This article departs substantially from both prevailing approaches to state buildingand decay, suggesting that neither has a generalizable or direct causal effect on thedevelopment of state institutions. 4 By contrast, I contend that the role of wars andwealth in building states is at best conditional; what really matters is a set of logicallyprior social and political variables that help to define the path down which politicaldevelopment will proceed.

war and natural resource wealth. Thefirst considers external threats to be propellants that supersede divisions among com-peting domestic elites, make substantial increases in revenue extraction politicallyviable, and open the door to the modernization of military and state institutions and thecentralization of authority.1 The second approach is more recent, and it is negativelyframed. It suggests that natural resource wealth is, paradoxically, detrimental to politi-cal development.

nstead, it argues that a set of logically priorconditions—the social relations that govern the principal economic sectors and thepattern or intraelite conflict or compromise—launch path-dependent processes thathelp determine when, and if, either strategic conflict or resource wealth contributeto, or impede, institutional development.

milarly, it challengesthe conventional wisdom that abundant exportable natural resource wealth is likelyto provoke institutional atrophy.

This manuscript departs strongly from conventional accounts that ascribe a centralrole to war and the threat of war in Third World state building. S

The purpose of this article has been simply to account for the part playedby regionalism in accounting for variation and to question the conventionalwisdom about the part played by ethnic diversity.

Second, my focushas been on the policies determined by political institutions and not theirimplementation: To the extent that policies are incompletely implemented,the causal framework developed in this article will over-predict levelsof public good provision in cases where regionalism is not salient.

First,although it can account for national government policy in regards to publicgood provision, it cannot account for the differential provision by subna-tional and local authorities, which is sizable in some cases.

findings do not support the possibility that regionalism acts as an interveningvariable or causal mechanism linking ethnic diversity to low levels of publicgood provision

Third, I have shown that the absence ofurban primacy (a proxy for regionalism) was associated with low levels ofpublic good provision across Spanish South America during the first centuryafter independence, with correlation coefficients as strong as or stronger thanthose found for ethnic heterogeneity in this set of cases, and that decliningregionalism in the case of Venezuela was associated with increased publicgood provision

In the Colombiancase, I showed that self-sufficient regions shaped an elite consensus that thenational state should not prioritize public good provision and should insteadleave its development to the regions.

The Ecuador case showed thatregions have distinct public good preferences, which generates politicaldeadlock, conflict, and low levels of public good provision.

The case of Venezuela thus shows a shift in public good provision thatcoincides with a decline in regionalism and the rise of Caracas as a singlenational center. This shift over time provides further support for the impor-tance of regionalism as a heretofore unexplored factor in accounting for vari-ation in public good provision. Moreover, the increased provision of publicgoods cannot be traced to a decline in ethnic diversity: As shown by Wright(1990), the country continued to be characterized by a “multi-racial majority”(p. 79), and elite discourse continued to view race as a fluid, largely socioeco-nomic category

Additional evidence for the regionalism explanation I have advanced can beseen in a cross-time examination of trends in regionalism and public goodprovision in Venezuela, where a sharp increase in the primacy of Caracas inVenezuela, without a concomittant decrease in ethnic diversity, was associ-ated with an increase in public good provision. Venezuela’s over-time trajec-tory thus confirms the power of the regionalist argument and shows that itoperates independent of the ethnic diversity account

Unlike this measure of ethnic diver-sity, the urban primacy measure of regionalism uses data from an earlierdate than the public good provision indicators and avoids this problem.This fact, combined with the historiographical evidence presented above

Table 2 also shows the correlations of country scores on railroad den-sity and literacy for two indicators of ethnic diversity—the Alesina et al.(2003) scores and the ELF scores (drawn from Fearon & Laitin, 2003). Ascan be seen at a glance in Figure 1, correlations of the Alesina et al.(2003) measure are equally strong as those for urban primacy, while thoseusing the ELF score with levels of public good provision are far weaker.Yet one cannot conclude from this that fractionalization causes low levelsof public good provision, given that the fractionalization measure, whichis calculated for the 1980s or 1990s in the Alesina et al. (2003) data set, ismeasured after the proposed outcome of public good provision

Ruggedness has a stronger relationship with literacy thanwith railroad density. Thus, there is reason to cast doubt on these geographicexplanations as the underlying cause of the correlation between urban primacyand public good provision in Latin America

In each case, higher urban pri-macy scores are associated with stronger states. These correlations remainsizable (in results not shown here) even with the exclusion of Uruguay, whichis an outlier in its high primacy score. Confidence about the direction of theproposed relationship is increased by the fact that primacy is measured forthe earliest post-independence date possible and at an earlier point in timethan the literacy rate or railroad density. Thus, we can rule out the possibilitythat public good provision could affect urban primacy—for example, bydrawing population to the capital city.1

0.25 (if the four cities are equal insize) to 1 (if the largest city contains the entire population of the country). InLatin America, urban primacy scores vary from Uruguay at 0.84 to Venezuelaand Paraguay at 0.39. The regional average is 0.54, with a standard deviationof 0.15.

In this cross-national analysis, I use urban primacy as a proxy for region-alism, using population share as a crude but telling indicator of political andeconomic importance

These include a common colonial background and a commonindependence process, broadly similar political economies centered on pri-mary product exports through the first century after independence, and thecommon absence of the sorts of wars and territorial conflicts that promotedstate building and public good provision elsewhere (Centeno, 2002).

Having shown that the proposed mechanisms can be found in various cases,I now turn to broader correlational evidence that salient regionalism is associ-ated with low public good provision. Spanish South America is an appropri-ate locus for studying the effects of regionalism because its countries sharesome core properties in common, and thus one can rule these out as sourcesof variation.

The striking hegemony of Colombian anti-statist views about develop-ment derived from the self-sufficiency of the country’s regions, which ledpolicy makers to see “order and progress”—the sorts of goals that droveprojects of public good provision elsewhere in the region—as goals to bepursued at the regional, rather than national, level.

olombia was an outlier in the absence ofuniversal schooling throughout its first century after independence, makingno attempt to standardize school curricula and promote national educationlike many of its neighbors

The fact that Colombia was “eco-nomically invertebrate” (Safford, 1988, p. 53) made the payoffs from estab-lishing a powerful state unclear and made the path to development throughthe fostering of regional self-sufficiency seem plausible. As a result, “theintegration of a national market, and therefore transportation improvement,may have been less fundamentally important to Colombians” than it was topolitical elites in our other cases (Safford, 1988, p. 37). Regions were tiedinto the world economy “without the presence and involvement of the centralstate” (Palacios, 1980, p. ix)

There was simply no vision ofColombia as a national economic unit centered on Bogotá and “there waslittle to draw [elites] automatically to the capital” (Christie, 1979, p. 44). Asa result of its lack of urban primacy, Colombia was unique “in the deepimprint its regionalist sentiment has left on the pattern of national develop-ment” (Park, 1985, p. 7).

alacios(2006) describes Colombia as a “mosaic of isolated regions” (p. 5).

Thus,although Alesina et al., (2003) would claim Colombia as a case fitting theirmodel—its fractionalization score is one of the highest in Latin America—Ishow that correlation in this case does not imply causation

A review of Colombia’s political and economic history shows that regional-ism caused low levels of public good provision, which are reflected both inits development of railroad infrastructure (second-lowest railroad density inthe region) and its relatively low literacy rate, as well as in many other aspectsof state activity (Deas, 1982).

In this context, the railroad, with the transfer ofresources from coast to highlands that it implied, became a palatable, if sub-optimal, strategy to increase the labor supply in the cacao sector.Although elites in the two regions did not share a common vision of thebroad project of national development, the railroad served the goals of eachgroup, and a narrow consensus was forged that facilitated its development

Highlanders sought a railroad from Quito across the spine of the Andes tothe coast. Coastal planters, who relied on river transport but were the princi-pal contributors to the fisc through import taxation, did not need the railroadfor their export trade to thrive (Henderson, 1997, p. 172). Guayaquil madesignificant efforts to develop river navigation and steamship commerce alongthe coast, but had little interest in extending infrastructure into the highlands(Deler, 1994, p. 312ff). Coastal resistance in Congress prevented progress onrailroad development, and a rail link connecting Quito and Guayaquil wascompleted (as discussed below) only in 1908

The sharp difference in regional economies prompted divergent preferencesabout public good provision. These manifested in Guayaquil’s displeasureover bearing the burden of most of the customs duties while infrastructurespending was concentrated in the highlands.

Ecuadoreanstate performed poorly in the provision of public goods: It had the lowestrailroad density of any country in the region and a below-median literacyrate. This is a country that might appear at first glance to have deep identity-based divisions; it has one of the most ethnically diverse populations in LatinAmerica, as seen in its relatively high scores on both the ethno-linguisticfractionalization (EFL) and the Alesina et al. (2003) fractionalization mea-sures of ethnic diversity. Yet I show that low levels of public good provisionin this case are not explained by disagreement across ethnic groups aboutpublic good preferences, but instead by longstanding patterns of interregionaldisagreement among elites from the mountain and coastal regions.

Low levels of public good provision by thenational state can therefore result from the disinclination of state leaders topursue it, which is likely to emerge in a context of salient and economicallyindependent regions.

Where the gains from national economic inte-gration are expected to be minimal, given existing technology, I argue thatpublic good provision comes to be seen by central state leaders as less propi-tious for economic development, political stability, and social peace. Underthese conditions, states will more likely pursue development through policiesthat allow already salient regional economies to flourish.

Where intra-regional trade and economic exchange is limited, and there is little potentialfor it to grow, there are fewer gains to be reaped from public goods that couldknit the country together into a single national unit (Spruyt, 1994)

Here lies the second mechanism linking regionalism to low state provisionof public goods: The absence of urban primacy is often accompanied by thepresence of salient regional economies, which operate largely independent ofone another and are separately integrated into the world market.

his mechanism centers onregionalism’s role in shaping the normative commitments of the state to publicgood provision; the place of public goods in the ideology of development

First, because public goods are by their nature fundamentallyspatial, the claim that people in different places have different preferencesover them requires fewer assumptions about the origins of those preferencesthan does the claim that preferences about public good provision are shapedby ethnicity.

I argue that distinct regions, defined and characterized by the organizationof economic activity around distinct urban centers, are likely to have divergentpreferences about the location and type of public goods provided by the state.3

Notably, these region-specific accounts are in tension with the ethnicdiversity arguments because of the mismatch between the reified nature ofethnic categories in those theories and the more deeply constructed, plastic,and fundamentally socioeconomic nature of ethnic categories highlighted inaccounts of ethnic diversity in Latin America.

Nevertheless, the argument developed in this article draws on the purelydemographic and economic component of regionalism, showing that the dis-tribution of population and economic activity, independent of identity-basedmechanisms, is sufficient to reduce public good provision by national states.

For the purposes of this article, I use the term “regionalism” to refer to theexistence of multiple urban concentrations with associated economies withina single country. Regionalism refers, then, to the opposite of urban primacy,which describes countries that have a disproportionate share of the popula-tion living in or near a single major city.

I identify two mechanisms—divergent public good preferences andregional self-sufficiency—that underpin this relationship, and trace themthrough case studies of Ecuador and Colombia, two countries that score highon standard measures of ethnic diversity and that saw little public good provi-sion during the first century after independence. In both cases, I show thatregionalism, rather than ethnic diversity, explains the low levels of publicgood provision

I show that even in the absence of cultural fractionalization, the very exis-tence of urban concentrations outside the capital can shape public good pro-vision.

Instead of ethnic diversity, this article shows that variation in public goodprovision is determined by the salience of demographic and economic region-alism—conceptualized as the presence of multiple large population centersand distinct economic units within a single country’s borders.

Moreover, this variation is patterned in ways predicted by theethnic diversity hypothesis; high diversity is correlated with lower levels ofpublic good provision. Yet this article finds little evidence of ethnic diversi-ty’s dampening effect on public good provision in Latin America. This chal-lenges not only the claims about these cases but also the broader cross-nationalcorrelational evidence that includes them

A plausible case for this aspect ofgeography as a cause of low public good provision poses a challenge to thescholarship attributing it to ethnic diversity

Two mechanisms—divergent publicgood preferences and regional self-sufficiency—underpin this relationship.

This article argues that variation in public good provision is determined bythe salience of demographic and economic regionalism, conceptualized asthe presence of multiple large population centers and distinct economicunits within a single country’s borders.

Accordingly, it is reasonable to concludethat methodological, not specific, differences account for performance of thetwo lexigram-competent cohorts.33

As an 8-year-old, Kanzi evidenced comprehension of spoken English (74percent correct) comparable to that of Alia, a 2-year-old human child (65percent correct). He readily responded to novel verbal instructions such as“Put the melon in the potty” and “Go outdoors and get the pine needles.” Hisperformance fell off dramatically (33 percent success) when asked to give twodifferent objects collectively to an animate receiver or to perform two differ-ent actions with different objects. The researchers tested Kanzi and Alia withmore than seven types of sentence, one of which (type 5C) had end phraserecursion. Kanzi’s score (77 percent correct) on type 5C requests (Go get ob-ject X that’s in location Y) was higher than Alia’s score (52 percent correct).327

Rumbaugh et al. concluded “it was through Kanzi’s access to the patternedexperiences afforded by the logic structure of his environments (e.g., the speechof the experimenters and their use of word-lexigrams on a keyboard thatstructured his mother’s instructional sessions) that he perceptually discernedand learned the relationships between symbols and events that provided forhim the basic pro cesses and competencies with language

Undeterred by skeptics and critics, Savage-Rumbaugh et al. proceeded toadvance the field notably via lexigram studies with bonobos: Kanzi, his half-sisters Mulika (1983– 86) and Panbanisha, and Panbanisha’s son Nyota (b.1998). Kanzi, Panbanisha, and Nyota reside with four lexigram-naïve bonobosat the Bonobo Hope Sanctuary that was founded in 2005 at the Great ApeTrust of Iowa to accommodate bonobos from the Language Research Centerat Georgia State University (LRCGSU), Atlanta.322 Highly productive stud-ies continue at LRCGSU with a lexigram-proficient chimpanzee, Panzee.323She was coreared with Panbanisha and like her and other lexigram-proficientbonobos understands human speech (Chapter 10).

those whofostered their subjects (the Gardners, Fouts, Terrace, Miles, and Patterson)and those who elected more traditional psychology laboratory settings (thePremacks, Rumbaugh, and Savage-Rumbaugh). 3

on symbolic communication in naturalistic social relations instead offocusing on food and bourgeois children’s toys and activities.

Accordingly, Savage-Rumbaugh et al. con-cluded that the Sherman-Austin study evidenced:• Great apes can comprehend symbols, but production does not leadspontaneously to comprehension.• In order to function representationally, the symbols must become de-contextualized and freed for use in novel situations.• Great apes can communicate with one another via symbols if they developjoint attention in an environment that requires mutual cooperation.• Great apes can make informative statements regarding their intendedfuture actions.• Referential comprehension and usage are prerequisites to the develop-ment of syntactic competence.3

Savage-Rumbaugh et al. cogently argued that Washoe, Sarah, Lana, Nim,and other chimpanzees that used symbols in ways that emulate human usagedid not comprehend their representational functions

Savage-Rumbaugh et al. challenged that Sarah’s successful choices of signsglossed as “name of,” negation, and so on were often due to conditionalmatch-to-sample strategies—that is, Sarah’s tasks were not communicative.301Like Ameslan signs and lexigrams, the plastic signs might not be symboli-cally equivalent to human words. 302 They simply showed that she couldsolve “complex conceptual tasks, but the relation between Sarah’s per for-mance on such tasks and language as utilized by human speakers remainedvague.”303 The Premacks were not persuaded by their arguments, and contin-ued to reason that, like human words, the plastic signs represent concepts inchimpanzees.3

Clearly, apes do not easily imitate human spoken words even when theyare immersed in human society and monotonously tutored and tormented.Other approaches had to be sought

Savage-Rumbaughet al. described spontaneous forelimb gestures that orchestrated copulatorypositions between captive bonobos. 244 They included positioning movementsin which one individual touched another, thereby indicating how it was toposture its limbs, and various iconic hand motions that showed its partnerwhat to do. Bonobos commonly engage in ventroventral copulatory posi-tions, including sustained eye contact that clearly distinguishes them fromchimpanzees and gorillas.

Accounts from the field and reports based on captives have evidenced thatbonobos have some unique postures, gestures, and vocalizations. 212 Forinstance, an alarmed arboreally fleeing female bonobo uttered eerie, high-pitched screams similar to the calls of gulls. MacKinnon concluded thatbonobos are generally quiet and appear to have no long-range contact call orterritorial signal. 21

First is a hoot series, during which the displayer might pluck a leaf or sprigand place it between his lips. Just before the climax of the hoot series, he risesbipedally and often tears off and flings a handful of vegetation into the air.The hoots merge into a growl as he rapidly tattoos his inflated chest. Thelaryngeal sacs, which commonly extend into the pectoral region, resonatepercussions that can be heard up to 1.6 km away. While slapping his chestwith slightly cupped hands, the displayer might kick sideways with one limband tear foliage with his foot. Then he runs a short distance sideways, oftentearing and slapping at vegetation en route. He might also assault foliage atthe end of his run. As a finale, he sharply smacks the ground with one orboth palms. The entire display takes thirty seconds and often much less.

Marler noted remarkable similarity between the vocal repertoires of goril-las and chimpanzees.192 He matched all thirteen chimpanzee vocalizationswith calls from gorillas, leaving roars, growls, whines, and whinnies as pecu-liar to gorillas among African apes.

Needless to say, such a social pattern coupledwith an even more sophisticated system of communication would serve ourprogenitors very well indeed, especially were they to pool and share collectedfoods, daily experiences, and knowledge of proximate changes in theirhabitat

Individual and commu-nity well-being and survival depend on their ability to distribute themselvesover the landscape in order to forage economically, often while keeping con-tact with other subgroups, and to quickly assess others and adjust sociallywhen subgroups meet before dispersing again, often as part of a subgroupwith a different composition.

Whereas most quiet communication of orangutans is performedintimately by dyads and gibbons interact closely with relatively few individu-als during their lives, chimpanzees are challenged to a remarkable variety ofdyadic and communal interactions with individuals of all ages and bothsexes because of their fission-fusion social pattern

Moreover, there are few differ-ences between the facial musculature of chimpanzees and humans.165Intensive studies on the visual signals, and especially postures and facialexpressions, of captive chimpanzees complement the available field observa-tions, providing a solid ethogram for the species.166 Some of the visual signalsaccompany particular vocalizations, including some that might be necessaryfor the production of certain sounds.

ant-hoots occur in various con-texts. Chimpanzees seem to listen carefully to distant pant-hoots and thenanswer them.

Chimpanzees are perhaps the noisiest and most demonstrative of apes.

Some Bornean and Sumatran orangutans housed socially in three Euro-pean zoos have repertoires of gestures that appear to convey intentionalmeanings. They employed forty gestures predictably to achieve one of sixsocial goals: to initiate contact, grooming, or play, to request objects, to shareobjects, to instigate locomotion, to cause a partner to move back, or to stopan action. Further, they consistently used twenty-nine of the gestures with asingle meaning

Very annoyed females and subadult males uttered lork calls and displayedviolently. Lorks probably also advertise an adult female much as the long callserves adult males.106 Kisses, grumphs, gorkums, and lorks seem to indicateincreasing levels of annoyance.

Less dominant males oftenfled terrestrially or arboreally upon hearing long calls of more dominant males,the foremost of which never moved away from a caller.

few subordinate targets ex-pressed agitation by bubbling, kiss-squeaking, or shaking branches, but theirdisplays were unimpressive

Most (82 percent) of the dominant male’s 157 long calls appeared to bespontaneous: eight followed loud branch or tree falls, six immediately pre-ceded 67 percent of nine copulations, six overlapped playback calls of anothermale, one occurred as he chased another adult male, and three immediatelyfollowed similar chases. He called significantly more often on the days afterhe had been with females than on days before and during bisexual associa-tions and when he was alone.8

In comparison with the problems faced by students of wild gibbon vocaliza-tions, facial expressions, and gesticulations, those of field researchers whowould decipher the semantics and pragmatics of orangutan signals are com-pounded by the solitary lives and dispersal of their subjects

Food sharing in simple societies is both an index of cooperationand a key symbol of what it is to be human. More generally, anethic of communalism and equal access to resource productionis highly developed

At least four tactics are available to a male gorilla that would head his owngroup:• Mature in and remain as a follower until he can head the group of a se-nescent or deceased silverback that may be his father.454• Depose the silverback of an established group.• Emigrate and attract transferring females, perhaps accelerating the pro-cess by raiding established groups

Watts tallied six additional infanticides, one attempt, and three suspectedinfanticides in Karisoke gorillas

In conclusion, bonobos do not merit the epithet pansexual apes. Further,the variety of acts that have been labeled homosexual hardly matches thatachieved by gay, lesbian, and heterosexual human lovers. Many interactionsthat appear to be sexually motivated are in fact greetings, modulations ofsocial relations, and reaffirmations of group cohesion, much as the butt-pats,hugs, and kisses among politicians and athletes are essentially asexual, atleast on the playing field

n wild bonobos, sexual coercion of females is subtle, if it exists at all.

Observations of three adult female and three adult male bonobos at Plack-endael Wild Animal Park indicated that coalitions functioned to maintainexisting ranks, to acquire ranks, to reduce tension, and to test or to strengthenbonds.

White and Burgman further con-firmed that, in contrast with androcentric chimpanzees society, bonobo soci-ety is characterized by strong affiliation among females and between malesand females, but not among males. 305 Party composition varied with partysize. On average, females outnumbered males, but the proportion of malesincreased as party size increased. Cohesion between males and females ap-peared to increase with party size; in contrast with relatively common partiesof females without infants, White observed no all-male bands

Clearly, bonobo society is unique amongape societies and should not be considered a mere variant of chimpanzee so-ciety or an amalgam of chimpanzee and gorilla social patterns.

Like Wamba, in Lomako Forest bisexual groups were the norm. Theytended to contain approximately equal numbers of adult females and males,but as group size increased, females outnumbered males.

Nonetheless,unlike chimpanzees, bonobo males merely engaged in displays toward low-ranging males and supplanted adolescents from feeding sites; they rarelylaunched a severe physical attack.

A young immigrant female would select an olderresident female that she would follow frequently, groom, and engage in frot-tage. The immigrant female was the more common initiator of the affinitivebehaviors that facilitated her integration into the unit-group. Resident malesapproach and mate with immigrant females, which appears to cement theirrelationships with them.

The ranges of neighboring unit-groups overlapped, especially where foodwas plentiful. 275 When bands from different unit-groups came into audiblerange, they avoided one another; this reaction contrasted with parties of thesame unit-group that generally merged noisily within one to ten minutes ofmutual awareness. Direct agonistic engagements were usually averted. Vocalexchanges probably allowed band members to determine whether anotherband belonged to their unit-group and, if not, whether the rivals might besupplanted versus avoided

Grooming competition is positively correlated withdominance rank; the highest-ranking females performed the most displace-ments over allogrooming.

However, excited bonobos tended to beless aggressive than Gombe chimpanzees.

Like Pan troglodytes, Wamba Pan paniscus had unit-groups of 50 to 120individuals that usually foraged in smaller subunits (mean: 17 individuals;range: 2–54 individuals; n = 147 subunits)

Among hominoids, only humans and bonobos have beenverified to make peace with different groups.

After four decades of arguments back and forth, the predominant opinionamong evolutionary biologists and some anthropologists favored the sexualselection hypothesis that infanticide is a tactic of males to ensure that theysired newborn members of their group

During much of our existence, human beings have engaged in infanticide,selective termination, and prevention of pregnancy, in addition to experi-encing stressful conditions that induce abortion.

Only an alpha male appears to gain reproductively from the position thathe obtains via energetic displays, coalitions, and occasional fights or coercion.Other males are reduced to• opportunistic copulation, which apparently is not the most reliable meansfor impregnation,• breeding with young nulliparae that commonly are less fertile thanmature females, and• leading prime breeders away from the group, where copulation mightlead to conception

Watts noted that pairs or trios of top-ranking males in the Kibale Ngogocommunity commonly engaged in coalitionary mate guarding, wherein theywere cooperatively aggressive to prevent periovulatory females from matingwith other males, though they tolerated matings of their coalitionary partners.The male coalitions occurred in large mating parties, perhaps because theycontained too many males for a single male to maintain exclusive access tothe estrous females.

The vast majority of copulations oc-curred opportunistically, and the remaining 25 percent were with possessive,high-ranking males.

At Gombe, consortships accounted for seven of nine (78 percent) concep-tions that occurred during restrictive matings, while only 2 percent of 1,137copulations involved consorting pairs.

Four hypotheses to explain high frequencies of copulation in chimpan-zees, bonobos, and other mammals merit consideration, especially insofar asthey might benefit female fitness.• Best male. Frequent copulation increases competition by males, therebyassisting females to identify partners with high fitness potential.• Social passport. Mating partners might support females in agonistic sit-uations, protect and otherwise care for their infants, and provide accessto food and nest sites.• Paternity confusion. If males are ignorant of paternity, they might pre-vent infanticide of a female’s offspring.• Optimal period for producing next offspring. Females may evince moreproceptive behavior toward copulation.22

In African apes, especially those in fission-fusion societies—namely, chim-panzees and bonobos—reconciliation likely serves to undo the damage thataggression has inflicted on valuable social relationships between aggressorsand victims, but consolation by third parties might not serve the same func-tion. 203

As attested by observations of bite wounds in vivo and on skeletons, chim-panzees can severely injure one another.198 Although they sometimes engagein brutal attacks on strangers and members of their own communities, torestore relative calm in the group most intragroup altercations are followedby one or both of the following interactions: postconflict reconciliation inthe form of an affiliative reunion between opponents, or consolation in theform of postconflict affiliative interaction directed to the recipient of aggres-sion by a third individual

Chimpanzees are exceptional among social mammals, and primates inparticular, in females being the dispersing sex.194 Although chimpanzee fe-males apparently are not averse to mating in the natal group, they are pre-dominantly the individuals to emigrate to other groups, despite the notablerisks of aggression by resident females and the initial or longer-term limitedaccess to the best resources in the new residence because of low social status

Williams et al. concluded that Gombe males proxi-mately defend a feeding territory for their females and protect them fromsexual harassment, though a large range might eventually attract more fe-males.191 At Gombe, the number of males was not correlated with communityrange size

Overt competition among females is usually less dramatic than that ofmales, but females also act violently, even murderously, to maintain theintegrity of the core areas upon which their reproductive success depends

High-ranking femaleshad narrower dietary breadth and higher quality food than the subordinatefemales that foraged less efficiently in lower quality habitats

ndeed, inter-female aggression increased fourfold with the arrival of immigrant females

Analyses of six years of Kanyawara data indicated a relationship betweensubgroup size and density and distribution of chimpanzee food sources;however, food density and travel costs appeared to determine subgroup sizeof males more than that of females. For females, the benefits of being in asubgroup did not exceed the costs, even when ecological conditions seemedto minimize subgroup foraging costs.174

Inbrief, although Gombe male society appeared to be relatively open and fluid,it was not haphazard

The bond between mother and offspringis reinforced not only by tender loving care, including fiddling with the in-fant’s genitalia, but also gentle play which decreases in frequency and mutualgrooming which intensifies as the offspring matures.

There was an alpha, fol-lowed by six others whose ranks varied according to which other males werepresent. Four males were subordinate to the first seven; relative rank amongthem also varied with the presence of specific males

At Gombe, regardless of their location within small groups, dominantindividuals generally regulated the group’s movements, unless they movedrapidly, in which case the leader would be at the forefront. Leaders were notapparent in large, mixed groups and male parties.121 Rank was indicated bypriority of access to favored foods, right of way on the paths, and freedomfrom attack by other chimpanzees. Healthy adult males were dominant overall adult females, and both classes were dominant over youngsters.121 In con-trast, at Kasoje some females were dominant to two adult males

The Reynoldses termed the loosely organized, unstable congeries “bands.”Larger bands (n > 15) occurred during major fruit seasons than when fruitwas scarce (n = 3– 4, and singletons also). They recognized• bisexual adult bands that occasionally included adolescents but noyounger individuals,• all-male bands,• mother bands that occasionally included females unaccompanied byyoungsters, and• mixed bands that were amalgams of mother and all-male bands.

The pre-dominantly adult bisexual groups generally were composed of >20 members,and nursery groups usually included up to 15 individuals. The former werethe more wide-ranging and demonstrative, and individuals freely joinedother groups. Fission and fusion were manifest in their foraging rounds, andthere was no particular order during group travel.

he emerging picture is one of remarkable flexibility, re-sourcefulness, and adaptability by orangutans, surely facilitated by keenperception and cognitive ability and versatile bodies that allow them to ne-gotiate complex arboreal habitats, to find and to subsist on a notable varietyof foods, and to adjust physiologically and hormonally to wide fluctuationsin food availability (females) and the presence of powerful rivals (males). Inbrief, the orangutan is not just elusive—it does not exist. Likewise, becauseof pervasive phenotypic, behavioral and genetic variation the chimpanzee,the bonobo, the gorilla, the gibbon, and the human do not exist

Forty-four percent of 207 copulations were betweenadult females and the subadult males that initiated 99 percent of them. Fe-males resisted 36 percent of their attempts and did not resist in 28 percent ofthem. The status of resistance was indeterminate for the remaining 36percent of cases

Whether an individual engaged in forced or complaisant copulationseemed to be related to male rank and size. Most copulation between sub-adult males and adult females was forced. The alpha male employed sexualcoercion once, but did not force eight other copulations. Four subordinateadult males forced twelve copulations and engaged in seven that appeared tobe consensual.

Schürmann confirmed that Ketambe females prefer to mate with an alphamale.86 He chronicled the sexual history of a young female that initially ac-cepted subadult males but later progressed to higher-ranking mates. As anadolescent, she was keenly interested in an alpha male, but it took nearly fiveyears for her to become his consort. He gradually showed interest in her—for example, allowing her to take bits of fruit from his hand and mouth.When she presented her genitalia close to his face, he sometimes sniffed.Further, she masturbated in front of him and handled his penis. For severalmonths he ignored or gently rebuffed her advances, but finally he consum-mated the union. She manually assisted penile intromission and performedthe pelvic thrusts while he remained inert

MacKinnon’s claim that “adult males also sometimes indulge in rape” isunsupported.76 Prime males mainly attract females and defend their priori-ties via long calls, but as they decline from prime, they often adopt the roleof guardian over their progeny with which intruders might compete forfood.6

That both sexes initiated foreplay is more persuasive evidence forconsortship. His further claim for long-term relationships between adultmales and females, to the extent that bisexual family units exist, is not sup-ported by longitudinal studies on known individuals

At Segama, seven of eight incidents of apparent sexual behavior that MacK-innon termed “rape sessions” entailed aggressive males chasing and assault-ing screaming, resistant females.67 Youngsters of the victims also screamed inaddition to biting, striking, and pulling the attacker’s hair. Males hunchedover the females irrespective of their postures and seemed not to achievevaginal intromission. The most dramatic battle between the sexes began in atreetop and ended on the ground. The eighth copulation was between young-sters.68

Despite relatively high population density, there was no tendency for Ket-ambe orangutans to band together into larger social groups or to coordinatetheir movements through the area with reference to long calls by high-ranking males.

The food-attraction hypothesis has more support; orangutandensity and the presence of nonresidents of both sexes were higher whenfruit was available.44Table 11.1. Estimates of home range size of female and male orangutansHome Range (ha)Study Site Study Area Size (ha) Female Flanged MaleLokan SaEK 390 up to 2,070 65 520Mentoko-1 SaEK 300 40– 60 60–120Mentoko-2 SaEK 300 >150 >FemalesMentoko-3 SaEK Unclear >150 500–700Kinabatangan SaEK 600 180 >225Tuanan CWK 500 250–300 >FemalesSabangau CWK 900 250–330 >560Tanjung Puting CWK 3,500 350– 600 >FemalesGunung Palung CWK 2,100 600 >650Ketambe-1 S 150 150–200 >FemalesKetambe-2 S 350 300– 400 >FemalesSuaq Balimbing S 500 up to >200 >850 >2,500Source: Modified from Singleton et al. 2009 and Utami Atmoko, Singleton et al. 2009.Abbreviations: CWK, Central and West Kalimantan; ha, hectares; S, Sumatra; SaEK, Sabah and

Te Boekhorst et al. tested two hypotheses— sexual attraction and foodattraction—to explain why nonresidents enter areas of the forest that are oc-cupied by longer-term residents.44 Their analysis is based on twelve years ofdata from forty-three identified residents and nonresidents at Ketambe. Thepredominant view has orangutan society comprising one dominant residentadult male, adult females, and subadults and adolescents of both sexes (andpresumably juveniles and infants). Females occupy overlapping home rangescontaining their core areas within a larger home range of the principal breed-ing prime adult male. Alternatively, orangutan society is a community inwhich subgroups of larger traveling bands, centered around an adult male,move in the same general direction and share a range rather that occupyingdistinct parts of it. Nonresidents are transient in both scenarios.

Peer contacts of youngsters in groups appeared to be “as impor-tant, if not more so, than direct genealogical ties in determining female asso-ciations in adulthood.

Untilthe fifth or sixth years, youngsters remain in the vicinity of their mothers.Juveniles eventually lag farther behind, engage in distinct activities at theirown pace, and are seldom groomed. Older juvenile males seem to strike outon their own more readily than do females, who tend to remain near theirmothers.14 It is not known whether this is related to stronger negative behav-ior from the mother toward male offspring

Occasionally she might play with it or groom itbriefly. She introduces it to solid foods by allowing it to sample her meals,including bits from her mouth.During the second and third years, an infant learns to build nests and isincreasingly adventurous while climbing independently in the canopy. Itmight play briefly with an older sibling, often with little body contact; usu-ally mothers rebuff attempts to touch her infant. Often by the time a young-ster is independent enough to play vigorously— at 2.5 to 3.0 years old—nosiblings are available. When a mother/young unit encounters other orang-utans, the latter usually do not handle her infant.13

Why are orangutans so soli-tary? Or more correctly phrased, Given their capacity in some forests to be so-ciable and to travel together for relatively brief periods between fruit sources,why aren’t they more gregarious or group living?5

individual females with dependent youngsters,• lone adult and subadult males, and• various small groupings of adolescents or singletons

The way in which state actors incorporated labor movements was condi-tioned by the political strength of the oligarchy at a time when governmentsrepresenting middle sector groups initiated reform programs to transform pre-existing oligarchic states.

he most important ante-cedent conditions in the authors' argument is the way in which capitalistdevelopment shapes the power of subordinate classes relative to dominantclasses, or what the authors call the "relative balance of class power."

Even without the United States, land reform wasunacceptable to the military, whose entire organizational existence was linkedto the polarized rural economy. Thus, when Arbenz moved to redistribute landand mobilize new rural organizations, the military was poised to seize poweron its own (Handy 1994: 179-90).

argue that the choice of either a radical policy option or a reform policyoption was a crucial component in the creation of these types of liberalism. Ifit could be carried out to its logical completion, a radical policy option wouldlead to radical liberalism, as happened in Guatemala and E1 Salvador. Like-wise, a reform option--when successfully implemented--would lead to re-formist liberalism, as happened in Costa Rica. By contrast, if a particular policyoption could not be successfully implemented, as occurred in Honduras andNicaragua, the outcome would be aborted liberalism

There is substantial empirical support for this argument. In Guatemala, lib-eral leader Barrios faced major challenges from still powerful conservativeactors, and these challenges led him to pursue massive military expansion be-fore implementing land and labor policies (see McCreery 1994: 180; McClintock1985b: 10-11). 7 In 1873, he created Guatemala's first professional army todefend his position in the face of persistent uprisings, and this military mod-ernization later provided a foundation for his enactment of forced labor poli-cies and violent land privatization.

De-spite its initial plausibility, however, this demographic argument overlooks thefact that population density (population per sq. mile) was slightly lower inNicaragua (where a radical policy option was followed) than in Costa Rica(where a reform policy option was followed) for much of the 19th century. 6Indeed, in Costa Rica there was a fairly dense settlement pattern in the CentralValley, precisely the area where the richest coffee lands were located. By themid-19th century, Costa Rican liberals such as Braulio Carrillo could haveeasily promoted the spread of plantations in this area, but they did not.

What factors led liberals to select one set of reforms and not another? Iargue that these policy choices were not determined by structural or demo-graphic factors present before the liberal reform period; instead, I maintainthat immediate political contingencies during the reform period itself best ex-plain liberal choices

The second dimension distinguishing policy options concerns the size ofthe commercial landholding estate promoted by liberals. The extreme case hereis Guatemala, where Barrios made explicit efforts to promote large-scale plan-tations at the expense of peasant farms. In converting community-controlledland into private plots, Guatemalan liberals imposed strict deadlines for occu-pants to buy their land, and they set the price of land acquisition out of reach ofmost occupants (McCreery 1994: 203; Herrick 1974:231). At the other end ofthis dimension is Costa Rica, where liberals showed no preference for largeestates and, in fact, actively promoted small farms. From the outset, CostaRican liberals sought to encourage the productive use of farms by small pro-ducers who already controlled a particular plot of land.

n promoting commercial agriculture and agrarian development, liberals hadto select from a range of specific land and labor provisions. As Table 1 sug-gests, these varying provisions can be grouped into two overall packages ofpolicies: a reform policy option and a radical policy option. Radical and re-form policy options are differentiated along three dimensions: scope ofprivatization, size of estate promoted, and level of state coercion

Hence, when liberals wrested power from conservatives,the Central American countries were poised to undergo a major advance ofcommercial agriculture and primary product exportation

Although by the mid-19th century both political factions drewmembership from the same wealthy classes and concurred on the need for ex-port agriculture (Gudmundson and Lindo-Fuentes 1995), they remained di-vided over the pace and scope of economic change.

In sum, a path-dependent approach emphasizes how actor choices createinstitutions at critical moments, how these institutions in turn shape subse-quent actor behaviors, and how these actor responses in turn culminate in thedevelopment of new institutional patterns

Reactive sequences are chains of temporally ordered and causally connectedevents. They are marked by what Abbott (1992: 445) calls an "inherent logic ofevents," whereby one event triggers another through predictable reaction-counterreaction dynamics

To explain criticaljunctures, therefore, path-dependent analysts often must focus on small events,human agency, or historical peculiarities that lie outside of available theoreti-cal frameworks.

Second, once a particular option is selected, it becomes progressivelymore difficult to return to the initial point when multiple alternatives were stillavailable. Whereas before a critical juncture a broad range of outcomes maybe possible, after a critical juncture the range of possible outcomes is nar-rowed considerably

I argue that theseregime outcomes can be traced back to the 19th century and early 20th centuryliberal reform period. During this period, liberal elites made key choices abouthow to modernize state apparatuses and agrarian economies. Depending onwhether they adopted what I will call a "radical" or "reform" policy option,and depending on whether foreign intervention intersected with the liberal re-form period, different kinds of state and class structures were constructed bythe early 20th century.

This discussion suggests that path depen-dence occurs when the choices of key actors at critical juncture points lead tothe formation of institutions that have self-reproducing properties.

These works havedemonstrated that major outcomes cannot always be explained in terms of short-term processes or unique and predictable equilibria. Rather, trajectories of de-velopment are sometimes punctuated by critical periods in which relatively smallo r contingent events have a profound influence on subsequent events and pat-terns of change.

he article argues thatthe 19th-century liberal reform period was a critical juncture that locked the Cen-tral American countries onto divergent paths of long-term development, culminat-ing in sharply contrasting regime outcomes.

irst reviews the central components of path-dependent explanation, including the concepts of critical juncture and legacy.

Overall, post-war Nigeria exhibited a pattern. Nationalist politicianspreyed upon agricultural exports and directed marketing board surplusesto political cronies, often under the guise of industrial modernization.Credit inaccessibility persisted for export producers, just as it did in Colom-bia. These malignant interventions were ultimately by-products of coali-tional politics, and politically marginalized export producers remainedstarved of credit

Northern villagesresorted to savings cooperatives to foment credit accessibility, but areapeasants had limited deposit potential.

To placate the nationalists, the British held elections in Nigeria in 1951,which launched regional political parties: the Northern Peoples’ Congress(NPC) in the Hausa-Fulani areas, the Action Group (AG) in western Nige-ria, and the aforementioned NCNC in eastern Nigeria (Sklar, 1963: 87–140).Once these parties established control over their respective regions, theybegan pressing the British to form regional (i.e., non-commodity-specific)marketing boards, give party leaders discretion in spending accumu-lated reserves, and create regional government bureaucracies (Ezera, 1960:191–4). The British, lacking the coercive power to resist the burgeoning na-tionalism, relented in 1954. The nationalist leaders then used these region-specific marketing boards to extend patronage to urban constituencies,while neglecting rural inhabitants (Bates, 1981)

Monetary policy was thus zero-sum, as it sustained the Conservativeruling coalition, while harming Liberal coffee producers

N ́u ̃nez used tariffs and lavish spending to solidify the ruling coalition.The key to his side-payment strategy was an inflationary paper moneyregime carried out by the Banco Nacional, founded by N ́u ̃nez in 1880.

The Colombian government also intervened in its credit market amid thecountry’s first major commodity boom – in coffee. Unlike in Chile and Ar-gentina, however, Colombian coffee producers were politically marginal-ized, so their desire for state assistance with credit accessibility went un-heeded. The government harmed them while using loose monetary policyto shore up the ruling coalition of artisans and traditional (non-export-oriented) agriculturalists (Bergquist, 1978: 51). These groups were partof the Conservative Party, which captured power amid macroeconomicproblems in 1880. The party was led by Rafael N ́u ̃nez, who inaugurateda period known as la Regeneraci ́on (the Regeneration). N ́u ̃nez’s ascent tothe Presidency ended three decades of dominance by the Liberal Party,which represented merchants and export-oriented landowners, particu-larly in Cundinamarca and Santander departments, where coffee exportsfirst flourished (Delpar, 1981: 26–38)

n 1872, the government formed a mortgage bank, the Banco Hipotecariode la Provincia de Buenos Aires, which attracted British investment in itsc ́edulas (Ferns, 1960: 370–1). The Buenos Aires mortgage bank and theBanco de la Provincia de Buenos Aires were largely extensions of exportinterests, with wool raisers, cattle ranchers and merchants ‘among (their)organizers, directors, and board members’. Large Buenos Aires ranchersbenefited most from these institutions because they could more readilyand cheaply obtain loans than could smaller pastoralists (Sabato, 1990:273–4)

The Argentine government’s credit market intervention parallels Chile’sexperience: the government aided credit accessibility for exporters in theruling coalition, but rebuked those who were politically marginalized.These groups were, respectively, the ranchers (estancieros) from BuenosAires province, who dominated the ruling coalition, and a competinggroup of ranchers from the neighbouring provinces of Entre R ́ıos andSanta Fe

Government intervention in the credit market reflected the ruling coali-tion’s concerns. In 1855, the state created a mortgage bank, the Caja deCr ́edito Hipotecario. The bank extended borrowers credit up to one-halfof the value of their land at low interest rates. This provision advantagedthe landed elites near Santiago because of the area’s relatively high landvalues, and the bank directed the vast majority of its early loans to Santiagoand Colchagua provinces, the core of the northern central valley (Bauer,1975: 90–6). To assuage the Valpara ́ıso merchants’ worries, legislators dis-allowed the bank from having paid-in capital or lending directly. Thebank instead issued fixed-value, interest-bearing letters of credit, knownas c ́edulas; borrowers then sold c ́edulas on the open market for cash (i.e.,their loan) (Pregger Rom ́an, 1978: 80–3). The Caja was arguably the mostimportant factor in the development of Chile’s commercial agriculture(Edwards, 1932: 150

Concepci ́on. Government policy helped landed elites access credit via anew mortgage bank, while southern agriculturalists received little benefitand had to turn to commercial banks, which lent at higher rates of interest.The state’s actions aimed to confer a distributive economic gain on thenorthern central valley landed elites, who formed the core of the rulingcoalition

Thus, because credit is relatively scarce throughout the developingworld (and especially so in prior historical eras), its extension is at oncea political as well as an economic issue. Those who receive credit ob-tain an opportunity to increase their economic power – and, therefore,may subsequently enhance their political power vis-‘a-vis other groups.

n short, government officials are sensitive to the distri-butional potential of their policy choices and craft policies that serve theirpolitical supporters.

The failure to improve credit accessibility not onlyinjures exporters, but, arguably, harms the state as well

Credit accessibility problems can provoke export-oriented actors to seekgovernment intervention in the credit sector.

Second, exporters could possibly meet their credit needs by turning toforeign lenders, including foreign merchant firms operating domestically.Foreign sources of capital have often been difficult to obtain, however.

Agrarian societies also regularly lack the financial andother risk-bearing institutional frameworks that can foster capital accu-mulation. They are missing an efficient institutional means to link saverswith investors, in other words (e.g., Gerschenkron, 1962: 45; North, 1990:125). In short, export-oriented actors will frequently have to look beyondprivate domestic sources for credit

First, they could acquire credit from private domesticsources

redit availability regularly proves difficult in agrarian soci-eties, however. Indeed, the contemporary fascination with micro-lendingunderscores just how difficult it can be for rural producers to readily obtaincredit, and such problems were acute in the historical eras examined here.

ommodity booms are bountiful times for exporters, and they are gener-ally good for governments, too, as export growth provides new revenuefrom consumption and fiscal linkages (Hirschman, 1977)

I analyse how commodity booms naturally spur economic ac-tors to try to exploit favourable world market conditions, which regularlygive rise to new demand for credit

The coalition upon which a govern-ment relies matters greatly, perhaps especially in weakly institutionalizedcontexts.By undervaluing the centrality of coalitional politics, mainstream studiesof the resource curse problematically presume that rent capture by thegovernment is both automatic and sequestered to mineral exporters. Yet,rents only accrue to the state because of prior policy decisions that haveeither nationalized or imposed rapacious taxes on mineral production(Karl, 1997: 48–9; Ross, 2012: 4–5

he emphasis placed on the moderating influence of strong institutionsimplies that, in their absence, governments will be infected by rent-seeking.Certainly, there is a noteworthy confluence of mineral export wealth andpredatory government behaviour (Karl, 1997: 236–7). But simply analysingwhether countries possess strong institutions at the onset of boom timesoverlooks those well-institutionalized countries, such as Australia, whichbuilt effective institutions while reliant on mineral wealth (Kurtz, 2009:482–4). Botswana, Chile and Mauritius are examples of developing coun-tries that strengthened institutions while simultaneously relying on re-source exports. One differentiating aspect of these countries, comparedto archetypical cases of the resource curse, is that they featured rulingcoalitions which included export-oriented actors.

This measure strives to probe whether a country enjoys state capacity, orwhat Michael Mann (1984: 189) calls infrastructural power: ‘the capacityof the state to actually penetrate civil society, and to implement logisticallypolitical decisions throughout the realm’

I arguesomewhat against a prevailing world-view that countries should simply‘remove politics’ from resource wealth management. It seems that a morefruitful approach should consider coalitional politics in order to gaugehow various strategies of resource management will be consonant, or inconflict, with the side-payment strategies that bind prevailing coalitionstogether

These cases were selected because they display variation in my indepen-dent and dependent variables (political coalitions and government creditpolicy, respectively).

In sum, ruling coalitions in all four countries regardedcredit as a zero-sum resource. They avoided channelling credit to polit-ically marginalized actors because to do so would have amounted to arelative redistribution of economic power away from coalition members

When states intervened in credit markets, they did so inways that worked to the distributional advantage of coalition membersand at the expense of politically marginalized exporters. In Chile, forinstance, the powerful landed elites wanted the state to authorize thecountry’s first formal banking and joint stock institutions to facilitate theirwheat exports. The government complied by creating a mortgage bankto direct credit to the landed elites on favourable terms. It simultaneouslyforestalled the analogous aims of small-scale wheat exporters in the southand copper mine owners in the north, who were not part of the rulingcoalition. The Argentine state likewise toiled to funnel credit to politicallypowerful ranchers in Buenos Aires province, but ignored pleas for creditfrom competing ranchers in the country’s other littoral provinces. Thecredit policy was purposefully prejudicial in both countries.

Overall, this paper helps to extend existing lines of inquiry by emphasiz-ing a non-institutional factor – coalitional politics – to account for variantgovernment responses to resource booms.

This papermaintains that the composition of ruling political coalitions underlies suchcontrasting attitudes on the part of governments. I illustrate this argumentby looking at how coalitional politics guided policymakers’ responses toresource booms in four countries. In particular, I analyse how commoditybooms led exporters to seek state assistance in obtaining capital. I findthat the way in which policymakers reacted to their requests hinged onwhether exporters were part of the ruling coalition

One prominent study finds thatcountries with considerable commodity exports and weak institutions doindeed feature poor economic growth, but countries with stronger institu-tional profiles exhibit no significant association between resource wealthand growth (

The disap-pointing economic growth associated with abundant natural resources isperhaps the best known facet of the ‘resource curse’. Researchers specu-late on a range of causes for this comparatively sluggish growth, including‘Dutch Disease’ 1 and countervailing protectionist trade policy (Sachs andWarner, 1995), low investment in human capital (Gylfason, 2001), rent-seeking (Karl, 1997) and myopic borrowing by governments experiencingresource booms (Manzano and Rigob ́on, 2007; Nooruddin, 2008).

These findingssuggest that the resource curse may pivot on coalitional politics in importantrespects.

Where exporters were members of the ruling coalition (Chile and Argentina),their needs sparked credit sector reform and government help to accesscredit

This paper argues that certain political coalitions can serve a similarfunction in the context of weak institutions. To explicate this argument, thispaper examines how international commodity booms regularly create creditdemand that surpasses available supply, often impelling exporters to seekgovernment assistance with obtaining credit.

Scholars of the ‘resource curse’ increasingly agree that strong institutions canhelp countries avoid the pitfalls associated with abundant natural resourcewealth

In agrarianpostcolonial societies, types of war and the type end scope of mobilization oftheruralpoor during stateformatinn shapedinstitutions, ciail-military rela-tions, and regime outcornes.

Europe taught us a similar lesson: conflict and political developmentsbefore the switch to commercial agriculture determined outcomes

Did the expansion of commercial agriculture affect democracy? Thetiming of the switchtoward commercial agriculture was a crucial factorin Barrin5on Moore's model. Rueschemeyer, Stephens, and Stephens(t9gz:8) have emphasized even more strongly the empowerment of thesubordinate classes as a necessary factor for democracy in Latin Amer-ica and Europe, arguing that the working class was the most "consis-tently pro-democratic force" and much more "insulated" from the he-gemony of the dominant classes than the rural lower strata. Theseau*rors' strong emphasis on the switch merits a quote: "If no signifi-cant export expansion occurred, no movement towards democracyoccurred either, and if expansion started late, movement toward de-mocracy was delayed. . . . Democracy depended on the sequences ofsocial structural and political institutional developments set in motionby export expansion."

The point is thatdespite frequent instances of corruption and clientelism, Latin Ameri-can "democracies" resembled the liberal democratic model more thanany other.

Unlike the other two cases, Buenos Aires was able tolaunch aggressive military campaigns to dominate the hinterlands.State and army developed a close partnership, which ultimately en-couraged more frequent military intervention and the formation ofmovements rather than parties.

ndependence, in many ways, repre-sented the "ruralization" of the pollty; cities remained militarized, butthey were forced to share military power with rural areas. In our cases,we can identify two distinct patterns. \7hen the recruiflnent of militiasand the formation of the army decentralized the power of the city, themilitary tended to develop a loose sense of autonomy and did not nec-essarily identify with the central power. The state grew weaker and theparties stronger, as inUruguay and Colombia. \il(rhenthe armyretainedsome degree of autonomy and military power remained associatedwith the urban gentry, the state grew stronger and the parties weaker, asin Argentina

7hen one looks into individual cases, it is clearttrat despite a corunon reaction against Spanish domination, the dif-ferent timing of independence responded to a complex set of localfactors.3a This book views the period of independence as a time ofpolitical revolutions that set different scenarios of institution building.3sIn other words, state formation can be understood as a process ofpostrevolutionary reconstruction.

Some of the entitlements granted to themilitary endured, such as the fuero militar, a controversial privilegegranted during the colonial period that exempted officers from theofficial system of justice, and that, wittr slight modifications, was main-tained after independence.3

nlike in Europe, where monarchs frequently hired for-eign mercenaries, in our cases, the army recruited almost exclusivelyfrom the local poor, in a context of labor scarcity.

The Americas differed from Europe in thetiming of party formation in relation to state building. In Europe, par-ties arrived later on the political scene, when states were already strongand consolidated, which places North America and Latin Americacloser together in the larger picture of state formation

Despitethe influence of a European military ideology that stressed the advan-tages of a strong central army, after independence, the Latin Americancases studied here deviated from that model. As in the Ottoman case,power centralization took place "mostly through negotiation and [the]incorporation" of armies that had emerged in revolts during prior at-tempts at centralization, going back to the late eighteenth century.2

Like Ottoman state makers, Latin Americans also struck flexibledeals with local patrons and militias who occasionally played the role ofbandits, not in the romantic style of EricJ. Hobsbawm (1981) butmoreinthe crude sense described byBarkey (rSSS).Armybuilding includedthe flexible absorption of dissident "generals" who often resorted tobanditry to maintain their militias, but who were opposed by the centralpower-whenever convenient-in order to defend property rights andthe safety of the local population.

excludedgroups (mestizos, blacks, and Indians) into armies and parties, as wellas the connections beNveen the army and the political elite, contributeto explain paths of state making.

State autonomy in each case examined here remained directly tied toaspects of army recruitrnent and composition, and to the constructionof lines of authority between officers, parties, and state makers.

These contrasting scenarios of rural insurrection beg further treat-ment of at least three key questions about state making in Latin Amer-ica. The first is, \7hy were alliances between the central power and thelower classes so infrequent? The second and related question is, ril7hydid local landlords and the rural poor tend to develop warring associa-tions at the national level, with enornous consequences for partyformation? The third is, \(rhy were these rebellious alliances almostalways alliances of class groups from very different sectors of the econ-omy that, at times, obtained part of their strength from the urbanmiddle class?

Most literature has pictured power centralization as a game inwhich three major actors-the state, ttre upper classes, and the lowerclasses-coalesced or clashed, determining the relationship between thestate and civil society and influencing the institutional design finallyadopted.l

for instance, offered a lesson familiar to state makers inLatin America: the failure of the rebellious nobility to coalesce againstthe central power accelerated power centralization.