A federal bargain is struck, that is, when two conditions are met.First, there exists a desire on the part of those offering the bargain toexpand territory by combining constituent governments into a new po-litical entity in order to secure a public good such as security or a com-mon market. Second, for those accepting the bargain, there must besome willingness to sacrifice political control in exchange for access tothe public good provided by the new federal government.1

tates are formed to secure public goods such as commonsecurity and a national market, but at the moment a federal state isfounded, a dilemma emerges. How can a political core be strongenough to forge a union but not be so powerful as to overawe the con-stituent states, thereby forming a unitary state?

federation is never the fi rst choice of all members in a group, and those members with dispropor-tionately high potential relationship-specifi c investments prefer federa-tion to IO, while the others prefer an international organization as their fi rst choice

he failure of the Federation of the West Indies initially led to the col-lapse of regional cooperation, but eventually Jamaica and Trinidad took the lead in developing several regional organizations, including, eventually, the Caribbean Community (CARICOM)—an IO that governs a reason-ably integrated common market—as well as several specialized agencies that administer policy coordination on diverse subjects, including health and education. Formed in 1973 and including Jamaica, Trinidad and To-bago, and many of the smaller states, CARICOM has increasingly taken on more functions, including external economic relationships, and the member states (as of 2008) were in the process of adopting a common passpor

Trinidad, unlike Jamaica, had oil, and even before the rise in oil prices in the 1970s, this resource gave the island economy access to inexpensive energy and to a steady stream of revenue that it could use to fi nance in-dustrial development—in particular industrial development related not just to oil drilling but also to modern tanker and container shipping. By 1960, as a result, Trinidad had a denser industrial base than any of the other islands and a greater interest than any of them in forming a regional customs union that would have a common external tariff.

Manley and profederal newspapers mostly defended federation as a means to achieve administrative gains: a larger state, drawing on a broader tax base and pooling the talents of elites throughout the region, would be more capable of making strategic invest-ments in new and growing industries and better at achieving the newly independent states’ ambitious goals for agricultural development.

The constitution had been designed to give national politicians incen-tives to invest their careers in federal politics, and so leaders were pro-hibited from holding elective offi ce at both the state and the federal level simultaneously. The idea was that, as in other federal systems, the drive to organize parties around the goal of maintaining federal offi ces would serve as a mechanism to keep leaders engaged in the federal project. In the 1958 elections, however, Jamaica’s main nationalist leaders did not run for federal offi ce, indicating their unwillingness to invest in the federal institution. Although the federation existed on paper, then, a true federal system was never actually implemented

British economic planners after World War II anticipated that the West Indian colonies would eventually become independent but were con-cerned that the new states would be too small to be economically viable. Following what they saw as the successes of federal integration in Austra-lia and Canada, they suggested that regional federation in the Caribbean would lead to better growth through a larger common market and a better ability to coordinate development projects such as health and education infrastructure. I

The unraveling of the federal proposal happened in two stages. First, after not quite three years of operation, Jamaica opted out of the federal union. Then, though the remaining states considered continuing as a fed-eral union, eventually the project collapsed when the next largest island, Trinidad, changed its position and, following Jamaica, exited the union. After Trinidad left, the project came to an end. Two separate questions merit answers. First, why did Jamaica opt out despite the economic ben-efi ts that, in principle, federation could have brought? Second, why did Trinidad’s leaders switch from supporting a strongly centralized federal union to opposing federation as soon as Jamaica opted out?

Tanganyika had originally been profederal because it understood that it and Uganda would be the vulnerable states in any system of economic cooperation (the third and fourth observable implications of the theory). With Uganda on its side in the federal parliament advocating for the con-tinuation of transfers, even though Kenya would have a slight plurality and would likely be represented in the ministry, Tanganyika would be as-sured that Kenya’s political system would be tied to the federation by a federal party system that would force Kenyans to forge ties with elites from the other states in order to maintain power. After Uganda left, however, that calculation became trickier. It was no longer clear that Kenya’s elite would be forced to maintain ties to Tanganyikan parties, and there was therefore little reason for TANU to have confi dence in the maintenance of the union

Ultimately, Obote was not convinced that even the federal institutions Kenya seemed willing to accept would prevent an erosion of Ugandan in-terests over the long run. With real, but only moderate, potential benefi ts to federation, and a danger of long-term exploitation, Obote chose to pursue self-suffi ciency, competing with Kenya for investment rather than tying itself to an uncertain partner

In June 1960, Nyerere put East African federation at the top of the political agenda by publicly proclaiming it as a goal. His argument was that federation would bring the benefi ts of development and integration while preserving the independence of the region from the great powers; thus he couched the move to federate in nationalist terms (Nye 1965; Rothchild 1968).

The initial push for federation happened during a time when there was a strong consensus among elites that regional economic integration could, in principle, help the leaders achieve their developmental goals by increasing the amount of foreign investment capital the states would be able to attract. These gains, however, involved relationship-specifi c assets. Foreign investment in the region was expected to be concentrated around Nairobi because of its advantage in both education and transportation; nationalist leaders proposed a system of tax redistribution by which Kenya would pay Tanganyika and Uganda out of the subsequent revenues. But this unequal industrial growth (which would be oriented toward export-ing goods from the region) not only would create a regional dependence on Kenyan industry but also raise the potential benefi ts to Kenya of exit-ing an agreement

states with unequal potential investments in relationship-specifi c assets can face commitment problems that make cooperation impossible unless they enter into a federation

These institutional exit costs are low or even negligible in an international organization but are higher in a federation. In federations, they arise from the institution-specifi c investments that political parties and individual politicians make in securing power at the federal level, inve

They have, of course, some drawbacks as well. For one, the cases are not randomly selected but chosen on the basis of variation on the dependent and independent variables, at least within a certain range

They can, however, use po-litical institutions to create what I call contrived symmetry, artifi cially en-suring that all sides have something to lose if cooperation ends. Because federal agreements make exit costly for all members, vulnerable states are put at less of a bargaining disadvantage. These states have an incentive to refuse to cooperate in international organizations, and when cooperation is valuable, their potential partners have an incentive to accept a federal constitution as the price of cooperation

In the literature on federation only one study explicitly asks why states form federations instead of alliances. Peter Ordeshook and Emerson Niou (1998) suggest that an inability to commit to a potential alliance partner may drive states to form federations instead of alliances for security when the military stakes are high.

A complete theory of federa-tion, therefore, cannot logically rest only on the observation, however true, that the size of a region for which a public good is produced matters

Size can have drawbacks as well, however. There may be declining re-turns to scale as a state’s administrative apparatus grows past a certain size (Lake 1992). Alesina and Spolaore assume that because citizens have di-verse tastes for public goods, in a large state many citizens will end up with public goods that are not well suited to their tastes; they conclude that states past a certain size will be ineffi ciently larg

These three examples illustrate the strengths of existing scholarly the-ories about the origins of federations but also reveal a key problem: states do not have a binary choice between federal union and complete indepen-dence. The European states and China and Taiwan can get the benefi ts of economic integration without forming a federal union, and Iraqi regions could in principle cooperate on security without a reconstituted Iraqi state. Nevertheless, most scholars explain federation simply by noting that there are gains from cooperation: William Riker (1975) and Murray Forsyth (1981, 160) maintain that states federate when they can be made more secure by pooling their resources.

An early federal theorist, Albert Venn Dicey, concluded that the condi-tions in which states choose federation are that “they must desire union and must not desire unity” (1885, 141).

It is, rather, a prior choice that states make about whether to have a common government at all (

Second, although the formal constitutional features of federal systems differ, one necessary feature of a stable federal union is a party system that joins local and national politicians (Truman 1962; Wildavsky 1967), since parties infl uence the incentives of politicians and other elites t

I also explain how states in pursuit of their individual interests can reach a federal agreement in the fi rst place. States that would be at risk of extortion in a relationship avoid it by turning the extortion around, demanding up front, as the price of cooperation, political integration that is as deep as they can get from their partners. Their partners give in when the benefi ts of cooperation are great enough that they are willing to sub-ordinate themselves to political integration, thereby getting otherwise un-attainable economic or security gains

tes can get the benefi ts of size in many other ways, however, including cooperation through international organizations such as cus-toms unions or alliances. A complete theory of federation must explain why states choose to federate when there are several other possible options.

The Empire Trap also highlights the growing role of investor insurance in allowingforeign direct investors to hedge against a variety of risks. Organizations such as theOverseas Private Investment Corporation (OPIC) and the Multilateral InvestmentGuarantee Agency (MIGA, bankrolled by the World Bank) pay insured companies ifthey are expropriated and undertake sometimes heroic efforts to recover the moneyfrom the host country, with the backing of the U.S. government. Using this tool, IOCscan even securitize themselves against changes in contracts and currency volatility.

Host governments can therefore exploit this phenomenonand pounce when the time is right. Especially when the oil price rises, governmentsface incentives to increase royalties and introduce windfall taxes, if not expropriateforeign oil firms altogether.

Plants, machinery, and storagefacilities associated with oil exploration, drilling, and transportation can be confis-cated quite easily. Investors must make expensive investments in these assets severalyears before the oil starts flowing and several years before they can realize a returnon their investment.

ts endemic weakness.Indeed, Subterranean Struggles points to the possibility that the myriad prob-lems faced by the citizens of weak states in the Latin American periphery are ulti-mately due to the state’s neglect or incompetence and not to natural resources perse. It argues that “at their core, most struggles over extraction reflect demands for astronger state—for stronger regulatory presence, for planning, for protection ofhuman rights and environmental assets, and for predictability in the lived environ-ment of rural populations” (24)

The rulers of weak states that attempt to consolidate power and raise rev-enues when commitment problems are thorny and fiscal transaction costs high turnto the deliberate construction of an oil sector to accomplish both tasks simultane-ously.

I adduce evidence for several empirical implications generated by this theory.Revenue-starved states with low capacity explore for oil more intensively. Theycourt larger volumes of capital in the oil sector. They produce oil rents at greaterrates by goosing the production of extant wells. They also export and tax it at higherrates. These results hold after controlling for geological endowments, oil prices, andproduction costs, among other factors

Ross’s Oil Curse is not your grandfather’s resource curse theory. The booktranscends, if not usurps, the older rentier state view, which evolved from thestudy of state building and democratization in Western Europe. This paradigmholds that if a ruler wishes to collect revenues by taxing the citizens, the ruler mustmake concessions in exchange for their compliance: provide public goods andgrant them representation (Tilly 1975). This is particularly likely when capital ishard to tax because it is liquid and mobile (Bates 1991). It therefore follows thattaxation of modern, formal sectors, composed of investors, salaried white collarworkers, and wage laborers, is more likely in the absence of natural resources (Ross1999)

widely known that there is a negative correlation between natural resourcereliance and political and economic developmen

If African states could avail themselves to ways of regulating flows of people, money, or goods from neighboring states, they might be able to strengthen their presence in outlying areas Coithout depending on the rather expensive deployment of coercion. In fact, buffer mechanisms were Jargely absent in Africa:

ather, the opposite appears to be true: states rose and fell, expanded and contracted, largely in relation to the amount of coercion they were able to broadcast from the center

Wilks the Ashanti conceived of their empire as radiating out in all r twenty days walking or roughly a month for a messenger to reports that directions fo m the capital Kumasi and return from the, frontier. Any farther d too difficult to administer because thé lag in responsetime Mg MNS FY AZAR AMASSIAK PHS depart fro was deeme : : 7 ; was too great.

orrespondingly, it was natural that many precolonial African states were exceptionally dynamic. Political organizations were created, rose, and fell naturally in response to opportunities and challenges

Of course, the slave trade only contributed ra ont on that true power was not synonymous with the control of to me Ieny African states continually tried to make forays into the inte- lane ue red the continual demand for slaves but did not try to extend an 4 ontrol into those areas because it was not profitable

In many precolonial societies, a significant difference existed between what control of the political center meant and what the partial exercise © power amounted to in the hinterland. Formal political control in pre- colonial Africa was difficult and had to be earned through the construc- tion of loyalties, the use of coercion, and the creation of an infrastructur:

The demographic and political realities in precolonial Africa had a pro- found effect on the contours of the states. In particular, as control of particular pieces of land was not critical to African societies, “there was no indigenous map-making in Africa or . . . its presence is so insignificant as to justify that generalisation.”

In addition to the sheer abundance of land, African farmers also de- pended almost completely on rain-fed agriculture and therefore invested little in any particular piece of territory. As a result, few specific pieces of land were that valuable or worth a great Price to defend. African agricul- ture was extensive, with low fixed investments in any particular piece of land, because the fundamental tool of intensive farming, the plough, spread throughout Eurasia without ever reaching Africa (except, it should not be a surprise to learn, Ethiopia). Goody describes the results of the failure to develop the plough:

In fact, states and state systems can be discerned in precolonial Africa i preconceptions based on Europe’s recent history are discarded. Affican states did broadcast authority, did have firm notions of what consolida- tion of power meant, and did develop conventions for relations between states.

focus on control of land as the basis of state authority is not surprising in Europe where, due to population densities, land was in short supply. Indeed, an assump- tion that land was scarce, and therefore control of territory was an impor- tant indication of power, underlies much of Weber’s own analysis.* Sim- ilarly, control of territory is the basis of international understandings of state authority since states in Europe fought about land. Thus, Gilpin argues, “The control and division of territory constitute the basic mecha- nism governing the distribution of scarce resources among the states in an international system.”*

I will demonstrate that, fundamentall there is nothing exotic about African politics. Rather, as elsewhere/ politi cal outcomes are the result of human agency interacting with powerfu geographic and historical forces,A

However, most scholars would also agree with Crawford Young that the imposition of colonial rule changed) everything:

orris Silver develop a model of state development where the state expands until average costs are rising, an adaption of venerable cost models in economics. However, this model ignores the fact that by changing the nature of the international system, states can profoundly alter the costs of gaining and consolidating control over land. Since inter- national society determines what state control actually means, to not in- vestigate the nature of the state system is to fail to grasp the fundamental dynamics of state consolidation. Of course, pure economic models of state expansion cannot incorporate such overarching design considera- tions because they are based on the assumption that states are system- takers, just as firms are assumed to be price-takers, and therefore unable to affect the economic system within which they operate.

In fact, the system of territorial boundaries, as this book demonstrates ~ in détail, has been critical to the particular patterns of African state con- solidation and has been seen as a tremendous(asset by African leaders, both in the colonial and independence periods. Far from being a hin- drance to state consolidation, African boundaries have beefyperhaps.the - _.critical foundation upon which leaders have built their state$. In addition, the territorial boundaries help shape other buffer institutions that also insulate polities from international pressures. These other buffer institu- tions vary considerably but include currency exchange mechanisms (which define the means of conversion between domestic and international units of money) and citizenship rules (which determine the difference between citizens and foreigners). Chapters seven and eight demonstrate that these boundary systems have also had significant effects on African “state consolidation.

African leaders were radically different than those made by Europeans: in Africa, conquest and consolidation must be understood as different pro- cesses, with different cost structures, because wars were primarily not over territory and the end of wars did not leave the organizational and infrastructure residual that was typical in Europe.

s, African leaders seeking to extend their power face a vari- ety of costs: some inescapable in the short-term associated with conquest; some that are contingent on the scale of the changes being brought about; and some that can be postponed to the long-term that are largely associated with the consolidation of authority.”

ll leaders face costs when trying to expand their writ of authority

I suggest that state consolidation in Africa can be understood by examining three basic dynamics: the assessment of the _ amuics: “costs of expansion by individual leaders; the nature of buffer mechanisms ~ established b by the state; and the nature of the¢egional state system. Only by understanding all three levels is a complete analysis ofthe consolida- tion of power in Africa possible.{I

This question goes to the very essence of politics because, as Weber and others have repeatedly noted, the signal characteristic of a state is its monopoly on the legitimate use of physical force in the territory it is said to control.” It is also an issue that has been analyzed almost entirely by focusing on Europe. Scholars have concentrated on the European experience when trying to understand state development despite the fact that Europe contains only a small per- centage of the states formed throughout history. In part, this myopia is due to the fact that the rise of European states is well documented.

In particular, African states have never had the security im- perative to physically control the hinterlands in the face of competition from hostile neighbors.(Since the external imperative for capitals to con- solidate authority was largely absent, African leaders have had to devise an entirely different set of strategies to exe control ¢ over their territories.

Due to low population densities and the large amount of open land in Africay4yars of territorial zonquest, as chapters two, three, and four will HISEiiSs seston At seldom been a Significant aspect of Gealite, not ant which was available fo 5 al¥. In contrast to European states that, ; at léast at some points in their. istories, needed to mobilize tremendous resources from their own populations to fight wars and were therefore forced to develop profound eh pel their owx hinterlands, precolonial African leaders mainly exploited™ people outs their own polity because the point of war was to take women, cattle;‘and slaves

Even the wave of democratization that swept Africa in the 1990s has not breached the center-periphery divide. The revolts since 1989 against African authoritarianism were largely urban affairs, with little participa- tion by any organized rural group.** Not surprisingly, few if any of the political parties that have come into existence since 1989 have strong ' rural roots.” It still appears too difficult to organize the peasants qua " peasants, despite the fact that spatial location is an excellent determinant ties of life chances in much of Africa.

den argues that because “the state does not really enter into the solution of his [the African peasant’s] existential problems” there is‘“a definite limit . . . to how far enforcement of state _ policies can go in the’ context of peasant productio n.?

As Robert H. Bates documented, African politicians traditionally @quated their political survival with appeasing their urban populations via subsidies even if the much larger, and poorer, rural populations had to be taxed.”

The Europeans, after formally colonizing Africa in the late-nineteenth century, did create many urban areas. However, these cities did not serve as the basis of state creation in the same mannet\as occurred in Europe_ because the colonizers were not interested in dup dcating the power infra- structure which bound city to hinterland in their homelands. Rather, the cities were mainly designed to service the needs of the colonizers. Partic- ularly telling are the location of the capitals the colonialists created. By 1900, twenty-eight of the forty-four colonial capitals were located on the coast, demonstrating the low priority of extending power inland com- pared to the need for easy communication and transport links with Eu- rope.” R

Successful European state development was therefore characterized by profound links between the cities#—the core political areas—and the sur- rounding territories. Indeed, the growth of states was Closely correlated with the development of Significant urban areas. As Tilly has argued, “The commercial and demographic impact.of cities ‘Made a significant _ difference to state formation. .. . The existence of intensive rural-urban ~ trade provided an opportunity for rulers to collect revenues through cus- toms and excise taxes, while the relatively commercialized economy made it easier for monarchs to bypass great landlords as they extended royal power to towns.and viltages.”’*

Because European states were forged with iron and blood, it was criti- cal for the capital to physically control its hinterland. Tilly notes, “as rulers bargained directly with their subject populations for massive taxes, military service, and cooperation in state programs, most states took fur- ther steps of profound importance: a movement toward direct rule that reduced the role of local or regional patrons and places representatives of the national state in every community, and expansion of popular consul- tation in the form of elections, plebiscites, and legislatures.

Charles Tilly notes that one of the central reasons for the creation of relatively centralized state apparatuses in Eu-_ rope was the “continuous aggressive competition for trade and territory among changing states of unequal size, which made war a driving force in European history.”!°

In Europe, . through the fourteenth century, population densities were not high enough to put immediate pressure on land and compel territorial competition. As ~ Mattingly notes, “In the fourteenth and fifteenth centuries, the continental space of Western Europe still impeded any degree of political organization efficient enough to create a system of continuous diplomatic pressures.

of issues“~when /building their states: the cost_of€xpanding the domestic power infra- structure; the nature of ational, boundaries; and the“design of state sys- _tems.

irst, African countries have’quite varied environmental conditions, Ecological differences across provinces of a country in West Africa, which can be coastal, forest, savan- nah, or near-desert, are greater than in any European country.® Therefore, the models of control an African state must develop for these highly differentiated zones are more varied, and thus more costly, than what a government in Europe or Asia must implement in order to rule over their more homogenous rural areas.

(Retail low population! densities in Africa have automatically meant that it always has been mor expensive for states to exert control over a given number of people com- pared to Europe and other densely settled areas.

HE FUNDAMENTAL PROBLEM facing state-builders in Africa—be they pre- \ colonial kings, colonial governors, or presidents in the independent era— has been to project authority over inhospitable territories that contain - £elatively low densities of people.

This findingsuggests that infants’ morally relevant own behavior (altruistic vs.selfish) is tightly linked to their third-party evaluation of morallyrelevant situations: altruists pay attention to normative (moral)issues of fairness, whereas selfish infants are interested in non-moral physical aspects of social interaction

Twenty-six out of 38 infants(68%) shared one of the toys: 12 infants shared the preferred toy(32%; ‘‘altruistic sharers’’), 14 infants shared the non-preferred toy(37%; ‘‘selfish sharers’’), and 12 infants did not respond at all(32%; non-responders)

possibility that a decline in attention over the course of theexperiment led to differential findings across the test and post-testphases.

Infants’ mean looking times to the test and post-test outcomes inthe VOE paradigm are depicted in Figure 2. Infants’ expectationsin the VOE paradigm were assessed by computing a repeatedmeasures analysis of variance (ANOVA) on averaged lookingtimes with phase (test vs. post-test) and trial-type (fair/symmetricalvs. unfair/asymmetrical outcome) as within-subjects factors. Thisanalysis yielded a significant main effect of phase, F(1, 46) = 12.52,p,.005, gp2 = .21, and a significant interaction of phase and trial-type, F(1, 46) = 4.68, p,.05, gp2 = .09.

We investigated 15-month-old infants’ sensitivity to third-partyfairness using a resource distribution task in a violation-of-expectation (VOE) paradigm, and infants’ explicit behavioralresponses in a sharing task, in which they could choose to shareresources with an unfamiliar adult altruistically (share a preferredtoy), selfishly (share a non-preferred toy), or not at all. By assessingfairness and altruism in infants via both a violation-of-expectancyparadigm and behavioral measures, we sought to empiricallyvalidate the hypothesized theoretical interdependence betweenthese two constructs [22,24], and to understand the underlyingnature of infants’ potential fairness expectation

nd that aversion to unfair offers is stronglyrelated to amygdala activity indicating an automatic emotionalresponse to unfairness [

We investigated 15-month-old infants’ sensitivity to fairness, andtheir altruistic behavior, assessed via infants’ reactions to a third-party resource distribution task, and via a sharing task. Ourresults challenge current models of the development of fairness and altruism in two ways. First, in contrast to past worksuggesting that fairness and altruism may not emerge until early to mid-childhood, 15-month-old infants are sensitive tofairness and can engage in altruistic sharing. Second, infants’ degree of sensitivity to fairness as a third-party observer wasrelated to whether they shared toys altruistically or selfishly, indicating that moral evaluations and prosocial behavior areheavily interconnected from early in development. Our results present the first evidence that the roots of a basic sense offairness and altruism can be found in infancy, and that these other-regarding preferences develop in a parallel andinterwoven fashion. These findings support arguments for an evolutionary basis – most likely in dialectical manner includingboth biological and cultural mechanisms – of human egalitarianism given the rapidly developing nature of other-regardingpreferences and their role in the evolution of human-specific forms of cooperation

Our findings add to this growing literature and suggest that inthe second half of the first year, infants’social evaluations, likethose of adults, become based on more than rigid and simplisticrules of “if helpful, then positive; if unhelpful, then negative,”butrather depend crucially on the contexts in which such behaviors areperformed.

Our toddlers were willing to ap-proach (rather than avoid) individuals who had behaved antiso-cially, overcoming their aversion to antisocial others (4–6, 42) todirect a negative behavior toward them.

To sum up, toddlers asked to perform a positive actionpreferentially chose the Prosocial character as their target,whereas those asked to perform a negative action preferentiallychose the Antisocial character as their targ

. Thispattern of choice differs significantly from 8-mo-olds’choices inthe Antisocial Target condition of experiment 1 (Fisher’s ExactTest, P <0.005).Taken together, results from experiments 1 and 2 suggest that by8 mo of age, infants are capable of evaluating third-party socialbehaviors in a manner that goes beyond simple “prosocial = good;antisocial = bad”judgments. They positively view not only thosewho behave positively toward prosocial individuals, but also thosewho mete out negative treatment to antisocial individuals

Thus, by 8 mo of age, our subjects preferred individuals whoacted positively toward prosocial others and preferred those whoacted negatively toward antisocial others. These results suggestthat by this age, the value of a social act is not determined solely byits positive or negative effect upon a recipient, but also on thatrecipient’s own status as a positive or negative individual. In con-trast, 5-mo-olds evaluated actors solely on the basis of the localvalence of their actions

To address this question, subjects were placed into one of twoconditions immediately after observing the above interactions.Subjects in the Prosocial Target condition saw new interactions inwhich the formerly prosocial puppet was now playing with a bal

For adults then, social evaluation often goes beyond analyzing theimmediate, local valence of a behavior enacted toward another.We consider our previous evaluations of a target to determine theoverall, what we might call the global valence of the action*

Neu-rological reward systems are activated during punishment, sug-gesting it is individually reinforcing (20

Adults, however, are not limited even to these expanded heu-ristics. Under at least some circumstances, people are motivated toapproach individuals who have been intentionally harmful in thepast—to punish them

Infants in their first year of life will approachindividuals who have acted positively toward others and avoidthose who have acted negatively (4–6). Infants also expect othersto respond in this manner—to approach those who have helpedthem and avoid those who have harmed them (3, 4, 7)

We find that although 5-mo-old infantsuniformly prefer individuals who act positively toward othersregardless of the status of the target, 8-mo-old infants selectivelyprefer characters who act positively toward prosocial individuals andcharacters who act negatively toward antisocial individuals

Although adults generally prefer helpful behaviors and those whoperform them, there are situations (in particular, when the target ofan action is disliked) in which overt antisocial acts are seen asappropriate, and those whoperform them are viewedpositively.

In sum, when a distributor divided items between an ingroupand an outgroup recipient, 1.5- and 2.5-y-olds expected fairnesswhen there were sufficient items for all parties involved, but theyexpected ingroup support to trump fairness otherwise. Fromearly in life, charity is understood to begin with the ingroup.

These results suggest that by the second year of life, childrenare already capable of sophisticated moral computations thatweave together several different factors

They expected fairness to prevail when therewere as many toys as puppets (three-item condition), but theyexpected ingroup support to prevail when there were fewer toysthan puppets (two-item condition)

First, to increase the perceptual variety of ourstimuli, we used elephants and monkeys as puppets, and we usedtoys (e.g., alphabet blocks) as allocation items. Second, in thethree-item condition, the distributor left the third item on thetray, instead of taking it with her when she exited the apparatus.

Infants were highly attentive during the initial phase of the testtrial (M =98%). Looking times during the final phase (Fig. 2)were analyzed using an ANOVA with event (equal, nonmixed-equal, or nonmixed-unequal) as a between-subject factor. The maineffect of event was significant, F(2, 21) =5.68, P =0.011, ηp2 =0.35.As predicted, infants looked significantly longer if shown the equalevent (M =27.23, SD =13.57) or the nonmixed-unequal event(M =29.86, SD =13.65), as opposed to the nonmixed-equal event(M =12.13, SD =4.07), F(1, 21) =11.00, P =0.003, ηp2 =0.34;looking times at the equal and nonmixed-unequal events did notdiffer significantly, F(1, 21) =0.21, P =0.651, d =−0.19. AKruskal–Wallis test confirmed the main effect of event in infants’looking times, χ2 (2) =7.82, P =0.020. Thus, when a distributorfrom one group divided two cookies between two recipients from adifferent group, infants expected the distributor to divide thecookies equally between the two recipients, as in prior research(17–23). When the distributor faced an ingroup and an outgrouprecipient, however, infants expected the distributor to reserve bothcookies for her group, as in Exp. 1

Infants varied whether they prioritized fairness or ingroup sup-port depending on how many resources were available. When therewere as many cookies as puppets (three-item condition), infantsexpected fairness to prevail: They detected a violation when thedistributor gave two cookies to one recipient and none to the other,regardless of whether the ingroup or the outgroup puppet wasadvantaged. However, when there were just enough cookies for thedistributor’s group (two-item condition), infants expected ingroupsupport to prevail: They detected a violation when the distributorgave any of the cookies to the outgroup puppet.

Infants were highly attentive during the initial phase of the testtrial; across conditions and events, they looked, on average, for98% of the initial phase. The final phase of the test trial endedwhen infants either looked away for two consecutive seconds afterhaving looked for at least five cumulative seconds or looked for amaximum of 45 cumulative seconds; the 5-s minimum value gaveinfants the opportunity to continue processing the event after itended.

We reasoned that if infants (i) assigned the monkey and giraffepuppets to distinct groups, based on the various cues provided,(ii) brought to bear considerations of fairness and ingroup supportwhen forming expectations about the distributor’s actions, and(iii) rank-ordered these considerations in a context-sensitive man-ner that took into account how many allocation items were avail-able, then different responses might be observed in the 2- andthree-item conditions. In the research with 3- to 5-y-olds describedearlier (34, 36), children acted as distributors and compared thenumber of items available to the number of ingroup recipientswhen deciding whether to produce a fairness-based or an ingroup-based allocation

ogether, these findings suggest thatbeginning early in life, infants can form groups based on a variety ofcriteria, and they apply an abstract expectation of ingroup support tothese groups: Individuals are expected to prefer and align withingroup members and to give them help and comfort when needed.

Across tasks, infants demonstrated sensitivity to fairness in sce-narios involving a wide range of characters, including humans,animal puppets, and shapes with eyes.

Children in the two-item condition looked significantly longer at the equal or favors-outgroup event than at the favors-ingroup event, suggesting thatwhen there were only enough items for the group to which thedistributor belonged, children detected a violation if she gave anyof the items to the outgroup puppet. In the three-item condition, incontrast, children looked significantly longer at the favors-ingroupor favors-outgroup event than at the equal event, suggesting thatwhen there were enough items for all puppets present, childrendetected a violation if the distributor chose to give two items toone recipient and none to the other, regardless of which recipientwas advantaged. Thus, infants and toddlers expected fairness toprevail when there were as many items as puppets, but theyexpected ingroup support to trump fairness otherwise

The notion that observed behavioral differences in infants’evaluations of prosocial and antisocial agents are markers of orprecursors to more complex moral cognition remains highlycontentious (13, 15). This study provides compelling evidencethat infants and toddlers discriminate between the prosocial andantisocial actions of others.

These results illustrate continuity in early social evaluation. Thoseinfants who prefer prosocial characters in one behavioral task showthe greatest neural differentiation to prosocial versus antisocialactions in an implicit moral evaluation task. Additionally, infants’and toddlers’individual differences in Nc (300–500 ms) for viewingpositive versus bad actions or in LPP/PSW (600–1,000 ms) were notrelated to the expression of sharing behaviors. Although sharingbehaviors in infancy and toddlerhood appear to be common (47),many have questioned whether such early expressions are actuallyrepresentative of morality or rather early self-regulation, com-pliance, and effortful control (48, 49). Consistent with the latterargument, our results suggest that sharing propensity is directlyrelated to children’s temperamental effortful control.

Consistent with our hypotheses, relative asymmetry for leftversus right was greater for the perception of hindering thanhelping, implying the engagement of domain-general mechanismsof withdrawal/avoidance when confronted with aversive stimuli andthe reduction of withdrawal when viewing positive stimuli.

The current study failed to find such a preference inolder toddlers, suggesting early and strong individual, rather thangroup, differences in preferring a helper that are directly related tothe neural dynamics processing of perceiving prosocial and antiso-cial characters. This result provides a more nuanced view of de-veloping sociomoral evaluations

Parental sensitivity to injustice for others wasdirectly predictive of infants and toddlers’ERP differences in theperception of helping versus harmful scenes in the CMST between300 and 500 ms (Cz, r=0.357, P<0.05; Fig. 2). Individual dif-ferences in the 600–1,000 ms range were not significantly relatedto parental individual dispositions toward justice

Parental disposition in perspective taking from the Interper-sonal Reactivity Index (IRI) was positively related to the proba-bility of children sharing (r=0.284, P<0.05). Parental personaldistress (IRI) was inversely related to children’s probability ofsharing (r=–0.317, P<0.05).

The other task was an infant-friendly version of the Chicago moral sensitivity task (CMST), de-veloped by Cowell and Decety (22), in which two characters areinteracting in a variety of prosocial (e.g., sharing, helping) and an-tisocial (e.g., shoving, tripping, hitting) ways. Consistent with de-velopmental research, it was hypothesized that even the youngestinfants would differentiate between characters that helped anotherand characters that hindered another, as indexed by differences infrontal power density asymmetries during the SET, and several keyERP components in the CMST.

ome preliminary evidence suggests that infant’spost hoc evaluations of previously helping and hindering charactersare marked by midrange attentional event-related potential (ERP)differences (P400) (23). However, although these findings have be-gun to illuminate the complexity of the neural underpinnings of earlysocial sociomoral evaluations and prosocial behaviors, it is unclearwhether these two fields (social evaluations and behaviors) are re-lated in their emergence and whether there is online neural differ-entiation to the processing of harmful and helpful actions of others

suggesting that toddlers do not simply reactto others’emotional displays but actively interpret the intentionsand feeling states of others. Although there is obviously consid-erable development in prosocial and moral abilities between in-fancy and childhood, it is less clear what motivates this change

and be-havioral observations, children under 2 y of age appear to both actprosocially and prefer prosocial to antisocial others (4). For ex-ample, 3-mo-olds preferentially attend to a character who pre-viously acted in a prosocial manner toward another (5), suggestinga bias toward those that “do good things.”By 6 mo of age, thisvisual preference is expanded to the realm of behaviors

All children demonstrated a neural differentia-tion in both spectral EEG power density modulations and time-lockedERPs when perceiving prosocial or antisocial agents. Time-lockedneural differences predicted children’s preference for prosocial char-acters and were influenced by parental values regarding justice andfairnes

Overall, this investigation casts light on the fundamentalnature of moral cognition, including its underpinnings in generalprocesses such as attention and approach–withdrawal, providingplausible mechanisms of early change and a foundation for forwardmovement in the field of developmental social neuroscience

Although some argue that infants and tod-dlers possess a “moral sense”based on core knowledge of the socialworld, others suggest that social evaluations are hierarchical in na-ture and the product of an integration of rudimentary general pro-cesses such as attention allocation and approach and avoidance

Thus in both Hungaryand Poland, strong national assemblies acted to prevent the bureaucrat-ization that occurred in other late statebuilders such as the Germanstates in order to protect a kind of patrimonialism centered not aroundproprietary officials and tax farmers, as in Latin Europe, but aroundcontinued control by a locally based elite, in this case the landowningnobility. The most tangible consequence of this triumph of patrimonialconstitutionalism was a decline of military effectiveness in both countries,which led to a partial loss of independence in Hungary and the com-plete destruction of the state at the hands of its neighbors in Poland.

By August 1948, Karensin the army had mobilised in anti-government violence, seizing cities close toRangoon and eventually forcing U Nu to cede control over the importantTwante Canal linking the Rangoon port to the Irrawaddy delta. By September,swaths of Karen soldiers were defecting from the tatmadaw to join irregularforces and the KNU gained momentum in demanding an independent Karen-Mon state in the heartland of Lower Burma. By the new year of 1949, theKaren insurgency had escalated into large-scale violence as full-fledged battlesraged, culminating in the KNDO’s capture of Mandalay in March 1949(Lintner 2000: 13). In response, the tatmadaw embarked on a scorched-earthmilitary campaign to quell the separatist uprisings, rather than the kind of admin-istrative advances that characterised the Malayan government’s evolving reactionto leftist insurgency

This only makes sense, however, if one fails todistinguish between administrative standardisation and militarised pacification,and conflates efforts to decimate and expel minority populations with attemptsto administer and assimilate them. Normatively speaking, Scott is spot-on inhis denunciation of the Burmese military’s (or tatmadaw’s) brutality towardnon-state peoples. But analytically speaking, the tatmadaw’s approach to region-alist insurgencies has had precious little if anything to do with legibility, adminis-tration, standardisation, assimilation, or a ‘grid’of any conceivable sort.

In sum, the conduct of the Malayan Emergency eventually came to approxi-mate Scott’s vision of a state obsessed with the sedentarisation and standardis-ation of its populace. Yet this more administrative and interventionist approachto governance emerged very gradually –and very grudgingly –as the purelycoercive and expatriating approach to counterinsurgency proved stubbornlyinadequate to the political challeng

Before the outbreak of theEmergency, however, this illegibility was of no major concern to either Britishor Malay officialdom, as “weakly administered land regulations”allowed “thesquatters to remain largely untouched by authority in the years immediately fol-lowing the war”

British authorities showed no inclinationto confront the rebellions with anything but raw, militarised force. It was onlyafter abject failure at purely militarised pacification that the colonial Leviathanadopted a more administrative approach to countering the leftist threat, withthe support of a wide range of Malayan elites

If states are ‘hard-wired’to pursue the standardisationand homogenisation of their populations and territories, it is in insurgent zoneswhere we should be most likely to see such state campaigns taking place. Andsometimes, to be sure, we do. Yet the unevenness of such attempts and the ten-dency for states to conduct counterinsurgency through practices such as indirectrule, militarised pacification, divide and rule, and forcible expulsion, rather thanadministrative governance, calls into question any claim that states are alwayskeen to impose a bureaucratic grid over their subjects.

Our concept of standoffishness embraces a similar corrective to the con-ventional image of a Weberian state in its monopoly-maximising guise. It alsooffers a conceptual vocabulary to discuss myriad and messy state behavioursthat, like Staniland’s wartime political orders, are “not the domain of a fewobscure outliers, but instead are hiding in plain sight”(Staniland 2012: 248).

We suspect that this claim would sound very odd to the poorest and most vul-nerable citizens of national capitals throughout the world. Literally withinwalking distance of gleaming national government offices in capitals rangingfrom Jakarta to Cairo to Washington, DC, one finds densely populated commu-nities whose primary experiences with state authority are those of coercion andneglect.

The fundamental challenge of state power is commonly portrayed as one in whichthe political centre struggles to project its writ to the farthest reaches of nationalterritory. Those populations located close to the national capital would seeminglyhave little hope of avoiding state campaigns to render them ‘legible’through cul-tural homogenisation and administrative standardisation.

we argue that it is, instead, to minimise political challenges

thus arguethat the world is riddled with standoffish states, not just standardisingstates. Even in the zones where the potential costs of eschewing the pursuit oflegibility appear highest –those containing violent insurgencies –states canprove surprisingly disinterested in pursuing centralised governance in ahighly administrative manner. We highlight four alternative strategies –indirectrule, divide and conquer, militarised pacification, and forcible expulsion –thatstates commonly deploy to fulfil what we see as their most fundamental objective:preventing political challenges to the ruling centre

uch of this book can be read as a case against the imperialism of high-modernist, planned social order. I stress the word "imperialism" here because I am emphatically not making a blanket case against ei- ther bureaucratic planning or high-modernist ideology.

huge dams

In summary, we are convinced that a renewed focus onlegibility has the potential to yield novel theoretical and em-pirical insights into a range of outcomes of interest to politicalscientists. We believe that the present article has only begunto explore these exciting new research frontiers.

n other words, taking legibilityinto account can help to explain not only a quantitative in-crease in the state’s extractive capacity but also a qualitativeshift in the arsenal of extractive instruments at the state’s

aken together, the results of our subnational and cross-national exercises are consistent with the claim that legibilityis central to the state’s capacity to curb free-riding in col-lective endeavors. We caution our readers that our results aresuggestive and should not be interpreted as causal.2

Greaterlegibility is associated with lower infant mortality rates andhigher literacy and primary school enrollment rates.

We can return to our examples of Switzerland, France,and Sierra Leone and compute the Whipple score and Myersscore for each distribution. Recall that the minimum valuesfor Whipple and Myers are 100 and 0, respectively. Switzer-land has a Whipple score of 102.86 and a Myers score of 0.22,indicating almost no error in the reported data. France hasscores similar to Switzerland, even though its distribution isdiscontinuous due to war-related demographic shocks. Theselow scores corroborate what we saw graphically in figure 1. Incontrast, we calculate a Whipple score of 228.20 and a Myersscore of 26.48 for Sierra Leone in 2004. The much largervalues on the indices indicate a high degree of age heaping,which we can confirm visually in figure 1C

n summary, the lack of age awareness is likely to obtainwhere the state generally has very little contact with its cit-izens. In such environments, citizens have few incentives tolearn their true ages with any precision, and few opportu-nities to obtain documents that can help with age calculationand recall. These considerations imply that when age infor-mation is inaccurate or missing, other types of informationabout the population useful for state administration are likelyto be missing as well, making census accuracy a proxy forlegibility more broadly.

elow weintroduce a method to quantify the magnitude of theseerrors. Before doing so, however, we first turn to a discussionof the data-generating process, as this touches upon the in-terpretation of our census measure.

In 1628, the Swedish state undertook what somehave called the first cadastral mapping in the Western world,which consisted of the production of around 12,000 mapsdelineating the boundaries of villages, freeholds, and farms,as well as details on land ownership, tenure, cultivation, andvaluation (Hallenberg 2012; Kain and Baigent 1992). In addi-tion, the Ecclesiastical Laws of 1686, which tasked the clergywith keeping lists of their parishioners’tax duties, furtherreinforced the state’s capacity to gather fiscal information.The legacy of these efforts can be seen in the fact that, in theyear 1920, already 80% of the economically active populationin Sweden was registered with the tax authority

To reduce confusion, a second designation could be added,indicating occupation (e.g., Smith, Baker), place of residence(e.g., Hill, Edgewood), father’s given name (e.g., Johnson,O’Higgins), or a personal characteristic (e.g., Short, Strong).Although such naming conventions usually sufficed for localidentification, they presented an administrative nightmareto central state agents who, naturally, were not privy to thislocal knowledge.

We show that, controlling foreconomic, geographic, and temporal factors, areas with higherlegibility also have more effective tax collection and enjoygreater public goods provision. Together the evidence sup-ports our contention that legibility is a crucial, yet largelyunderstudied, component of state capacity and administra-tion.

Legibility implies both (a) that the state possesses in-formation about local practices and (b) that this informationis rendered in standardized forms (e.g., cadastral maps, birthcertificates, property registers)

We contribute to this stream of scholarship by highlight-ing the importance of the state’s information gathering andprocessing functions

uch areas represented a reliable zone of refuge for those who lived there or who chose to go there

Colonial regimes, though they worked mightily to construct all-weather roads and bridges, were thwarted in much the same way as the indigenous states they replaced. In the long, arduous campaign to occupy upper Burma, the progress made by colonial troops (mostly from India) in the dry season was often undone by the rains and, it seems, by the diseases of the wet season as well. An account of the effort in 1885 to clear Minbu, in upper Burma, of rebels and bandits, revealed that the rains forced a withdrawal of British troops: “And by the end of August the whole of the western part of the district was in the hands of the rebels and nothing remained to us but a narrow strip along the river-bank. The rains and the deadly season which succeeds them in the water-logged country at the foot of the Yoma [Pegu-Yoma mountain range] . . . prevented extended operations from being undertaken before the end of the year [again the dry season].”

s a rule of thumb, hilly areas above three hundred meters in elevation were not a part of “Burma” proper. We must therefore consider precolonial Burma as a flatland phenomenon, rarely venturing out of its irrigation-adapted ecological niche.

n 1714, the Dutch estimated that Johor could bring into battle 6,500 men and 233 vessels of all types. In Vietnam, by contrast, the Nguyen army was tallied at 22,740 men, including 6,400 marines and 3,280 infantry.”24 The earliest cautionary tale of maritime-state vulnerability is, of course, Thucydides’ Peloponnesian War, in which a resolutely maritime Athens is, finally, undone by its more agrarian rivals, Sparta and Syracuse.

The Malay trading port is the classical example, typically lying athwart a river junction or estuary, allowing its ruler to monopolize trade in upstream (hulu) export products and simi-larly to control the hinterland’s access to trade goods from downstream (hilir) coastal and international commerce. The Straits of Malacca were, in the same fashion, a choke point for long-distance trade between the Indian Ocean and China and thus a uniquely privileged space for state-making.

A map in which the unit of measurement is not distance but the time of travel is, in fact, far more in accord with vernacular practices than the more abstract, standardized concept of kilometers or miles. If you ask a Southeast Asian peasant how far it is to the next village, say, the answer will probably be in units of time, not of linear distance.

As Charles Tilly has noted, “Before the later nineteenth cen-tury, land transport was so expensive everywhere in Europe that no country could afford to supply a large army or big city with grain and other heavy goods without having efficient water transport. Rulers fed major inland cities such as Berlin and Madrid only at great effort and great cost to their hinter-lands. The exceptional efficiency of waterways in the Netherlands undoubt-edly gave the Dutch great advantages at peace and war.”18 The daunting military obstacles presented by travel over very rugged terrain, even in the mid-twentieth century, was never more evident than in the conquest of Tibet by the China’s People’s Liberation Army in 1951. Tibetan delegates and party representatives who signed the agreement in Beijing traveled back to Lhasa via “the quicker route”: namely by sea to Calcutta, then by train and horseback through Sikkim.

Extrapolating from a more generous estimate of 32 kilometers a day by foot, F. K. Lehman estimates that the precolonial state’s maximum size could not have been much more than 160 kilometers in diameter, although Mataram in Java was considerably broader. Assuming a court roughly in the center of a circular kingdom with a diame-ter of, say, 240 kilometers, the distance to the kingdom’s edge would be 120 kilometers.16 Much beyond this point, even in flat terrain, state power would fade, giving way to the sway of another kingdom or to local strongmen and/or bandit gangs. (See map 3 for an illustration of the effect of terrain on effective distances.)

The calculus is slightly modified in premodern Southeast Asia, and particularly in warfare, by the use of elephants, which could carry baggage and negotiate difficult terrain, but their numbers were modest and no military campaign depended essentially on them.1

hus Burmese royal edicts of 1598 and 1643, respectively, ordered that each soldier remain in his habitual place of residence, near the court center, and required all palace guards not on duty to cultivate their fields.8 The constant injunctions against moving or leaving the fields fallow are, if we read such edicts “against the grain,” evidence that achieving these goals met with a good deal of resistance.

Most important of all, if the army and/or the tax collector arrive on the scene when the crop is ripe, they can confiscate as much of the crop as they wish.3 Grain, then, as compared with root crops, is both legible to the state and relatively appropriable.

The non-grain-producing elites, artisans, and specialists at the state’s core must, then, be fed by cultivators who are relatively near. The concentration of manpower in the Southeast Asian context is, in turn, particularly imperative, and particularly difficult, given the historical low population-to-land ratio that favors demo-graphic dispersal. Thus the kingdom’s core and its ruler must be defended and maintained, as well as fed, by a labor supply that is assembled relatively close at hand.

Your task, crudely put, is to devise an ideal “state space”: that is to say, an ideal space of appropriation. Insofar as the state depends on taxes or rents in the largest possible sense of the term (foodstuffs, corvée labor, soldiers, tribute, tradable goods, specie), the question becomes: what arrangements are most likely to guarantee the ruler a substantial and reliable surplus of manpower and grain at least cost?

Moving to the southern hemisphere, luz Marina Arias (2013) argues that new external military threats gave elites in eighteenth-century Mexico the incentive to overcome local tax free-riding. She claims that, to increase state capacity, the Spanish monarchy made a fiscal deal with corporate elites (e.g., guild leaders, clergyman) who derived rents from the traditional economic system. Thus, in contrast to seventeenth-century England, a parliamentary bargain did not accompany fiscal development in Mexico. Miguel centeno (1997) argues that the logic of “war makes states” (Tilly, 1992) does not apply to nineteenth-century South America. To have war-related state development, pre-conditions—namely, the need to rely on internal taxation, a minimum amount of administrative infrastructure, and support from local elites—must be satisfied.

araman and Pamuk (2010) show that, after two centuries of stagnation, Ottoman revenues grew more than 15 times from 1780 to 1913 as the state made central-izing reforms.

On the contrary, the historical evidence indicates that state effective-ness—or lack thereof—had a major influence on long-run economic outcomes. O’Brien (2011) claims that England’s early establishment of a state capable of providing domestic property rights protection and external security was an important reason why it became the first indus-trial nation.

A third group of scholars claims that the focus on institutional reforms overlooks other important features of state formation. Avner Greif and Murat Iyigun (2013) emphasize the role of local social safety nets, which can reduce violence and promote risk-taking. They show that English counties that provided more poor relief saw greater political stability and technological innovations from the late 1600s to the early 1800s.

Stephen Quinn (2001) argues that greater government borrowing in the aftermath of the Glorious Revolution discouraged private investment.

The establishment of effective states—a process not completed until the nine-teenth century—was associated with a dramatic increase in the state’s capacity to tax relative to any reforms undertaken during the Old Regime

oyal moral hazard helps explain why monarchs were nearly always in battle; wars involving major powers were underway in 78 to 95 percent of all years between 1500 and 1800 (Tilly 1992)

National representative bodies—culled from certain provinces and social groups, and called to order at the monarch’s request—had control over tax policy, but monarchs had control over spending. Parliamentary elites did not trust monarchs to spend public funds in line with their preferences. They called for institutional reforms that would grant them budgetary oversigh

A key feature of the Old Regime fiscal landscape was tax free-riding (Dincecco 2009, 2011). There was a close relationship between local control over taxation and political autonomy. local elites had strong incentives to oppose tax reforms at the national level that would under-mine their traditional fiscal and political rights. In this context, monarchs had to bargain place-by-place over individual tax rates.

I argue that the creation of effective states was the institutional foundation upon which twentieth-century welfare states were built. I conclude by drawing historical lessons for political and economic development today. I base this account on my recent research (e.g., Dincecco 2011).

rst, the national (i.e., sovereign) government must have the ability to implement a uniform tax system throughout its territory. Second, within the national government itself, there must be a veto player (i.e., parliament) with the ability to regularly monitor the state’s budget. States that satisfy these two political conditions are “effective”:

Becauseboth turf war and cartel–state conflict have occurred in the absence of the other, Ihave argued that a valid first step is to analyze each in isolation. Nonetheless,future research must grapple with the complicated interactions between these twoforms of conflict

Conversely, many examples of violence against state actors, suchas high-level assassinations of police or judicial officers, may be understood as sig-nals whose intended recipients are also part of the state. In these cases, signalingforms part of a larger strategy of violent corruption or violent lobbying, ultimatelyaimed at affecting policy at either the de jure or de facto level

In practice, violent lobbying often seems to fail: the Caldero ́ n government, forexample, unsurprisingly rejected the Familia Michoacana’s calls for dialogue or itsdemands to remove the federal police. This suggests overconfidence on the part ofcartels as a potentially important trigger

Strategies of influence over policy outcomes.

Crossing these dimensions yields four distinct strategies (Figure 3). Hidingstrategies correspond to ‘‘conventional’’ corruption and lobbying.24 In violentcorruption, cartels use violence to affect enforcement, usually inducing enforcersto accept (smaller) bribes.25 In violent lobbying, cartels use violence to pressure lead-ers for de jure policy change, usually reduced repression along a specific dimension.

Cartel–state conflict isfar rarer—most traffickers in most places avoid confrontation with the state—butin Colombia, Mexico, and Brazil, it has been of long duration once initiated. The typology suggests an explanatory hypothesis: cartel–state conflict only beginswhen traffickers see weathering an ongoing armed confrontation as optimal, so pos-itive cases will tend to be drawn out and involve only large and well-equipped drug-trafficking organization

For example, in the Opium Wars, Britain both annexed portions of Chineseterritory (conquest) and forced key policy concessions (constraint). Foreign-ledcoups may blur the line between outright conquest and the desire to constrain theactions of an ally or neighbor. Conflicts can also change over time: what beginsas a revolutionary insurgency, with both sides truly fighting for control of the state,can deteriorate into a criminal war in which jungle-bound rebels subsist on drugprofits for decades with no ambition to capture the center or secede

First, in these settings, decisive victory is probably either unrea-listic or undesirable, and hence not a meaningful solution to commitment prob-lems. For cartels, even outright military victory over the state would not lock intheir preferred policy or distribution of rents. In the current international system,an overt narco-state granting itself a monopoly on cross-border trafficking is basi-cally unthinkable, whereas nascent insurgencies can reasonably hope to not onlyprevail militarily (Fearon and Laitin 2007, 16) but to win international legitimacyand support.

Likewise, states cannot meaningfully ‘‘conquer,’’ even provisionally, cartels’illicit rents from trafficking, nor can fighting realistically lock in states’ preferredoutcome. One might object that states nonetheless aim to destroy, not constrain, car-tels. Again, though, we must distinguish states’ immediate, tactical goals (such asdismantling a particular cartel) from their larger strategic aims in fighting drug wars.Identifying those aims is complicated by rhetoric: international drug prohibitiontreaties tout total eradication as a supposedly plausible goal, and drug warriorsfrequently equate dismantling cartels with ‘‘victory’’ (e.g., Bonner 2010). Yet theempirical record is nearly devoid of examples of sustained eradication of an other-wise viable drug trade

n short, any reasonably non-myopic state must understand its drug-war aims to be, at best, constraint—not total elimination—of illicit drug markets.

Conversely, cartels cannot easily ‘‘take what they want’’ from the state, because‘‘what they want’’ is not something the state has, but rather something they wouldlike the state to refrain from doing. Even where cartels wield significant territorialcontrol, such as Rio’s favelas, they do not seek formal secession or the removalof all vestiges of state presence; their interest remains firmly fixed on constrainingstate repression of their illicit economic activities

Generalizing, then, the proximate aim of fighting in ‘‘classic’’ insurgency is con-quest, in which belligerents aim to definitively expropriate from or replace oppo-nents. When fighting states, this implies competitive state building, if only withina portion of the national territory. Cartel–state conflict, by contrast, is a war of con-straint: belligerents aim ‘‘merely’’ to coerce opponents into changing their behavior(i.e., their policies, when fighting states). The distinction is meant to characterize,not define: just as insurgents could fight wars of constraint but rarely do, drug cartelscould seek to seize formal state power, but have not. Distinguishing conflicts basedon this aspect of how they are fought can, I claim, help us theorize about why

repression or elimination of one cartel generally helps itscompetitors. This is unlikely to be true of kidnapping rings, or, for that matter,most insurgencies.

I show that theterror tactics associated with violent lobbying were indeed far more prominentin Colombia than Mexico or Brazil. Finally, I argue that turf-war dynamics likeinternal and competitive signaling are unlikely to generate antistate violence ontheir own, but could be complementary to violent corruption.

political leaders in order to induce changes in de jure policy; the emblematic case isColombian drug lord Pablo Escobar’s all-out war on the state between 1984 and1991. In violent corruption, cartels target enforcement agents in order to intimidatethem and reduce the price of bribes; this is the logic expressed in Escobar’s phraseplato o plomo (the bullet or the bribe), but the logic applies equally to cases like Riowhere the initiative for bribery, and the majority of the bargaining power, lies withcorrupt state forces.

Conversely, belligerents in wars of constrainteither cannot or do not aim to definitively conquer or replace their opponent; bruteforce may play a defensive, tactical role, but it is subjugated to an overarching aimthat is fundamentally coercive

When cartels turn to fighting strategies, I argue, their aim is not to conquer thestate but to constrain it—to change its behavior, which in the case of states meanspolicy outcomes. In wars of constraint, the function of violence is generally coer-cive, in Schelling’s (1966, 240) sense of the power to hurt: ‘‘To inflict sufferinggains nothing and saves nothing directly; it can only make people behave to avoidit.’’8

Both insurgency and turf war obey a logic of conquest that is well theorized: inweakly institutionalized settings, actors will use violence to physically appropriatemutually prized territory or resources when they cannot reach stable bargained solu-tions (Fearon 1995; Powell 2002; Wagner 1994; Walter 2009). Critically for thisbody of theory, conquest can be decisive, producing definitive outcomes that are dif-ficult or costly to reverse

Byeliding critical differences in rebels’ and cartels’ battle aims, it obscures ratherthan illuminates a fundamental puzzle: why fight the state if not to topple orsecede from it? The answer ‘‘to keep the state off one’s back’’ is insufficient:criminal groups everywhere, including powerful mafia and ‘‘institutionalized’’street gangs (Hagedorn 2005), would like the state off their back, but rarely resortto violence.5 This is the puzzle.

cartels do not seek to topple the government andseize formal power. However, with no clear conceptual alternative, the tendencyhas been to treat Mexico’s conflict as a crim

A useful conceptualization of criminal war should be based on observablecharacteristics that illuminate differences in underlying logics of violence. Failureto appreciate such fundamental differences can make states’ efforts to curb cartelviolence, often counterinsurgency inspired, disastrously counterproductiv

The very contentiousness of classifying drug-related conflicts as civil war con-stitutes a barrier to understanding. Politically, ‘‘civil war’’ carries a possibly unde-served stigma of chaos and loss of state control3 (Grillo 2011, 202-6) and couldhave consequences for international intervention (Bergal 2011). Analytically, itwould seem to corroborate a persistent but controversial current in conflict studiesthat attributes ‘‘greedy’’ motives to rebels (Collier and Hoeffler 2004) or evenequates insurgency with organized crime (Collier 2000).

In the study of war, ‘‘criminal’’ may be the new ‘‘civil.’’ Since 2006, Mexico’sdrug war has claimed 60–70,000 lives (Shirk and Wallman 2015), as many as16,000 a year (Shirk et al. 2013)

. Whereas rebels fight states, and cartels fight with one another, toconquer mutually prized territory and resources, cartels fight states ‘‘merely’’ toconstrain their behavior and influence policy outcomes.

What kind of war is Mexico’s drug war?

First, governments themselves can alter their ideological projects, whether as a result of major regime changes or shifts over time within regimes. Second, armed groups can adjust their ideological position relative to the government, either radi calizing or moderating. T

nation or institutionalization (Christia 2012). The operational needs required to embrace a strange bedfellow militia are severe: insurgents marching on the capital and no-holds-barred militarized elections, for instance, can trigger collusion with ideologically incompatible groups to stave off existential threats. These are part ne

Mortal enemies are targeted with large-scale commitments of state intelligence and coercio

Undesirables are generally contained. They are objects of repression and moni toring, but not the full hammer blows of state power. Governments have scarce coer cive capacity, and they are unlikely to expend it all on groups that are seen as a mild political challenge, even if these groups are large, wealthy, or well armed. Instead, we see recurrent raids, vigilance, and a general attempt to keep a lid on the activities of undesirable

Ideological projects are most stable and coherent when governments are unified around a common idea of what the state exists for and how the political arena should be structured. Divided state apparatuses (with civil-military, central-local, or intra military tensions) may be associated with competing ideological projects. Factiona lized security forces or political elites that control those security forces can have heterogeneous notions of appropriate roles for armed groups, and state policy will therefore vary according to which arm of the state is acting. In these circumstances, the "government" will need to be disaggregate

A group can occupy an ally, enemy, or gray zone ideological position in the eyes of rulers. Ally militias are those that mobilize symbols, cleavages, and demands that can be easily accommodated within the political arena desired by a regime. For instance, paramilitaries that battle leftist insurgents are likely to be compatible with a regime dominated by oligarchic, anticommunist landlords. Enemy militias are those that mobilize goals in direct conflict with a ruling regime's ideological tem plate. These need not be antistate insurgents: a workers' militia demanding redistri bution may not be actively fighting the state, but nevertheless be viewed as an illegitimate, unacceptable threat to the government's favored political order. Finally, militias can occupy a "gray zone"5 that mixes elements of political conflict and compromise. This is where many militias exist, with their own political agendas that straddle

d defend.4 Governments attempt to create political arenas that reflect these preferences, whether a linguistically homogenized polity or a communist party state, though these projects are often contested and may disastrously fail. The laws, boundaries of political discourse, norms about acceptable behavior, and insti tutions that regimes try to put into plac

Incorporation aims to demobilize a militia by formally integrating it into "normal" politics. It is a form of state making that seeks to eliminate non-state violence through absorption rather than annihilation. This is a strategy of peace negotiation in the case of insurgents or of formal demobilization or transition in the case of pro-state militias. Incorporation can lead to full transitions out of violent activity on the part of the group or can involve the armed group continuing to carry guns but now as formally part of the state. T

Collusion can be a short-term expedient against a mutual enemy that later breaks down into suppression, a trust-building way-station en route to incorporation, or a long-term outcome in which the state and a militia develop clear rules of interaction and

Collusion is a strategy of active, sustained cooperation between a state and an orga nized armed actor, ranging from explicitly holding back police and military acti