolo Mauro (1994), accounts that any one the dfferent proxes of poltcal stablty that he analyzes s sgnfcantly postvely correlated wth prvate nvestment and economc growth. In ths paper we wll face the problem from a dfferent perspectve; we wll propose that there also exsts a relaton from economc growth to poltcal stablty. In drecton to ths goal we wll center our attenton on mltary coups d'etat. The study of mltary coups d'etat has not receved enough attenton n the publc choce lterature; actually, snce the poneerng works of Downs (1957), Buchanan and Tullock (1962), and Rker (1962) most of the publc choce lterature have been developed under a democratc framework. The frst paper n the publc choce lterature developed under a non-democratc framework was presented by Ireland n 1967. Ths work, as well as the Tullock's (1971) paper, opened a new framework to the study of nondemocratc changes of government. Untl Ireland's and Tullock's works, the study of revolutons was an exclusve feld of poltcal scentsts, who focus ther nterest on the publc good aspect of the revolutons. Snce the appearance of Ireland's and Tullock's works a group of scholars (Letes and Wolf, 1970; Tullock, 1974; Slver, 1974; Cao Garca, 1983; Cartwrght, Delorme and Wood, 1985; etc.) have challenged ths romantc noton of revoluton usng the assumptons and methodology provded by the economc theory. The by-product desgnaton of ths self nterest theory s credted to Tullock (1971), who used the term followng Olson (1965), whose analyss of the motvatons of an agent as an actve partcpant n a collectve acton can be extended to the revolutonary actvty. Whle most of the publc choce lterature n non-democratc changes of government center ther

4 nterest n the so called "mass revolutons" (Ireland, 1967; Letes and Wolf, 1970; Tullock, 1971; Cartwrght, Delorme and Wood, 1985; Kuran, 1989; Grossman, 1991; etc.), most of the actual rregular executve transfers are mltary coups d'etat. To the best of my knowledge, only Tullock (1974), Slver (1974), Cao Garca (1983), Mbaku and Paul (1989), Zablotsky (1992) and Morón (1994) analyze coups d'etat. Of these scholars only Tullock and Zablotsky explctly study mltary coups d'etat by means of a mcroeconomc analyss of benefts and costs. Ths paper provdes further nsghts on the ssue; t s dvded n three sectons. Secton 2 proposes an alternatve defnton of a mltary coup d'etat that characterzes mltary coups d'etat that overthrow democratc regmes better than the usual defntons. We wll show that n order to understand the behavor of the army offcers who face the decson to partcpate n a coup t s nsuffcent to analyze the prvate nterest motvatons that they may have, as t s stated by the by-product theory of revolutons; t s also necessary to analyze the publc good rewards that the hgh rank offcers may consder. Secton 3 s devoted to develop our proposed hypothess, and to provde prelmnary evdence n order to llustrate ts feasblty. II. AN ALTERNATIVE DEFINITION OF A MILITARY COUP D' ETAT. Ths secton s devoted to propose an alternatve defnton of a mltary coup d'etat; ths defnton allows us to understand better the behavor of the army offcers who face the decson to partcpate n a

5 coup that overthrows a democratc regme. It s usually argued that the man dfference between a revoluton and a coup d'etat s that n the former case a sgnfcant proporton of the revolutonares are not members of the government or of the rulng coalton, whle n the latter the members of the plot are part of the government. 2 In actualty, ths defnton fully apples to most, but not every type of coup d'etat; the mltary coups d'etat that overthrow democratc regmes should be consdered an excepton, gven that they are headed by hgh rankng offcers who only supposedly are part of the government. The army offcers are professonals, they are nether elected offcers nor are they part of the governmental coalton; therefore, the usual defnton: n a coup d'etat the members of the plot are part of the government or of the rulng coalton, s nadequate to characterze ths type of rregular executve transfer. In order to characterze adequately ths class of non-democratc change of government I wll ntroduce the followng alternatve defnton: "A mltary coup d'etat that overthrows a democratc regme s characterzed by the fact that ts actors 2 For example, Cao Garca (1983), p. 77, states, "In contrast to revolutons, whch are actvtes organzed by persons outsde the government, a coup d'etat s an attempt of a subset of ths rulng coalton to overthrow from offce the head of a government, together wth a subset of hs supportng coalton, by means of poltcal volence... The basc dfference between a revoluton and a coup d'etat, therefore, s that, whle revolutonary actvtes are made, ex defnto, by ndvduals outsde government, coups d'etat are carred out by government offcals."

6 are supposedly, but not n fact, members of the government." Ths subtle dfference s of great relevance n the understandng of the role played by publc good consderatons on the behavor of the army offcers. For example, Tullock (1974) sustans that publc good consderatons are apt to play as small a part n the decson to partcpate n a coup as n the partcpaton n any outsde revoluton (Tullock, 1974, p. 62); ndeed, t s clear from hs arguments that he bases ths concluson n the tradtonal defnton of a coup. However, t can be shown that under my alternatve defnton t s not possble to reach such a concluson; n order to demonstrate ths pont, we wll explcate Tullock's lne of argumentaton. Tullock studes mltary coups d'etat by means of a mcroeconomc analyss of benefts and costs, analyzng structural factors that affects the partcpaton of the army offcers n the coup. He proposes a framework where, n order to choose hs poston, every army offcer compares the total expected payoff that he would receve f he jons the coup (Pr), f he stays loyal to the government by jonng the represson (Pd), and f he remans neutral (Pn). The army offcer wll jon the coup f: smlarly, he wll jon the represson f: Pr > Pd and Pr > Pn Pd > Pr and Pd > Pn otherwse, he wll choose to reman neutral. Where: Pn = Pg Lv - Np Pr = Pg (Lv + L ) + R (Lv + L ) - P [1 - (Lv + L )] - Lw Ir + E Pd = Pg (Lv - L ) + D [1 - (Lv - L )] - Pp (Lv - L ) - Lw Ir + E

7 and, Pg = Publc good generated by a successful coup. Lv = Lkelhood of a revolutonary vctory f the subject s neutral. Np = Punshment for remanng neutral. L = Change n the probablty of revolutonary success resultng from the subject partcpaton. R = Prvate reward to the subject for hs partcpaton n the mltary coup d'etat f the coup succeeds. P = Prvate penalty mposed on the ndvduals for hs partcpaton n the coup f t fals. Lw = Lkelhood of njury through the partcpaton n support of, or aganst, the coup. Ir = Injury suffered n acton. E = Entertanment value of partcpaton. Slver (1974) defnes ths term as the "psychc ncome from partcpaton," gven that t may nclude a wde varety of factors, lke the ndvdual's sense of duty to the law, race, humanty, the rulers, the revolutonary brotherhood, hs taste for conspracy, etc. D = Prvate reward to the ndvdual for hs partcpaton n puttng down the coup f the government wns. Pp = Prvate cost mposed on the defenders of the government f the coup succeeds. The publc good reward (Pg) wll have a relevant role n the decsons of the agents f, and only f, the change n the probablty of revolutonary success resultng from the partcpaton of the agent (L) s 3 sgnfcantly dfferent from zero. As n a mass revoluton the partcpaton of the subject wll have an 3 Pr - Pn = Pg L + R (Lv + L ) - P [1 - (Lv + L )] - Lw Ir + E + Np

8 nfntesmally small effect over the probablty of success of the acton (L. 0), then we can conclude that publc good consderatons do not play a role n the behavor of the agents; therefore, the subject wll be motvated to partcpate by the expectaton of a prvate return and the publc good reward generated by a successful revoluton must be nterpreted only as a by-product. Gordon Tullock also argues that publc good consderatons are not an mportant factor n explanng coups d'etat; n order to support ths statement he analyzes the followng facts: A. For most of the junor government offcals L wll be close to zero. B. For the hgh rank government offcals whle L wll be sgnfcantly dfferent from zero, the mportant governmental postons that they hold mply that they are bascally satsfed wth the government, such that f publc good consderatons (Pg) have any role at all, t wll be aganst ther partcpaton n the coup. 4 Pd - Pn = Pg L + D [1 - (Lv - L )] - Pp (Lv - L ) - Lw Ir + E + Np then, Pg wll play a role only f L 0. 4 "Note that L, the effect that the ndvdual may have on the coup, s not necessarly nfntesmally small for a government offcal. Under these crcumstances, the expresson Pg L may be more sgnfcant for the government offcal than t s for the prvate ctzen. Ths s dubous, however. Most of the junor government offcals wll stll have very small L's; therefore ths expresson should be close to zero. On the other hand, the senor government offcals, although they wll ndeed have somewhat larger L 's are also lkely to receve very large rewards or very large punshments n the prvate sphere from the success or falure of the coup. Under the crcumstances, t s lkely that for them, too, the publc good aspect of the coup s relatvely mnor. Another feature that must be emphaszed s that the partcpants n the coup or n defense aganst the coup are

The frst argument, whle fully correct, s generally not mportant for explanng mltary coups d'etat; 5 usually, gven the vertcalty of the army, 2 the only relevant behavor that explans a mltary coup d'etat s 6 the behavor of the senor offcers; offcers whose L 's wll be sgnfcantly dfferent from zero. The second Tullock's argument s based upon the fact that the senor offcers are members of the government; so, whle the argument s entrely correct for any coup d'etat that satsfes the tradtonal defnton of a coup, t s no longer satsfactory for mltary coups d'etat that overthrow cvlan regmes, gven 9 offcals of the government. They are that group of people who are least lkely to be unhappy about the polcy of the government. Further, the hgher rank they have, the hgher the L ; but at the same tme, the more lkely t s that they are bascally rather satsfed wth the exstng government, except nsofar as they would lke to have a hgher rank. Under the crcumstances, publc good consderatons - f they are of any mportance at all among government offcals consderng a coup - are more lkely to wegh n on the sde of retanng the present government than on the sde of attemptng to overthrow t. Thus, publc good consderatons are apt to play as small a part n the decson to partcpate n a coup as n partcpaton n any outsde revoluton." 5 As Tullock, 1974, p. 63, argues, Tullock, 1974, p. 62. "The nfantry prvate who s suddenly nformed by all of hs offcers that they have joned the revoluton probably fnds t extremely dangerous to do anythng except to agree wth them." 6 It s possble to fnd examples where the prvates, organzed nto battalons, resst orders (.e., the 1991 Moscow coup), but they are clearly the exceptons; exceptons that, for example, do not characterze the tradtonal Latn Amercan mltary coup d'etat.

10 that ths type of non-democratc change of government satsfes our alternatve defnton nstead of the tradtonal one. Therefore, n order to understand the behavor of the army offcers who face the decson to partcpate n a coup t s nsuffcent to analyze the prvate nterest motvatons that they may have, as t s stated by the by-product theory of revolutons; t s also necessary to analyze the publc good rewards that the hgh rank offcers may consder. There s no reason to assume that publc good consderatons are not a relevant factor n explanng a mltary coup d'etat that overthrows a democratc regme; thus, t seems approprate to vew prvate returns and publc good consderatons as complementary factors n the decsons of the army offcers. It s surprsng that the romantc publc good consderatons that are prevalent n most of the poltcal scence lterature on non-democratc changes of government cannot be rejected only n the less romantc type of rregular executve transfer: the mltary coup d'etat that overthrows a cvlan regme. Zablotsky (1992) makes use of the assumpton that prvate returns and publc good consderatons are complementary factors n the decson process of the army offcers who face the possblty to take part n a mltary coup d'etat; n the followng secton we wll make use of the same assumpton n order to obtan further nsghts on non-mltary factors that may affect the probablty of the coup. III. ECONOMIC GROWTH. A REQUIREMENT FOR POLITICAL STABILITY. Ths essay s devoted to present further results on non-mltary factors that may ncrease the probablty of a mltary coup d'etat. In drecton to ths goal I wll make use of the framework proposed by Zablotsky (1992). It wll closely follow the Tullock's approach to the subject but t also wll take nto

account the cvlan sde of the coup; the ncluson of cvlan consderatons consttutes the basc dfference between ths framework and that of Tullock, and radcally departs from the by-product theory of revolutons snce t provdes publc good consderatons, nstead of prvate nterest rewards, as the engne for the motvatons of the cvlan actors. These consderatons are a sde product of the pressure groups approach to the economc polcy developed snce the semnal work of Arthur Bentley (see Zablotsky, 1995). Our frst step wll consst to descrbe the mltary buldng block of the model. An army offcer may support a coup heavly, leadng t, or he may want to partcpate only as a follower n the event that most of hs colleagues partcpate. In the frst case hs level of support of the coup (X) wll be hgh, whle n the second t wll be small but postve. Smlarly, he may want to lead the represson, whch wll mply a large, n absolute value, but negatve (X), or he may want to partcpate n the represson as a follower whch wll mply a smaller, n absolute value, and negatve (X ). Obvously, neutralty mples X = 0. In order to choose hs optmal level of partcpaton n support of the coup, or of the represson, (X ), the army offcer wll take nto account the dfferent payoffs that he expects to receve f the coup succeeds (R, P ), or fals (D ), and hs own assessment of the probablty of success of the acton (L ). The army offcer expects to receve a prvate nterest payoff (R ) f the coup succeeds; t wll be postve for the army offcers who support the coup and negatve for the offcers who jon the represson, Secton 2). R = R (X ) R (0) = 0 dr /dx > 0 Each army offcer also expects to receve a publc good payoff (P ) f the coup succeeds (see By the same token, every offcer expects to receve a prvate nterest payoff (D ) f the coup fals; 11

12 t wll be postve for the army offcers who jon the represson and negatve for the offcers who support the coup, D = D(X ) D (0) = 0 dd /dx < 0 Then, n order to choose hs optmal level of partcpaton n support of the coup, or of the represson, each army offcer wll face the followng maxmzaton problem, T T -*t -*t Max E(U ) = L * U (R t + P t) e dt + (1 - L ) * U (D t) e dt {X } 0 0 In order to mantan the framework as smple as possble I wll assume as n Zablotsky (1992): 1. R = R, P = P, and D = D. Ths assumpton s also employed by Mran (1984), and Usher and t t t Engneer (1987), n frameworks where an agent face the possblty to partcpate n the producton of volent poltcal pressure (.e., rots, rebellons, etc.). 2. L = L (L) and dl /dl > 0, where (L) represents the probablty of success of the coup; a smlar assumpton s mplctly employed by Slver (1974) and O'Kane (1981), L = L(X,...,X ; V) ML/MX > 0 ML/MV > 0 1 n where (V) summarzes the exogenous factors that affect the probablty of success of a mltary coup d'etat for gven levels of partcpaton of the army offcers. An example of ths varable may be the partcpaton of cvlan groups n support of the coup. Under these assumptons the maxmzaton problem faced by each army offcer becomes, Max E(U ) = B {L (X,..., X ; V) U (R + P ) + [1 - L (X,...,X ; V)] U (D )} 1 n 1 n {X } T -*t where, B = * e dt

13 0 The next step wll consst to formalze the problem faced by the cvlan actors. The exact specfcaton of ths problem lacks of relevance as far as t contemplates the exstence of a postve margnal cost of partcpaton; ths cost wll rule out the partcpaton of any pressure group who does not affect the probablty of success of the coup to a perceptble degree. Consder, for example, that each pressure group faces the followng maxmzaton problem, Tj T j -*t -*t Max E(U j) = L j* U j(w jt + M jt - C jt) e dt + (1 - L j) * U j(w jt + D jt - F jt) e dt {Y j} 0 0 whch under smlar assumptons to the ones mposed to the mltary buldng block, 1. W = W, M = M, D = D, C = C, and F = F jt j jt j jt j jt j jt j 2. L = L (L), and dl /dl > 0 j j j becomes, Max E(U ) =, [L U (W + M - C ) + (1 - L ) U (W + D - F )] j j j j j j j j j j j {Y } j T j -*t where,, = * e dt 0 and, Y j = Level of partcpaton of each dentcal member of the group j n support of the coup (Y j > 0), or of the represson (Y < 0). j W = Income of the agent ndependent of government redstrbuton. j M = Government redstrbuton to each member of the group j under the rules of the redstrbutve game j

14 emboded n a mltary regme. C = Cost of partcpaton n support of the coup. j C = C(Y ) and dc /dy > 0 f Y > 0 j j j j j C(Y j) = 0 f Y j# 0 D = Government redstrbuton to each member of the group j under the rules of the redstrbutve game j emboded n a democratc regme. F = Cost of partcpaton n defense of the democratc regme. j F = F(Y ) and df /dy < 0 f Y < 0 j j j j j F(Y j) = 0 f Y j$ 0 The nteracton between the actors s modeled as a Cournot-Nash non-cooperatve game n ther level of partcpaton; then, the equlbrum s determned by the utlty maxmzng condton for each actor (mltary or cvlan) wth respect to hs level of partcpaton n support of the coup or of the represson, takng as gven the level of partcpaton of any other actor, ME(U)/MX = ML/MX [U(R + P) - U(D)] + L U'(R + P) R'+ (1 - L) U'(D) D'= 0 ME(U)/MY = ML/MY [U(W+M-C)-U(W+D-F)] - L U'(W+M-C) C'- (1-L) U'(W+D-F)F'= 0 where we are omttng from now on the subscrpts and j, and I am assumng B =, = 1. By comparng both sets of frst order condtons t becomes clear that ths framework would satsfy the stylzed fact that most army offcers take part n a coup whle most cvlan actors defer from dong so (see Zablotsky, 1992). The framework provdes army offcers not only wth publc good consderatons but also prvate nterest rewards; then, whle the total payoff expected by the army offcers s not ndependent of ther level of partcpaton, the total payoff expected by the cvlan actors s only based n

a publc good consderaton: the change n the outcome of the redstrbutve game emboded n the overthrowng of the democratc regme. Therefore, whle most army offcers wll choose to take part, most pressure groups wll choose to reman nactve, unless they can affect the probablty of nstallaton of the mltary regme to a perceptble degree, * ML/MY = 0 Y Y = 0 By means of a smlar argumentaton t s easy to show that the model also would satsfy the stylzed fact that n most of the mltary coups d'etat t s usually verfed some sort of support by part of the cvlan populaton but not any form of cvlan resstance. To contemplate ths emprcal asymmetry we have made use of a publc good theory - based upon the pressure groups approach to the economc polcy - gven that under ths framework the cvlan actors wll only choose to partcpate f they can sgnfcantly affect the probablty of success of the coup (see Zablotsky, 1992). Under ths scenaro f the partcpaton of some of the pressure groups beneftted by the change of poltcal regme affects the probablty of success of the coup, but the partcpaton of any of the groups harmed does not, the former groups would support the coup but the latter wll reman nactve. The maxmzaton problem faced by the actors allow them to choose ther optmal level of partcpaton n the contngent stage of a mltary coup d'etat, but t does not explan how the coup has begun. We wll assume, as t s also done by Tullock (1974), O'Kane (1981), and Zablotsky (1992), that an ncrease n the probablty of success wll ncrease the lkelhood that a subgroup of the army offcers would decde to begn the acton, C = C(L) and dc/dl > 0 where, (C) represents the probablty of a mltary coup d'etat. 15

16 Our next step wll consst to make use of the descrbed framework n order to obtan further nsghts on non-mltary factors that may affect the probablty of a mltary coup d'etat; n drecton to ths goal we wll propose the followng hypothess: A decrease n the ncome ndependent of government redstrbuton (W) of the cvlan actors beneftted by the change of poltcal regme wll ncrease, assumng decreasng margnal utlty, the benefts dspensed to the actors by a successful coup; ths would create an ncentve for ther partcpaton n ts support, consequently rasng the probablty of ths non-democratc change of government. Sgn MY/MW = Sgn {ML/MY [U'(W+M-C)-U'(W+D)]-L U''(W+M-C) C'} < 0 f *ML/MY [U'(W+M-C)-U'(W+D)]* > *L U''(W+M-C) C'* In order to llustrate the plausblty of ths hypothess t wll be necessary to dentfy a pressure group that would be beneftted by the change n the rules of the redstrbutve game emboded n the overthrowng of the democratc regmes, and then to analyze the behavor followed by some varables that may affect the ncome ndependent of government redstrbuton of ts members. Gven that there are no reasons to assume that n varous countres the pressure groups beneftted by the overthrow of democratc regmes wll be the same, I wll center my nterest on a specfc one: 7 Argentna, where the export (agrcultural) sector appears to have been beneftted by the modfcatons n the commercal polcy emboded n the overthrow of the democratc regmes (see Zablotsky, 1992). 7 Actually, there are no reasons even to assume that n a gven country the groups beneftted wll be the same across tme, snce dfferent varables that affect the outcome of the redstrbutve game may vary (see, for example, Becker, 1983).

17 Table 1 descrbes the behavor followed by the real foregn prce ndex of exports; ndex hghly assocated wth the ncome ndependent of government redstrbuton of the export sector. The ndex actually fell before the overthrow of each of the democratc regmes (1962, 1966, and 1976); ths fact motvates us to nvestgate further the plausblty of the hypothess by basng our analyss on the behavor followed by the nternatonal prce of the man agrcultural products.

18 TABLE 1 REAL FOREIGN PRICE INDEX OF EXPORTS (1960=100) Real Real Year Even Foregn Year Even Foregn t Prce t Prce Index of Index Exports of Exports (1) (1) 1960 100 1972 113 1961 97 1973 138 1962 Coup 91 1974 146 1963 95 1975 121 1964 107 1976 Coup 106 1965 105 1977 102 1966 Coup 102 1978 97 1967 97 1979 111 1968 92 1980 115 1969 89 1981 109 1970 93 1982 89 1971 103 1983 79 Source: Adolfo Sturzenegger, Wylan Otrera and Beatrz Martnez Mosquera, Trade, Exchange Rate, and Agrcultural Prcng Polces n Argentna, World Bank Comparatve Studes, The World Bank, May 1990. where, (1) Foregn Prce Index of Exports (CEPAL)/USA Wholesale Prce Index. Table 2 reports the relatve partcpaton of the man agrcultural products between 1960 and 1984, TABLE 2

19 RELATIVE PARTICIPATION OF THE MAIN AGRICULTURAL PRODUCTS 1960-1964 1970-1974 1980-1984 Product Mllon Percent Mllon Percent Mllon Percent of U$S of U$S of U$S Beef 1038 30 2137 30 4182 25 Dary 426 112 753 11 1781 11 Prod. Wheat 368 11 543 8 1656 10 Corn 241 7 751 11 1266 8 Total 2073 60 4184 60 8885 54 Source: Adolfo Sturzenegger, Wylan Otrera and Beatrz Martnez Mosquera, Trade, Exchange Rate, and Agrcultural Prcng Polces n Argentna, World Bank Comparatve Studes, The World Bank, May 1990. All of the products, wth the excepton of the dary ones, are traded goods; then, I wll focus my attenton n ther real nternatonal prces. To consder another ndependent source of nformaton, I also wll examne the behavor of an ndex elaborated by Adolfo Sturzenegger, Wylan Otrera and Beatrz Martnez Mosquera (1990): the relatve prce of each of these agrcultural products respect to nonagrcultural ones n absence of any form of government nterventon (drect or ndrect). where, The relatve prces n the absence of any form of governmental nterventon are defned as: * * FOB e P /P NA = {[(P * E ) - GP ]/1.03 - C }/P NA*

20 - P */P NA* = Relatve prce n the absence of government nterventon of agrcultural good respect to the nonagrcultural goods. - P = FOB prce of product. FOB e - E = Equlbrum adjusted nomnal exchange rate. - GP = Port costs for product. - 1.03 = Export commsson costs. - C = Transport and dstrbuton costs (from the farm to the port) of product. * - P NA = X nat I nat [d/(1+t nat)] + X s I s + X cc Icc * - d = E /E. Dvergence between the actual real exchange rate (E) and the sustanable equlbrum free- * trade/real-exchange rate (E ). - t = Estmated mplct tarff for the nonagrcultural tradable ndex. nat - I = Nonagrcultural tradable ndex. nat - I = Servce ndex. s - I = Cost of constructon ndex. cc - X = 0.36 nat - X = 0.57 s - X = 0.07 cc Table 3 summarzes all the relevant nformaton provded by both ndcators (n the Appendx there are reported the tme seres of each of them). TABLE 3

21 SUMMARY OF THE RESULTS Items Increasng Path Decreasng Path Real Internatonal Prces Beef 0 4 Wheat 1 3 Corn 2 2 Relatve Prces n the Absence of Government Interventon Beef 0 3 Wheat 0 3 Corn 1 2 Both ndcators have usually fallen before mltary coups d'etat that have overthrown democratc regmes (the real prce n 75 percent of the observatons and the relatve prce wthout any form of governmental nterventon n 89 percent of the cases). Therefore, t seems far to state that the evdence I have examned does not reject the proposed hypothess, because the ncome ndependent of government redstrbuton of the agrcultural sector seems to have fallen before coups that have overthrown democratc regmes; whch would have rsen, n terms of our framework, the probablty of ths non-democratc change of government. Morón (1994) also provdes emprcal evdence that may support the proposed hypothess. He

22 makes use of a panel data consstng of nformaton about 16 Latn Amercan countres (Argentna, Bolva, Brazl, Chle, Colomba, Domncan Republc, Ecuador, El Salvador, Guatemala, Honduras, Hat, Panama, Peru, Paraguay, Uruguay, and Venezuela) for the perod 1950-1990. He shows that the lack of dversfcaton n the export structure of a country contrbutes to explan successful coups d'etat. Countres that had not been able to dversfy the export sector are more lkely to suffer economc nstablty due to an external shock orgnated manly by sharp declnatons n the commodty prces. Therefore, world market condtons and n partcular export characterstcs form a very sound bass for developng a general explanaton for coups d'etat. Ths evdence s consstent wth our proposed hypothess snce, n a small country where the export sector s beneftted by the change n the rules of the redstrbutve game emboded n a successful coup, the probablty of ths event wll ncrease f the nternatonal prce of the export good decreases, because t wll decrease the ncome ndependent of government redstrbuton of the sector. Havng determned the role played by the ncome ndependent of government redstrbuton t becomes clear the role played by economc growth; economc growth may reduce the probablty of a mltary coup d'etat snce t would ncrease the ncome ndependent of government redstrbuton of the nterest groups beneftted by the event. It s nterestng to pont out that our hypothess satsfes the stylzed fact that hgh developed countres would be a less probable stage for mltary coups d'etat than low developed ones. Moron (1994) provdes evdence that sustans ths asserton; he shows that the poverty stuaton of a country affects the number of coups d'etat n the same country. Zehra Fatma Arat (1984) has bult an ndex of democratcness for selected countres whch allows

8 us to llustrate clearly the pont. From the Arat sample we have selected the 63 countres whch have been ncluded durng the whole perod and we have classfed 17 of them under the label of "frst world countres", and the remanng 46 under the label of "others"; from the later we have selected the 19 Latn Amercan countres (see Table 4). Whle the average score for the 17 "frst world countres" reached 19.40, t dropped for the 19 9 Latn Amercan countres to 10.19, and to only 8.04 for the 46 "non frst world countres" as a whole. 23 8 The measure of democratcness s based upon prncples whch lead to hgher levels of popular control. Ths control s perceved to have three components: poltcal partcpaton (whch measures the extent that popular wll s reflected at decson-makng nsttutons), compettveness (whch measures the compettveness of the poltcal system), and cvl and poltcal lbertes (whch measures the coercveness of the government). The estmated scores, whch are ranked n the (0-20) nterval, fluctuate between 0.55 and 18.91; the hgher the rank, the hgher the degree of democratcness. 9 We have classfed under the label of "frst world" the Western European countres n addton to the USA, Canada, Australa and New Zealand. We have classfed under the label of "others" the remanng forty sx countres: Afghanstan, Albana, Bulgara, Chna, Czechoslovaka, Ethopa, Greece, Hungary, Iran, Iraq, Jordan, Lebanon, Lbera, Mongola, Nepal, Phlppnes, Poland, Portugal, Romana, South Afrca, Saud Araba, Span, Thaland, Turkey, USSR, Yemen, A.R., Yugoslava plus the followng nneteen Latn Amercan countres: Argentna, Bolva, Brazl, Chle, Colomba, Costa Rca, Cuba, Domncan Republc, Ecuador, El Salvador, Hat, Honduras, Mexco, Ncaragua, Panama, Paraguay, Peru, Uruguay and Venezuela.

24 TABLE 4 SCORES OF DEMOCRATICNESS FOR 63 SELECTED COUNTRIES Average Frst World Others Latn Amerca Year (63) (17) (46) (19) 1950 11.32 19.39 8.34 11.10 1955 10.94 19.43 7.80 10.21 1960 11.42 19.35 8.49 11.70 1965 11.25 19.42 8.23 10.64 1970 10.94 19.36 7.83 9.96 1975 10.75 19.42 7.56 7.51 Average 11.10 19.40 8.04 10.19 Source: Compled from Zehra Fatma Arat, "The Vablty of Poltcal Democracy n Developng Countres. Ph.D. dssertaton, The Graduate School of the State Unversty of New York at Bnghamton, 1984. In concluson, ths paper has proposed the hypothess that there exsts a relaton from economc growth to poltcal stablty. Ths hypothess does not oppose but complement the usual one, whch states that poltcal stablty s a prerequste for economc growth. Under our hypothess the relaton between poltcal stablty and economc growth has to be understood as a two way relaton.

APPENDIX 25

26 TABLE A.1: BEEF Real Value of Exports Year Event (1951=100) Chlled Beef Corned Beef 1951 100 100 1952 104 115 1953 132 118 1954 Coup 133 112 1955 125 100 1956 95 94 1957 89 87 Source: I.M.F., Internatonal Fnancal Statstcs. Real Prce Relatve Year Event FOB Prce Buenos Ares wthout (1960=100) Interventon 1960 100 10.05 1961 95 9.56 1962 Coup 88 8.57 1963 87 7.15 1964 123 9.07 1965 143 11.67 1966 Coup 100 8.21 1967 90 7.65 1968 86 7.44 1969 81 7.11 1970 87 6.71

27 TABLE A.1--Contnued Real Prce Relatve Year Event FOB Prce Buenos Ares wthout (1960=100) Interventon 1971 119 8.51 1972 127 8.53 1973 165 9.01 1974 140 8.30 1975 92 6.31 1976 Coup 60 3.60 1977 85 5.66 1978 72 4.74 1979 112 6.25 1980 117 7.31 1981 120 9.28 1982 100 8.98 1983 91 8.67 Source: Adolfo Sturzenegger, Wylan Otrera and Beatrz Martnez Mosquera, Trade, Exchange Rate, and Agrcultural Prcng Polces n Argentna, World Bank Comparatve Studes, The World Bank, May 1990. Note: Prce FOB, Buenos Ares: annual, U$S/ton.

28 TABLE A.2: WHEAT Year Event Real Value of Exports (1951=100) 1951 100 1952 119 1953 122 1954 88 1955 Coup 87 1956 75 1957 71 Source: I.M.F., Internatonal Fnancal Statstcs. Real Prce Relatve Year Event FOB Prce Buenos Ares wthout (1960=100) Interventon 1960 100 1.64 1961 106 1.83 1962 Coup 107 1.41 1963 102 1.49 1964 113 1.41 1965 100 1.33 1966 Coup 82 1.10 1967 90 1.05 1968 96 1.49 1969 92 1.46 1970 87 1.21

29 TABLE 39.--Contnued Real Prce Relatve Year Event FOB Prce Buenos Ares wthout (1960=100) Interventon 1971 85 1.06 1972 85 1.22 1973 138 1.62 1974 221 2.62 1975 179 2.68 1976 Coup 124 1.44 1977 81 1.03 1978 94 1.22 1979 88 0.99 1980 119 1.44 1981 121 1.07 1982 99 1.20 1983 81 1.44 Source: Adolfo Sturzenegger, Wylan Otrera and Beatrz Martnez Mosquera, Trade, Exchange Rate, and Agrcultural Prcng Polces n Argentna, World Bank Comparatve Studes, The World Bank, May 1990. Note: Prce FOB, Buenos Ares: January (December or February durng 1966-1970, 1973 and 1975), U$S/ton.

30 TABLE A.3: CORN Year Event Real Value of Exports (1951=100) 1952 100 1953 68 1954 53 1955 Coup 63 1956 56 1957 52 Source: I.M.F., Internatonal Fnancal Statstcs. Real Prce Relatve Year Event FOB Prce Buenos Ares wthout (1960=100) Interventon 1960 100 1.37 1961 94 1.31 1962 Coup 96 1.23 1963 104 1.27 1964 104 1.06 1965 112 1.34 1966 Coup 93 1.02 1967 99 1.35 1968 89 1.17 1969 89 1.20 1970 96 1.11

31 TABLE A.3--Contnued Real Prce Relatve Year Event FOB Prce Buenos Ares wthout (1960=100) Interventon 1971 95 0.97 1972 97 1.03 1973 127 1.02 1974 128 1.16 1975 139 1.31 1976 Coup 125 1.45 1977 95 1.09 1978 95 1.07 1979 83 0.74 1980 102 0.97 1981 89 0.85 1982 70 0.82 1983 80 1.18 Source: Adolfo Sturzenegger, Wylan Otrera and Beatrz Martnez Mosquera, Trade, Exchange Rate, and Agrcultural Prcng Polces n Argentna, World Bank Comparatve Studes, The World Bank, May 1990. Note: Prce FOB, Buenos Ares: May (July n 1967), U$S/ton.

32 REFERENCES Alesna, Alberto and Roberto Perott. "Income Dstrbuton, Poltcal Instablty and Investment," mmeo, Harvard Unversty, September 1993. Arat, Zehra Fatma. "The Vabllty of Poltcal Democracy n Developng Countres," Ph.D. dssertaton, The graduate School of the State Unversty of New York at Bnghamton, 1984. Barro, Robert. "Economc Growth n a Cross Secton of Countres," The Quarterly Journal Economcs, of May 1991. Bentley, Arthur. The Process of Government, Chcago: The Unversty of Chcago Press, 1908. Buchanan, James and Gordon Tullock. The Calculus of Consent, Ann Arbor: Unversty Mchgan of Press, 1962. Cao Garca, Ramon. Exploratons Toward an Economc Theory of Poltcal Systems, New York: Unversty Press of Amerca, 1983. Cartwrght, P.; C. Delorme; and N. Wood. "The By-Product Theory of Revoluton: Some Emprcal Evdence," Publc Choce N. 46, Summer 1985. Downs, Anthony. An Economc Theory of Democracy, New York: Harper and Row, 1957. Grossman, Herschel. "A General Equlbrum Model of Insurrectons," Amercan Economc Revew N. 81, September 1991. Ireland, T. "The Ratonale of Revolt," Papers on Non Market Decson Makng N. 3, Fall 1967. Kuran, Tmur. "Sparks and Prare Fres: A Theory of Unantcpated Poltcal Revoluton," Publc Choce N. 61, Summer 1989.

33 Letes, N. and C. Wolf, Jr. Rebellon and Authorty, Chcago: Markham, 1970. Levne, Ross and Davd Renelt. "A Senstvty Analyss of Cross-Country Growth Regressons," The Amercan Economc Revew, September 1992. Mauro, Paolo. Corrupton, Country Rsk and Growth, paper presented at the XIII Latn Amercan Meetng of the Econometrc Socety, Venezuela, August 1994. Mbaku, John and Chrs Paul. "Poltcal Instablty n Afrca: A Rent Seekng Approach," Publc Choce N. 63, Wnter 1989. Mran, S. Kaveh. "Collectve Poltcal Volence and the Redstrbuton of Poltcal Income," Ph.D. dssertaton, The Unversty of Chcago, 1984. Morón, Eduardo. A Coup d'etat Model for Latn Amercan Data: A Panel Data Model wth a Posson Regresson, Unpublshed manuscrpt, UCLA, 1994. O'Kane, Rosemary H. "A Probablstc Approach to the Causes of Coups d'etat," Brtsh Journal of Poltcal Scence N. 3, July 1981.... The Lkelhood of Coups, Averbury, England, 1987. Olson, Mancur. The Logc of Collectve Acton: Publc Goods and the Theory of Groups, Cambrdge, Mass.: Harvard Unversty Press, 1965. Slver, M. "Poltcal Revoluton and Represson: An Economc Approach," Publc Choce N. 17, Sprng 1974. Sturzenegger, Adolfo, Wylan Otrera and Beatrz Martnez Mosquera, Trade, Exchange Rate, a n d

34 Agrcultural Prcng Polces n Argentna, World Bank Comparatve Studes, The World Bank, May 1990. Tullock, Gordon. "The Paradox of Revoluton," Publc Choce N. 11, Fall 1971.... "The Paradox of Revoluton," The Socal Dlemma, Blacksburg, VA: Unversty Publcatons, 1974.... Autocracy. Dordrecht, The Netherlands: Kluwer Academc Publshers, 1987. Usher D. and M. Engneer. "The Redstrbuton of Income n a Despotc Socety," Publc Choce N. 54, Summer 1987. Zablotsky, Edgardo. "An Economc Theory of Autocracy. A Publc Good Approach to Mltary Coups d'etat," Ph.D. dssertaton, The Unversty of Chcago, June 1992.... "An Interest-Groups Approach to Economc Growth," Workng Paper N. 100, CEMA, November 1994.... "The Process of Government," The New Economcs of Human Behavor, Marano Tommas and Kathryn Ierull (eds.), Cambrdge Unversty Press, 1995.