a theory of state formation and the origins of inequality

A THEORY OF STATE FORMATION AND THE ORIGINS OF INEQUALITY (*)Carles Boix (**)February 21, 2011(*) Previous versions of this paper were presented at the University of Virginia, the New Yorkmeeting of the “September Group” Princeton University, Duke University, Yale University, StanfordUniversity, the University of Chicago and Harvard University. I thank the comments of all theirparticipants, in particular to those of Sam Bowles, Dan Gingerich, Stathis Kalyvas, Carol Mershon,Thomas Romer and Milan Svolik. I am especially grateful to the comments of Alícia Adserà and toTeppei Yamamoto for his research assistance.(**) Professor of Politics and Public Affairs, Princeton University. E-mail: cboix@princeton.edu.

AbstractTo explain the transition from stateless, relatively equal communities to agrarian, unequal and stategovernedsocieties that has taken place since about 8,500 BC, this paper offers a model with thefollowing traits: income-maximizing agents, who may either choose between a productive strategyand an expropriation strategy, coordinate on peace without permanent political institutions providedtheir economic conditions are relatively equal; however, as soon as inequality rises (due to a biasedtechnological shock), their spontaneous coordination around peace becomes unfeasible; agents sortout into different types; and states are formed, either of a monarchical type (where the moreproductive agents make a transfer to the less productive ones in exchange for permanent protection)or a republican system (where the former invest directly on some defensive structures to deter thelatter from looting them). Those two solutions, crucially determined by military technology and theopportunity costs of not producing, lead to distinct distributional outcomes. The formal model isfollowed by a worldwide cross-sectional analysis of the sources of the ‘exogenous’ nature of theagricultural revolution, the latter’s effects on to the formation of the state, and the impact of militarytechnologies on the emergence and distribution of different political regime. The paper concludeswith a brief exploration of empirical evidence (mostly related to the distribution of height patterns)relating the type of political structure to the internal distribution of economic resources.1

Until about 10,000 years ago, mankind lived in small, relatively equal and statelesscommunities of foragers. From 8,500 B.C. onward, however, several regions of the world transitedto agriculture and sedentary human communities. In turn, the domestication of plants and animalsspurred or at least was correlated with two momentous events: the formation of the state (rangingfrom small urban polities to vast despotic monarchies) and the emergence of marked inequalities inthe distribution of wealth. Those changes had long-lasting political and economic consequences, atleast until the spread of industrial technologies compressed, with a few national exceptions andtemporal reversions, the distribution of incomes within countries, and arguably triggered a process ofdemocratization in the last 150 years or so.This article offers a theory to account for those joint processes of state formation and theorigins of inequality. The first two sections of the article develop a model in which incomemaximizingagents may pursue two kinds of strategies: either a productive strategy, consisting in theallocation of their resources to the production of goods and services, or an expropriatory orextractive strategy, which implies directing their resources to appropriate the assets or returns ofother individuals. Even though those agents may engage in conflict in one-shot interactions, overtime they may cooperate ‘spontaneously’, that is, without any permanent political institutions, if andonly if their economic conditions are relatively equal. As soon as the level of inequality rises above acertain threshold (generally following a biased technological shock in which some individualsbenefit differently from a similar innovation), the cooperation equilibrium breaks down. In such anunequal world, agents sort out into different types – with different payoff functions and differentstrategies. Whereas the economically advantaged individuals or ‘natural producers’ have a strongpreference for production, those who do not benefit from the technological shock prefer looting.Hence, to deter the latter, the ‘natural producers’ either invest on constructing and managing somedefensive structures (to stop the looters) or make some transfers to the ‘natural looters’ to appeasethem (and to obtain some permanent protection from them). Although both solutions give birth topermanent political institutions, i.e. the state, each strategy leads to different political regimes (self-2

governing or republican systems versus monarchical regimes). Each political outcome, which isstrongly affected by the nature and evolution of military technologies, has, in turn, distinctiveredistributive effects. Monarchical regimes tend to generate and sustain much higher levels ofinequality than republican institutions. Still, the overall distribution of income continues to dependon strictly economic (technological) variables too.The third part of this article explores the empirical validity of the theoretical model.Employing both anthropological and archaeological evidence, it first shows that foraging groupsboth have no permanent political authority (i.e. some kind of leadership stably holding the monopolyof coercion) and display very limited levels of social and economic stratification and similarinterpersonal patterns of consumption and individual wealth. Using systematic evidence across theworld and over time, it demonstrates that the agricultural revolution that took place about 10,000years ago was triggered by exogenous factors, that is, the biogeographical conditions of each regionin the world. It then relates the timing of the birth of state institutions to the emergence ofagriculture. It concludes by linking the prevailing type of state institutions (in agricultural societies)to the nature of military technologies.The fourth part of the article explores the determinants of inequality. After calibrating themodel (showing how economic and military technologies as well as population and the capacity ofstates shape the final distribution of income), it brings in some empirical evidence (mostly related tothe distribution of health patterns) relating the type of political structure to the internal distribution ofeconomic resources. Contrary to the standard research in political economy emphasizing either theeconomic foundations (Marx 1848) or the political roots of inequality (Rousseau 1755[1973],Andreski 1968), both exercises reveal that the patterns of wealth and income inequality are jointlydetermined by both technological and institutional parameters.This article makes several contributions to two areas of research in political economy: theprocess of state formation and the origins of inequality. Broadly speaking, the literature on stateformation includes two main strands of thought. Institutionalist (including neomarxist) models3

interpret the construction of political institutions as the outcome of a (voluntary or forced) deal inwhich bandits offer their protection to peasants in exchange for some payment or transfer (North1981; Olson 1993, 2000; Levi 1988; Bates 2001). Older, functionalist work sees states forming as aresponse to collective action problems and to the breakdown of a cooperation equilibrium. Bothapproaches suffer from several glaring weaknesses. Institutionalist models do not entertain thepossibility of states forming as self-governing polities. Functionalist approaches cannot account forthe distributive effects that accompany the emergence of political institutions. With the exception ofsome historical macrosociological accounts (McNeill 1982, Tilly 1990), both literatures bracket theeffects of economic and military technologies on the overall process of state formation and on theinstitutions upon which bandits and producers coordinate.Although there is a burgeoning literature on the effects of inequality on growth (Atkinsonand Bourguignon 2000, Lindert 2000, Piketty 2000, Kahhat 2007) and on political institutions andconflict (Boix 2003, Acemoglu and Robinson 2006, Geddes 2007, Boix 2008), much less work hasbeen done on the underlying causes of the evolution of inequality. From an empirical point of viewthey have focused their attention to the evolution of income distribution in advanced countries in thelast three decades (Gottschalk and Smeeding 1997), suggesting partial yet still heavily contestedexplanations (Wallerstein 1999, O’Rourke and Williamson 1999, Scheve and Stasavage 2007,Gordon and Dew-Becker 2008). From a theoretical point of view, most contributions are strictaccounting exercises or limit themselves to economic variables such as imperfect credit marketswhich are in turn left unexplained. 1 Moreover, there is little work on the choice of politicalinstitutions and the distribution of economic assets as jointly determined phenomena – even thoughthat question was a long-standing concern among classical political theorists such as Aristotle([1988], IV, 11), Machiavelli (1531 [1970], I, 55) or Rousseau (1755 [1973]).1 An exception is Rogowski and McRae (2008). For a review of the literature, se Kahhat (2007) and Boix(2010).4

1. INITIAL CONDITIONSSet-upTo understand the dynamics that led from equal, primitive economies to complex, unequalhistorical societies, consider a world populated by two types of agents, 1 and 2, each one with sometechnological knowledge (for example, some natural abilities to hunt, fight, etc.). These two agentscan be thought of as individuals living together or as individuals (or households) populating someterritory. They can also be thought of as representative individuals of homogeneous groups, that is,groups formed by agents who, because of their close ‘resemblance’ (in a sense to be made moreprecise shortly) to each other, can cooperate in a stable, permanent manner.All individuals have the same time L, which they can employ in two different types ofactivities. On the one hand, they may allocate some fraction of their time L to some ‘productive’activity (such as hunting) that generates some income y i in a standard Cobb-Douglas productionfunction y i =A i T α 1-αi L i for i={1,2} and where A denotes the state of technology and T stands for a fixedstock of land. 2 On the other hand, they may direct their time to predatory activities, that is, to stealother individuals’ production. In this latter instance, they obtain (1-1/θ L )y j , where y j is the return fromthe productive activities of individual j (where j≠i) and θ L is a technology parameter (θ L >1) thatmeasures the effectiveness of the time allocated to extraction. Hence, the income of individual iwould be, in principle, that is, in a decision-theory setting:r i = (1-λ) A i T α 1-αi L i + λ(1-1/θ L ) A j T α 1-αj L jwhere λ (with 0 ≤λ ≤1) is the fraction of time i spends looting.Assume that T and L are identical across all agents and define  n = A n T α 1-αn L n / A j T α j L 1-α j sothat (A i T α 1-αi L i / A j T α j L 1-α j )=  i and (A j T α 1-αj L j / A j T α j L 1-α j )=1. Then we can rewrite r i as:r i = (1-λ)  i + λ(1-1/θ L ) (1)2 Later I explore more precisely the nature of that productive or cooperative strategy, that is, the set ofactivities (and interpersonal activities) that go into it.5

Taking  i and θ L as given, agents i and j will choose λ that maximize their respectiveincomes in the context of the strategic setting in which both agents choose between looting andproducing. The payoffs of that game will be as follows. If both agents simply produce (λ = 0), theirpayoff will be  i and 1 respectively. Under a condition of generalized looting (that is, with both i andj looting the other side), all output not directly consumed by each agent will be destroyed. Hence,agent i will be only able to obtain (in fact, keep) the portion she produces directly or (1-λ)  i . In turn,j will only keep (1-λ). Finally, if one agent loots while the other does not, the income of the nonlooter(the part not expropriated by the looter) will be reduced by some proportion due to thedestructive effects of pillaging. That is, if agent i loots while agent j does not loot, i receives thepayoffs in (1) while j gets (1-β)(1-λ). The parameter β (with 0≤β≤1) measures the destructiveness ofpredation. In turn, if j loots and i only produces, i keeps (1-β)(1-λ) i In that instance, j gets (1-λ) +λ(1-1/θ L )  i .Anarchy, Equality of Conditions, and Self-sustaining PeaceTo determine the conditions under which both agents will coordinate on a productionstrategy, suppose that they play an infinite, repeated game, with some discount rate δ (identical forall players). To make the analysis simple (and focused on the problem at hand), assume that playershave two strategies available to them. They can follow a trigger strategy PP, producing always untillooting occurs and then responding with looting ever after. Or they can adopt a constant lootingstrategy LL. 33 Since (1-β)(1-λ)

To have sustained peace, the following conditions should take place. For player i, the valueof choosing a production strategy must be larger than looting a producer in the first period and thenfacing a looting solution always ever after:Aˆi> (1 − λ)Â1 − δiLiRearranging terms:ˆA i1> (1 − δ )(1 − )(2)θLSimilarly, for player j the following inequality (where producing always is better than looting theproducer and then moving into a permanent situation of conflict) must hold:1 δ+ λ(1- ) + (1 − λ)Âθ 1 − δ11> (1 − λ)+ λ(1- )Â1 − δθLiδ+ (1 − λ)1 − δHence:1Aˆi1> (1 − δ )(1 − )(3)θLequilibrium if:Putting (2) and (3) together, we can conclude that continuous production will be an1> Aˆi(1 − δ )(1 −1θ )L1> (1 − δ )(1 − )θL(4)Inequality (4) carries with it two main implications. In the first place, it shows that, even in aHobbesian world, that is, in a world where there is no authority in place and players may haveincentives to raid others, it is possible for individuals to sustain peace, avoid conflict and engage inproductive activities – again without having to resort to any centralized common mechanisms ofauthority. This is a result well known in the formal study of human cooperation (Axelrod 1984).In the second place, it also shows that the non-predatory equilibrium is only possible if thereis some fundamental equality of conditions among individuals or across human communities. If  ibecomes too large, that is, if agent i is much more productive (and, hence, better off) than agent j,7

then the left-hand side in (4) will not hold. Agent j has no incentive to cooperate and shifts to alooting strategy. Similarly, if  i becomes much lower than 1, then the right-hand side in (4) will nothold either. In that instance, it is agent i that shifts away from the production strategy. In short, in thecontext of a primitive economy as the one described so far, peace is only possible if the value ofnatural resources is uniformly distributed across the territory populated by i and j. 4 Otherwise, theproduction equilibrium collapses and conflict erupts. 5Open Conflict and the Exit ConditionThe bounds of the level of inequality within which cooperation is feasible vary with thediscount rate and the looting technology. First, the more players value the future (i.e. the closer thediscount rate δ is to 0), the higher inequality can be without jeopardizing a production equilibrium.Second, a predatory outcome is less likely when looting is inefficient (that is, when θ L gets closest to1). The effectiveness of looting is directly related to the military technology in the hands of thepredator. But it is also shaped by the strategies producers themselves have to reduce the value ofwhat a looter can steal. The efficiency of the looting technology declines (and the occurrence of openpredation does too) if producers can easily exit or move away in response to any potentialaggression, that is, if the costs of moving to a different territory are low. In fact, we know from theanthropological literature that abandoning one’s own community and creating or joining another is acommonly used mechanism in pre-state communities to lessen internal conflict (Leacock and Lee1982).4 In the model presented so far, the value of resources depends on the state of technology A. However, itcould also (and in fact it does) vary independently of the tools and skills of humans as a result of climate,quality of land, etc.5 Notice that the production equilibrium is compatible with higher levels of inequality if the biased shockthat generates more inequality (that is, that increases the income differential across individuals) increasesthe income of the poorer individual (in the production equilibrium) as well. [DEVELOP.]8

The feasibility of the exit option is negatively related to population densities and positivelycorrelated with the marginal value of land. The gradual territorial expansion of mankind across theglobe in the late Pleistocene can be explained as a rational response to manage competition forresources. Migration was the automatic, less costly response to avoid a permanent state of open war.However, as land filled up, the exit option became more costly and the relative value of lootingincreases. According to the model, social conditions had to approximate strict equality to avoid apredatory equilibrium.The Nature of “Cooperation”So far I have defined the production equilibrium by exclusion: as a set of social interactionswith no systematic looting. However, a state of nature with no open war may take several forms. Inthe most stripped-down interpretation of the model, it may consist of a set of isolated individuals,wandering alone in the forests, satisfying their needs on the spot, and clashing (or not) with eachother as a function of the parameters (their physical strength, roughly similar at the beginning oftimes, and the environmental distribution of resources, more or less concentrated) of the model. In amore elaborate application of the theory, it implies the existence of some communities of men andwomen living together under a condition of relative equality. The first type of state of nature, akin toRousseau’s vision of the first men as a savage men, “solitary, idle and always near danger”, does notsquare well with the evidence produced by anthropological research and, more recently, byevolutionary biology theory. Human beings appear to have lived in families and bands always (eventhough these groups tend to exhibit some considerable level of fluidity in primitive societies). 6 Thosesocial units may be partly glued by biological as well as by emotional ties (Ingold 2000). But theyare the result, especially at the band level, of optimizing individuals that join in groups to exploit6 On human sociality as a universal condition from an anthropological point of view, see Levi-Strauss(1961 [1955]), pp. 389-392, 397-398. For a review from the point of view of evolutionary biology, seeHanson (1992), Hawkes (1992) and Boone (1992).9

some economies of scale in production and/or to minimize the risks of foraging and harvestingthrough resource-sharing (Smith 1987, Boone 1992). In any case, that is, regardless of themotivations that lead to the emergence of those cooperative structures, absent any force or authority,they are only sustainable if their members are roughly equal.Notice then that the model can be thought as a building block in what may be labeled a‘nested’ theory of cooperation. At the lowest (individual) level, peace is possible among a subset ofactors provided they enjoy sufficiently similar payoffs: when they do, they can form a stable, orderlytribe, village or neighborhood. Similarly, this social unit can also sustain cooperative relations withother human bands or tribes if the same degree of relative equality (and looting technologies) appliesat that level. Peace breaks down at the point (in the continuum of agents) where inequality ofconditions becomes ‘excessive’, i.e., inequality (4) does not hold anymore. In short, the modelexplains why cooperation may exist at some but not other levels of social interactions.2. CONFLICT AND THE EMERGENCE OF STATE INSTITUTIONSInequality and Sorting of AgentsAs pointed above, inequality makes a cooperation equilibrium unfeasible because it altersthe ranking of preferences (and the corresponding optimal strategy) of each agent. The emergence ofinequality is the result of a process through which resources (or, more precisely, the value ofresources) become concentrated in dense and predictable patches and therefore the returns toindividuals vary depending on the territory they occupy. 7 This process can occur through at leastthree paths. First, the discovery of new resources (or, simply, the occupation of new areas that have amuch higher yield): this is the case of the gradual habitation by foraging communities of theNorthwest Pacific Coast, where fishing resources were concentrated in particular areas. Second, atechnological shock, such as the agricultural revolution that I describe and exploit empirically in the7 For a first analysis of the effects of resource concentration among animal populations, see Wilson(1964). For its extension to humans, see Boone (1992).10

following section, that changes the marginal productivity for some but not all territories. Finally,population growth leading to the occupation of marginal territories and to a correspondingdifferentiation in incomes between the initial core and the less valuable land.Besides breaking the production equilibrium, inequality also sorts agents out into differenteconomic classes with different ‘political’ goals. Following a biased shock that increases the returnsto production of agent i relative to those of agent j, each individual pursues now very differentstrategies. Whereas agent j now prefers looting, the economically advantaged agent i still has an(even stronger) incentive to sustain a production equilibrium.From Anarchy to Political InstitutionsTo overcome a generalized situation of looting, those who are better off under a productionequilibrium can pursue two alternative strategies. On the one hand, they can make a transfer (in theform of a tribute or lump-sum payment) to agents of type j such that the latter stop looting (andbenefit from the production efforts of i). In this instance, the potential looters or bandits (those whohave not benefited from the economic shock) control (govern) and protect the natural producers fromtheir own explicit violence and against other potential bandits. On the other hand, the ‘natural’producers (or agents i) may decide to make looting a much less rewarding activity (for agents of typej) than just following a cooperative, non-violent path. They do so by setting up some institutions orstructures, such as a common army and some stable leadership, that would allow them to defendtheir assets and lives against any potential bandits.Both solutions involve the formation of some organization or structure that controls themonopoly of violence over a given territory – that is, they involve the construction of a “state”. Butthe mechanisms underpinning each solution are different. The first road to peace, which we may11

efer to as a “monarchical solution”, follows Olson’s insight on the emergence of the state as “rovingbandits” become “stationary bandits” or landlords (Olson 1993, 2000). 8The second strategy, which has received no attention in the modern institutionalist literatureon state formation, leads instead to the construction of a state in which the “natural” producers alsodouble as defenders and rulers. We may wish to call the second strategy an instance of a “selfgoverning”polity or of a republican settlement. 9 Notice that a system in which the community ofproducers sets aside some portion of its income to hire an army or a leader to protect its members(while governing themselves through republican institutions) does not constitute a case of selfgovernmentor republicanism. Since the hired leader cannot credibly commit to preserve the terms ofthe contract once he has been appointed and takes control of the army, such a solution ends in (or isequivalent to) the subjection of the producers to a leader-monarch. In short, we should think of astable republican solution as one in which citizens have ultimate control over the means of defense. 10We do not need to think of bandits as ‘external’ enemies only. The term of looter refers aswell to individuals who live in a given community and may try to steal, free ride on others, etc.8 Writing from an anthropological perspective, Carneiro (1970) also offered a theory of state formationthat emphasizes the interaction of violence and exit options (the latter in turn deriving from populationdensity). Exit strategies and density are, however, not endogenized in his paper. As in Olson, he focuseson the emergence of autocratic states. See Wright (1977) for a review of state formation theories inanthropology and archaeology.9 Olson’s (and in fact most neoinstitutionalist accounts’) unique solution explains why his explanationabout democratic transitions is ad hoc and unconvincing. He claims that democracy emerges when thereis a vacuum of power in an already-formed state in which a set of notables have roughly equal claims topower (see Olson 2000: 28 ff.). But this state of things (i.e. the existence of balance of power) could beequally predicated for stateless societies.10 For a discussion of the moral hazard problems of hiring an external captain or condottiero in lateMedieval Italian cities, see McNeill (1982). The mechanisms through which citizens (or some subset ofcitizens) can control defense – and so limit exploitation – are multiple. For a formal analysis of howinstitutions may limit executive power, even in the context of non-democratic systems, see, for example,Boix and Svolik (2010).12

Individual(s) i will have then an incentive to set up some policing mechanism to dissuadeindividual(s) j from misbehaving.Monarchical Settlement versus Self-Governing SolutionLet me now characterize each political solution separately and then solve for the conditionsunder which one should prevail over the other. (All this discussion clears the way to examineanalytically (in the third part, after I explore the historical validity of the model) the ways in whichthose alternative paths of state formation have different consequences for income distribution andinequality – with the monarchical solution implying some redistribution from producers to the lordand the eventual rearrangement of income rankings within society.)Monarchical Settlement. In a ‘monarchical’ settlement, where agents of type i make atransfer to agents j to buy their protection, agent i’s payoff will be (1-ε) / (1-δ), where ε denotes thetransfer extracted by agent j, with 0≤ε≤1, and all time is now devoted to production. In turn, agent jreceives, provided he follows a non-looting strategy (which now includes an effort to protectindividuals of type i against other potential bandits), a transfer from the producer in proportion to thetime λ spent governing (as well as any payoff from his own production, if any): [(1-λ)+(ε-γ)λ i ] / (1-δ), where γ is the portion that the bandit-lord spends in governing and protecting the producer and0

Self-Governing Institutions. In a self-governing system the producers themselves takesteps to defend their assets and lives against bandits, that is, they have to spend some fraction of hertime to set up a military or defense structure (such as building a wall or a watchtower to observe thehorizon). That amount of time will depend on the efficiency with which it is employed to deter anylooter j from pillaging the producer. The level of efficiency will be a function of the militarytechnology in the hands of the producers (relative to the military technology of the looters).Therefore, the payoff of those agents of type i (or a community with individuals i such that all havean incentive to cooperate) that decide to defend themselves is [(1-(λ/θ D ))Â i ]/(1-δ)], where λ is thetime spent in defense and θ D is the efficiency of the technology with which producer i defendshimself against potential bandits – and where 0≤λ≤1 and θ D >λ. Agents i keep the fraction ofproduction not spent in defense. As i’s defensive capabilities increase, the producer can spend moretime generating Y=YA α i L 1-α i .Looked at from the point of view of the potential looter, the purpose of creating a defensestructure is equivalent to the principle of inflicting some cost on j to push the latter into peacefulbehavior. In other words, agent i must choose a parameter of defense θ D such that j has no incentiveto exploit the producer. Hence, the j’s payoff under a successful self-governing or republican regimewill be 1, that is, j will devote all his time to production and none to looting.Choice of Institutions. Once the level of inequality disrupts the possibility od spontaneouscoordination around peace, the natural producers choose between establishing a self-governing state,giving up their rights to a tyrant (a ‘stationary bandit’) or engaging in a looting strategy. In turn,agents j must decide whether to abide by the political settlement or to loot.Under what conditions will each solution prevail?1. In the first place, a state will only emerge if the payoffs under stable institutions (either atyrant or a republican structure) exceed the payoffs that come from looting the other side in the firstperiod and then slipping into a situation of permanent conflict. Let me examine first (in points 1A14

and 1B) by examining each alternative separately, that is, the value of a republican solution versusstraightforward looting and the value of monarchy versus looting. Later (in point 2), I consider theconditions under which a republican solution prevails over a monarchical solution (or viceversa)provided order is preferred to conflict.1. A. The ‘natural’ producers (agents i) would rather have a republican structure than looting(provided they prefer the republican solution over the monarchical settlement) if:(1 − λ θ ) AˆD1 − δi> (1 − λ)Aˆi1 δ+ λ(1− ) + (1 − λ)Aˆiθ 1 − δLRearranging the terms of this inequality gives us:ˆA i(1 −1θ L )> (1 − δ )(5)(1 −1θ )DThus, the incentives of any agent i to move to a republican solution increase when thealternative option of looting becomes less attractive – either because the looting technology θ Ldeclines in efficiency, the defensive technology θ D improves, or the production technology  iincreases.In turn, agent j will accept the self-governing structure set up by i if just producing leaveshim better off than trying the looting strategy:1> (1 − λ)+1 − δλθDÂiδ+ (1 − λ)1 − δRearranging this inequality then:θ D> Âi1 − δ(6)In short, j is more likely to accept the production strategy when the strategy capabilities of iincreases and when the value of i’s production declines.Putting (5) and (6) together, a republican settlement is feasible whenever:θ D ˆ (1 −1θ L )> Ai> (1 − δ )− δ(1 −1θ )(7)1 D15

In addition to (7), self-government will only be feasible if peace without transfers is not anequilibrium. 121. B. In the instance in which a monarchical solution is preferred over a republican one, thenatural producers will only accept to transfer some resources to a tyrant if that makes them better offthan looting:(1 − ε ) Aˆ1 − δi> (1 − λ)Aˆi1 δ+ λ(1− ) + (1 − λ)Aˆiθ 1 − δLRearranging the inequality:(1 − δ )(1 −1/θ L ) λλ −> εÂi(8)A monarchical regime becomes more attractive to producers when the efficiency of looters(measures through θ L ) and the extraction rate ε decline and the productivity of i increases.In turn, j will have an incentive to act as a monarch and not as a bandit if:(1 − λ)+ ( ε − γ ) λAˆ1 − δi1 δ> (1 − λ)+ (1 − ) λÂi + (1 − λ)θ 1 − δLHence:1ε > (1 − δ )(1 − ) + γθ L (9)Therefore, agents of type j will become monarchs when the extraction rate and the lootingcosts increase and when the costs of governing decline.Putting together (8) and (9), a monarchical settlement will be possible whenever:1 11λ − λ(1−δ)(1 − ) > ε > (1 −δ)(1 − ) + γθ ˆL A θL(10)iAs in the case of a republican settlement, in addition to (10), a monarchical settlement willonly be feasible if peace without transfers is not an equilibrium. 1312 This condition takes place, following (3), when  i >1/[(1-δ)(1-1/θ L )]. It is possible to show that there isa broad range of values for which both the latter inequality and (7) hold at the same time.16

2. For the space in which a state (either monarchical or republican) may take place(according to the conditions spelled out in 1.A and 1.B), the producer (agent i) will prefer arepublican over a monarchical solution whenever:1 − ( λ θ ˆD)) Ai1 − δ(1 − ε ) Aˆ>1 − δ(iSimplifying the expression:λε >θ D(11)According to inequality (11), the political solution depends, given a fixed time λ devoted tolooting, on the rate of extraction ε. The higher ε is, the higher the incentive to establish a republic.The political regime will be also a function of war technology. If producers are superior or evenrelatively equal to looters at war, the former will be able to deter any predator or set of predators andtherefore will associate in human communities where decisions are made in a consensual manner orwhere, at least, a broad section of the population has some impact over politics. However, as soon asθ D falls, that is, as soon as looters acquire some strong comparative advantage in the use of force,monarchical regimes will become the prevalent form of government. That may happen for multiplereasons. Looters may control a particular type of weapon or develop a new set of military tactics thatare particularly effective at waging war. More often, the invention of new weapons (such as swordsor horse chariots) requiring considerable training and expertise foster the process of specializationand professionalization of war-making. The opportunity costs of self-defense among producersescalate. By contrast, those individuals that engage in predatory activities gain in power andeffectiveness. They form a separate caste of warriors that encroaches upon the self-governinginstitutions preferred by the producers and that establishes centralized states with highly extractivefiscal systems.13 Again, there is a wide range of values that make these two conditions (and thus a monarchical regime)feasible.17

Notice that, if the producer sets up a republic, the potential looter j will abide by the solutionsince, by definition, a republic is only possible if producers have the actual military capacity to deterlooters from pillaging them (and, again, this deterrence strategy still is preferable to any othersolution).If the producers are, instead, better off under a monarchy, agents of type j will not accept themonarchical settlement if their payoff under a republic is larger than the monarchical transfer (1-ε)A– with the looting alternative precluded by our assumption that inequality (7) holds. In that case, theproducers will have to increase ε to the point that monarchy becomes attractive to j (provided that,even after that increase in the extraction rate, a republic would still leave agents i worse off than themonarchical solution).An Extension: The Possibility of Imperial Republic. In addition to establishing a selfgoverningstructure to pursue a purely defensive strategy, producers i may decide to engage in anoffensive one, both geared toward defending their land and attacking the looters. The offensivestrategy costs an extra fraction of i’s production above the price paid to simply defend their ownterritory. Therefore, since it implies a lower direct output for i, we can denote the offensivetechnology parameter as θ O with 0≤θ O ≤θ D . 14Since the use of an offensive strategy implies that agents i will apply sufficient force to makeindividuals j spend their time producing even when they experience some looting from i, the naturalproducers will follow that strategy (to establish an imperial republic) only if the benefits of looting jare larger that the portion of output lost in upgrading the military strategy from a defensive into an14 It seems plausible to think that even when the main endeavor of these agents is production, they stillneed to pay always some minimum costs of defense θ D to be ready to move to the fully militarizedstrategy. As a matter of fact, if θ D were equal to 0, then government institutions would only appear whenindividuals of type j chose a looting strategy. Otherwise, if individuals of type j did not loot, everyonewould simply cooperate spontaneously with no permanent structure governing them. This does not seemto be realistic.18

offensive one. This will depend on two factors. First, the incentives to set up an imperial republicwill decline whenever the offensive costs θ O increase with respect to the defensive costs θ D . Second,they will also fall the more productive the agents of type i become with respect to the agents of typej. The latter proposition derives from the fact that, as the i’s become wealthier, the opportunity costsof any expansion become higher. Or, to put it in slightly different terms, the more marginal lands andpersons (of type j) are, the less prone their (richer) neighbors i will be to attack and subject them. 153. HISTORICAL DISCUSSIONEgalitarian, Pre-state CommunitiesUntil about 8,500 BC all mankind lived in bands of hunters and gatherers. Sincearchaeological research only offers rather scant information about the nature and patterns of socialinteractions in those pre-state communities, we are forced to rely on contemporary anthropologicalresearch, mostly on foraging communities, to assess the nature of simple communities. 16With the exception of the fishing communities of the Northwest Pacific Coast, whichshowed considerable social differentiation, mostly related to a pattern of strong spatial concentrationof resources in discrete areas, foraging communities are small in size (the sum, at most, of a fewextended families). 17 They are also internally equal. Economic differences within bands are either15 Mathematically, i will choose to follow an offensive strategy if:( 1 − λ θ ˆ + λ θ − λ θ ˆO ) Ai( / O ) (1 / D ) Ai>1 − δ1 − δSimplifying the expression shows that the inequality holds if:λ λ λ> ( − )Âiθ θ θThis will happen as either  i becomes closer to ) or θ O approaches θ D (which in turns makes theexpression in the parenthesis close to 0).16 For a discussion of the extent to which current ethnographic evidence may be employed to makeinferences about past foraging communities, see Headland and Reid (1989) and O’Connell, Hwakes andJones (1988).17 Northwest pacific communities were also characterized by the construction and ownership ofsophisticated equipment, such as complex fish traps, which involved considerable labor (Rowley-Conwy2001).OOD19

absent or very limited. Food and, particularly, meat distribution is strictly egalitarian (except, insome communities, for very large preys) (Hawkes 2000). Differences in nutrition over the life cycleare minimal (at least within each gender) (Kelly 1995). Success in hunting is tracked by everyone(Hawkes 2000) and some social recognition is granted to the good hunter (Turnbull 1965; Kaplan &Hill 1985). But foraging communities have strict social norms that prevent the hunter from owningthe meat he has hunted and status recognition is extremely transient (Weissner 1996, Erdal andWhitten 1994). Individual success seems to lead at most to easier access to a few goods such assexual partners. However, there is too much variability in hunting performance over the medium runto lead to permanent social distinctions (Hawkes 2000). According to recent work comparing 21foraging, pastoralist and agricultural communities, inequality is extremely low among foragersexcept for embodied wealth (measured in terms of physical strength, hunting skills, etc.) (Smith et al.2010). The intergenerational transmission of wealth is also low except for traits that can be explainedin strict genetic terms. To the extent that there is some correlation in genetic endowment acrossgenerations, it does not come from more and better nutrition provided to the children of the bestendowed parents but rather from matching between good hunters and more hardworking women(Hawkes 2000). The existing archaeological evidence for Paleolithic sites seems to confirm similarpatterns of nutritional equality and little material differentiation (Formicola 2007; Vanhaeren &D’Errico 2005).Simple foraging societies lack any clear, stable authority structure and certainly no stableorganization (or group of individuals) holding the monopoly of violence in a given territory.Decisions affecting the band are arrived at through discussion in which all adults participate(Turney-High [1971]: 61-73; Silberbauer 1982). Even when there are some patterns of leadership,leaders have no formal authority and act as mere referees among different individuals or families orspend substantial time negotiating with, cajoling or persuading other members of the tribe (Clastres1972; Lee and Leacock 1982; Lee 1982). And although threats and actual violence are certainlyemployed within the tribe, they happen in a sporadic or at least non-institutionalized manner20

(Clastres 1972, 1974; Kelly 2000). In short, the structure and underlying dynamics of foragingsocieties match the model in section 1. A pattern of relative equality in foraging groups underpinsthe outcome of “spontaneous” social cooperation within each tribe or, to employ Clastres’ terms, theoutcome of “non-state politics”.Equality is the result of a particular (uniform) distribution of resources in the environment(in conjunction with a specific technology of production). But the outcome of social cooperation isthe result of systematic, deliberate and painstaking efforts directed at maintaining peace amongindividuals. At the end of the day, the production equilibrium is just one of the two equilibria thatmay take place in the repeated PD game discussed in part 1. Violence, looting and exploitation canerupt at any time, with catastrophic consequences for everyone, especially given the constraintsimposed by a rather harsh ecological environment. Hence pre-state communities need to spendconsiderable efforts at equalizing material outcomes and social status across individuals.The “creation” (or maintenance) of equality in pre-state communities does not result fromhaving a shared ideology accepted and followed out of idealistic convictions. It has none of theromantic qualities imagined by Marx and Engels in their depiction of primitive communism or laterattributed to foragers by a certain anthropological tradition. It is simply the result of the “behavior ofsociable and self-interested individuals whose motivations include a strong desire to get enough forthemselves coupled with a strong desire to make sure that no one else gets more than they do” (Erdal& Whiten 1994: 177). In fact, the distribution of food is often conflictual, involving arguments,yelling and clear signs of jealousy and envy. Any individual deviation from the mean is quicklynoticed and then hammered out through mocking, pleading and shaming. Writing about the San,Wiessner notes that “[it cannot] be argued that !Kung have little interest in material possessions.Most possessions … are highly desired; and many !Kung are really torn between the desire toaccumulate goods and the desire to remain within a secure system of mutual help.” Hence,individuals may work hard and even try to accumulate some material possessions only to “comeunder more and more pressure to give them away in hxaro, and finally give in and redistribute them”21

(1982: 81-82). Similarly, hunting parties are followed by an elaborate set of practices directed atleveling down the successful hunter. The latter rejoins the band announcing the yield of the huntmodestly, almost in a hush. His entry is then followed by the systematic efforts of the band to“downplay his merits” through teasing and mocking, to stress the randomness of the success, and to“emphasize his duty to provide meat for others” (Wiessner 1996: 179). 18 Across all simple societies,from the Inuit Eskimos to the !Kung in the Kalahari, the ideology and practice of equality turnsaround denouncing and avoiding two capital sins: stinginess, that is, “to hoard one’s goods jealouslyand secretively, guarding them ‘like a hyena’” (Lee 1982: 52); and arrogance, which “is actuallydangerous, since according to the !Kung ‘[someone’s] pride will make him kill someone” (Lee 1982:54)Maintaining equality implies exercising an inordinate level of psychological repression uponeach and every member of the community. As stressed before, the suppression of individualdifferences is often done very openly: by the use of constant deprecation toward oneself, byemploying a piercing irony toward others, and by constantly begging the more successful membersto engage in the proper behavior of sharing (or what someone has labeled “institutionalizedstealing”). But in some communities repression is above all about structuring the inner motivationsof every member of the group: although relationships between parents and children are hardlyauthoritarian and although the latter normally enjoy a very playful infancy, children are brought upin the expectation that they will uphold the tight social norms of equality and conformity in place.Accordingly, as Briggs writes about the Inuits, “in a society in which the expression of hostility isvery stringently controlled (…), it is more than likely that people project their own suppressedhostility onto others and, knowing that they themselves feel violent, will easily suspect others ofmurderous intents” (1982: 116).18 Similarly, Lee (1984) for the !Kung, Woodburn (1982: 440) on the Hadza, Hararo (1981: 536) on theMbuti, Clastres (1972) on the Ache, Lothrop (1928: 422) on the Ona, Balicki (19xx: chapter 9) onEskimos.22

Those societies, characterized by repressed fear and constant suspicion, witness unexpectedexplosions of individual anger, resulting from rather trivial, almost childish disputes with a marked“noninstrumental manner” (Knaupf 1991: 400). According to the existing anthropological research,violence is unrelated to any struggle for material goods or political control. In turn, male competitionfor females only explains part of the conflict and with different success across groups. In addition toindividual disputes and killings, there is considerable collective violence: individuals deemed todeviate from the boundaries of appropriate behavior are prone to accusations of black magic andwitchcraft and to some kind of socially sanctioned execution (Knauft 1987, 1991). Violent deathsare, as a result, extremely high in pre-state societies. The annual homicide rate, which stands ataround 20 to 30 individuals per 100,000 in American inner cities, has been estimated to approacharound 300 individuals per 100,000 among Agta Pygmies and over 400 per 100,000 among theCopper Eskimo and Papua-New Guinea’s Gebusi (Knauft 1987; Headland 1989; Gat 1999). About25 percent of adult males die from a violent death on average in foraging societies. That proportionmatches the percentage of buried skeletal remains from the Paleolithic showing traces of violence(Kelly 2000; 155).Nonetheless, that level of lethal conflict does not jeopardize social stability. Homicidescommitted by a single individual remain largely constrained. They never escalate into a generalconflict pitting one family against the other in the social group. 19 As Balikci (1970: 182) points outin his study on the Netsilik Eskimos, because “every murder signified the loss of a highly neededseal hunter”, the community always intervened to stop any revenge that could lead to a chain19 Homicides (committed to revenge a previous crime) are hardly interpreted as an injury to a wholegroup and do not generally lead to the collapse of social interaction within a given human band. Inforaging communities, “when felt anger or ill-feeling is expressed in interpersonal violence, theexpressive action typically runs its course (…) and may actually culminate in murder. However, theviolence that occurs is specific, not generalized, and it does not escalate beyond a sequence of events thatencompasses homicide followed by execution of the killer. Typically, a murder is an isolated event withno sequel” (Kelly 2000: 42-43).23

eaction. If generalized violence becomes unavoidable, most foraging bands manage it through aprocess of fission: a section of the group secedes and migrates to a new location, where it selforganizesagain along the standard patterns of relative equality and non-institutionalized politics. 20Transition to AgricultureThe domestication of animals and plants about 11,000 years ago put an end to a world solelypopulated with foraging communities. The invention of agriculture had two effects. It increased landproductivity: whereas a family of hunters and gatherers may need several square miles to survive, ahousehold of irrigation agriculturalists may live with less than one hectare (Bellwood 2005: 14-19).It raised land productivity in a non-uniform manner across territories, generating a pattern ofinequality across individuals and human communities (and probably within human groups also): asagriculture spread out, initially homogeneous regions (enjoying the same marginal productivity ashunting areas) acquired a differentiated resource gradient ranging from a ‘core’ of rich lands to a‘periphery’ of very marginal quality.The causes behind the adoption of agriculture have been thoroughly debated amongscholars. A first strand of the literature explains the domestication of plants and animals as aresponse to the declining marginal productivity of hunting and the related Malthusian constraintsfaced by growing populations of hunters and gatherers (Flannery 1965; Binford 1968; Cohen 1977).A second set of researchers interprets agriculture as a technological improvement that happened as aresult of a long process of learning (Rindos 1980). A third line of research traces it back to amomentous change in cultural norms among foraging societies (Bender 1978). 21 Regardless of theultimate causes of the agricultural revolution, it is obvious that the transition to agriculture could20 For an analysis of the exit option in the hortoculturalist society of the Yanomanos, see Chagnon (1997).For a broader conceptual analysis of the effect of strong exit options on income distribution and politicalstructures, see Hirschman (1981), chapter 10.21 For general reviews, see Watson (1995), Bar-Yosef and Meadow (1995), Bellwood (2005), Barker(2006).24

have only taken place in those areas that were characterized by quite precise geographical,environmental and climatic conditions. As pointed out by Diamond (1997), the agriculturalrevolution materialized in those regions of the world that had enough plants and animals suitable fordomestication and the proper climate to exploit those species.Partly building upon seminal empirical work by Hibbs and Olsson (2004) and Olsson andHibbs (2005), Table 1 shows that most of the variation in the time at which different parts of theworld transited to agriculture was indeed related to the set of biological and geographical conditionsof each of those areas. That result implies that the causes of the transition that led to the social andpolitical consequences (inequality and state formation) formalized in Section 1 were exogenous tothose outcomes.The estimations in Table 1 are of two kinds. Models 1 through 3 examine the impact of strictgeographical conditions on the number of domesticable plants and animals across the world. Models4 and 5 estimate the effect of those same geographical conditions on the years it took eachgeographical unit to transit to agriculture. The units of analysis are the geographical quadrantsdefined by 10 degrees in latitude and longitude (starting at the intersection of the Equator with theGreenwhich meridian) for which there is some land mass. There are 234 such observations – 203 ifwe exclude the Antarctica. 22 The number of species suitable for domestication are taken from Olssonand Hibbs (2005) and come for ten separate regions of the world. 23 The number of domesticableanimals ranges from 9 in the Middle East to 0 in Iceland. The number of plants goes from 33 to 0.The years to agriculture are the years it took agriculture to appear since the beginning of theHolocene (9,500 BC). 24 The data is built employing the information gathered in the surveys written22 In the Olsson-Hibbs work, the unit of observation is the contemporary state.23 Those regions are: Near East-Europe-North Africa, East Asia, Southeast Asia, Sub-Saharan Africa,North America, Central America, South America, Australia, Pacific Islands and Iceland.24 For a discussion on whether to code the origins of agriculture following either the first evidence ofplant (and animal) domestication or the moment at which horticultural practices give way to intensiveagriculture, see Shenk et al. (2010) and Thurston and Fischer (2007).25

y Bellwood (2005) and Parker (2006). For those areas where agriculture has yet to develop I assigna value of 11,500 years (9,500 plus 2,000 years).The geographical conditions are measured through four variables:1. Climate, measured following Köppen’s system of climatic classification and ranging from1 (worst climate for agriculture) to 4 (Mediterranean and West Coast climate). I code each unit ofobservation by looking at the geometrical center of the quadrant. 252. Latitude of the geometrical center of each observation. I also add the square of latitude tocapture any potential non-linearities in the effects of latitude.3. Major axis of continent, that is, longitudinal degrees between the extreme eastern andwestern points of the continent divided by the distance in latitudinal degrees between the extremenorthern and southern points of the continent. The axis ratio goes from 0.1 (in the Pacific Ocean) to3 (for Asia).4. Island, that is, whether the unit of observation (or most of it) belongs to an island or not.Both climate and latitude approximate the environmental conditions most favorable toagriculture. The axis ratio measures the fact that, as emphasized in Diamond (1997), a predominantEast-West orientation speeds up the diffusion of agricultural techniques within continents since newagriculturalists do not need to modify the received plants to new climatic conditions. The dummyisland is a second proxy to tap the effects of oceans as barriers to the adoption of agriculture.[Table 1 about here]The geographical variables of climate, axis and island are statistically significant, behave inthe expected direction and together explain about 55 percent of the variance in the number ofanimals (Model 1), plants (Model 2) and a composite of the two (Model 3). A one-point shift in the25 The coding is as follows: 1 for tundra and ice (climates of type E in Köppen’s classification), drytropical climates (climates of type B) and subarctic climates (Dc and Dd in Köppen’s); 2 for wet tropical(climates of type A); 3 for temperate humid subtropical and continental areas (types Caf, Caw, Da and Dbin Köppen’s system); and 4 for dry hot summers and wet winters (types Cb, Cc and Cs).26

climate variable changes the number of animals by 0.5 and the total of plants by 1.5. The axis ratiohas a strong impact on the number of species that can be domesticated: the maximum value of 3implies about 11 animals and 24 plants. Being an island reduces the number of domesticable plantsand animals by 3 and 10 respectively. Finally, latitude has some effect on the number of plants.Models 4 and 5 report the impact of the environmental conditions on the time to agriculture– with Model 5 excluding the cases of Antarctica. All the variables are strongly significant and the r-squared is 0.54. A shift in the climate index from the worst to the best conditions leads to theintroduction of agriculture 2,800 years earlier. The size of the coefficients for axis ratio and islandunderscore, in turn, the importance of geographical barriers to explain the agricultural revolution. Allother thing equal, agriculture spread 2,000 to 2,500 years earlier in Asia than in America. The seadelayed the process of diffusion by 1,500 to 2,250 years. Figure 1 displays the correlation betweenthe time at each agriculture was introduced and the composite index of biogeographical conditions(the abbreviations correspond to the contemporary country that matches the geometrical median ofthe area and so there could be more than one observation represented by the same abbreviation). 26[Figure 1 about here]State FormationThe process of economic differentiation of individuals and territories (due to a biasedtechnological shock such as the agricultural revolution or the occupation of new lands) is a necessarybut not sufficient condition for the emergence of states. The introduction of political institutionsdepends on the development of some permanent military structures, either directly by the ‘naturalproducers’ or provided by a ‘condottiero’ or pacified bandit. In turn, to set up those structures, thatis, to establish some kind of political hierarchy, the technologies of war must grow in their degree ofsophistication: if all individuals have equal access to very simple weapons (such as rocks and sticks),26 The index is an equally weighted average of climate, latitude, axis and island (set at their relativevalues with respect to the maximum value in each variable).27

they have no incentives either to associate together and spend some time in military training or toaccept the tyranny of a bandit.According to the existing archaeological research, political institutions emerged after (andnever before) the transition to agriculture (cf. Thurston and Fischer, ed. 2007, and the chapterstherein). In the most investigated region where agriculture appeared independently, the Middle East,there were several villages with an estimated population in each one of them of three to fourthousand inhabitants in the southern Levant already in the seventh millennium before Christ (Kuijt2000). Although there are some traces of labor specialization, most studies agree that householdswere the basic economic units in the agricultural Neolithic and that they were “homogeneous withregard to wealth and status” (Voigt 1990: 7). The existing non-domestic structures, such as thedefensive walls and the temple shrine of Jericho, from about 6,500 BC (Yadin 1963, Finer 1997),point to the existence of some collective authority. But they do not appear to embody any statusdifferentiation (Monahan 2007).The first and more systematic historical records attesting the formation of states (and a muchmore hierarchical political structure) only appeared later in time, at around 3,200 BC, that is, a fewthousand years after the domestication of plants and animals, mainly because writing was onlyinvented at the end of the fourth millennium before Christ. As discussed later, it is not coincidentalthat they appear with the use of metals to build weapons. All the first documented states invariablyformed in areas that had first transited to agriculture. At the end of the fourth millennium BC therewere a few city states, with several thousands of inhabitants, a centralized political authority andfortified walls, in southern Mesopotamia. Egypt became unified under a single monarch at around3,000 BC. By the end of the third millennium several palaces were built in Crete, spawning theMinoan civilization. A few hundred years later continental Greece witnessed the emergence ofseveral monarchies within what has been called the Mycenaean civilization. The Harrapan culture,with large urban agglomerations and fortified structures, flourished in the Indus valley in the middleof the third millennium BC. The first firm evidence of a state structure in East Asia has been put at28

around 1,850 BC with the Shang dynasty governing northeastern China. In America, urbansettlements and state structures developed in the two main cradles of agriculture: in Mesoamericaabout two thousand years ago and in the Andean area around AD 500-700 or perhaps earlier.[Figure 2 here]More systematically, Figure 2 explores the relationship between the spread of agriculture(horizontal axis) and the appearance of state institutions (vertical axis) in all the geographicalquadrants of the world (as defined earlier). A state is defined by the existence of a specializedorganization with the quasi-monopoly of coercion over a well defined territory. I code an area ashaving a state if there are contemporary writing materials that report the existence of those structuresor if, absent writing records, there are sufficiently large fortified structures that point to the existenceof armies and states. I employ the information provided in Kinder and Hilgeman (1974), Mann(1986) and Finer (1997) and several regional histories to date the approximate century at whichpolitical institutions formed. In coding the emergence of states I take a conservative approach: forexample I date the emergence of states at around 3,000 BC in Mesopotamia and the Near East (whenwriting records appear for the first time) even though Jericho seemed to have had some fortifiedstructures a few thousand years earlier. Pacific islands are not coded because it is unclear whether(and when) complex tribal chiefdoms were born before Europeans explored that area.Figure 2 reveals two main empirical patterns. First, the transition to agriculture alwayspreceded rather than followed the formation of states. Second, states came into being with a temporallag with respect to agriculture. This temporal gap, which shrank over time, must be related to twofacts: first, the dissemination of intensive agriculture (at around 4,100-4,000 BC according to[Barker 2006]); second, a shift in military technology. Until the end of the fourth millennium, menemployed stone, a material that was too cheap and easily available to give any strong advantage toany of its users, to build weapons. The introduction of copper helmets and swords revolutionized thedynamics of war – it certainly coincided with the appearance of the first Mesopotamian city-states. Afew hundred years later, the invention of bronze, an alloy of common copper and tin, exacerbated29

those trends toward more military specialization (and the creation of states) for two reasons: thehardiness of bronze gave a considerable military advantage to its users; the scarcity of tin and itsvery localized sources “made it a substance on which it was easy to levy high market prices andheavy transport tolls and taxes at the point of delivery” (Keegan 2004: 238), further fostering thatprocess of concentration of military might among certain strata.[Table 2 here]Table 2, which estimates the relationship between the transition to agriculture and theformation of states, confirms the previous discussion. Models 1 and 2 estimate the impact ofbiogeographical conditions on the birth of states, with and without Antarctica respectively. All thevariables are strongly significant and operate as expected. Models 3 and 4 examine the relationshipbetween year of transition to agriculture and year of birth of political institutions (again with andwithout Antarctica). The relationship is extremely strong with and r 2 around 0.6. Since one can claimthat diffusion effects may be important the process of state formation, Model 5 simply looks at thosecases where political institutions formed ‘originally’ in each area. The explained variance is aboutthe same.Violence and Population MigrationsBesides linking the formation of states to the agricultural revolution (and militarytechnologies), the model in Section 1 sheds light on the origins and direction of violence, with thepoor, non-agricultural populations looting the technologically advanced, sedentary societies thatemerged after the domestication of plants and animals. That prediction turns out to be a very strongregularity in human history. Populations in peripheral, less rich lands have tended to breed muchmore war-prone populations than fertile lands. Ancient Mesopotamia and Egypt, the first centers ofagriculture in the Middle East, withstood successive waves of invaders originating in the Caucasusand the Asian steppes such as the Hurrians in the third millennium and the Hyksos in the late secondmillennium. The Roman Empire fell under the pressure of Germanic tribes. Medieval Europe30

endured the attacks of periphery populations such as the Vikings, Slavs, Turks and Mongols.Similarly, China was raided and at times governed by several steppe tribes. In Mexico and CentralAmerica, once-periphery groups such as the Aztecs came to dominate the core lands of the region.Military Technology and Type of Political InstitutionsAs discussed in Section 1, the type of state, ranging from a ‘self-governing’ or republicanconstitution to a monarchical or dictatorial regime, depends crucially on the efficiency θ D with whichthe producers can defend themselves against potential looters. 27 To examine the impact of warmaking on the type of state institutions, I now look at the political consequences of what have beenconsidered five key technological innovations in military weaponry: bronze weapons, the horsechariot, the invention of iron, the stirrup and firepower arms. 28As already mentioned before, the introduction of bronze helmets, swords and ax sockets atthe end of the third millennium B.C. led to the collapse of the Mesopotamian city-states and to theformation of the first big monarchical states in the Middle East – Sargon of Akkad, for example,27 In addition to the type of war technology, the costs of self-defense for producers may be affected by thelevels of competition among looters. A bandit-lord who has to fend off the assault of other agents of hissame type will have fewer resources left to enslave other communities of producers. As emphasized byHansen (2000), many (if not most) of the thirty city-state systems that he has identified emerged in themargins of empires or kingdoms with insufficient means to exercise complete control over their nominalterritories. For example, after the collapse of the Carolingian empire, the exercise of political authoritywas substantially weak in the decentralized areas of the Lotharingia: that may explain why Europeancities emerged mostly in the axis Flanders-North Italy already in the early Middle Ages (Dhondt 1976).28 Although these technological innovations do not exhaust the list of invented weapons (since there areother tools such as the bow, fortifications, etc.), they are arguably the most important in military history(Andreski 1968, Keegan 2004). Besides any strictly technological innovations, there have beenconsiderable organizational and strategic changes in war making. I do not consider them directly forseveral reasons: they are often linked to technological changes (for example, the hegemony of cavalry isdirectly related to the introduction of the stirrup); they are harder to trace and more volatile in nature(with some of them related to the military creativity of a particular general); and they are arguably lessexogenous to politics than pure technological changes.31

controlled an empire of roughly the size of modern Iraq. The predominant use of bronze weaponswas associated with a sharp division between a warrior aristocracy and a peasant class in AncientGreece (Andreski 1968: 43-44).The introduction of two-wheeled horse chariots probably had an even stronger impact on thekind of political institutions. The invention of the spoked wheel and of light chariot bodies combinedwith the use of fast horses (in place of the former onagers) radically altered war making at the end ofthe second millennium B.C. The speed of transportation was multiplied by ten to about 20 miles perhour. Military effectiveness at the battle camp also increased extraordinarily. Since a single chariotcrew (one driver and one archer) could kill about six men a minute circling at a distance of 100 to200 yards, ten chariots (or twenty men) could cause over half a thousand casualties in about tenminutes. The pastoralist peoples in the margins of the Eurasian steppes were the first to employ thehorse chariot – they could easily apply all their riding and hunting skills to these new technologies ofwar (Keegan 2004: 159-66). Attracted by the much richer agricultural lands to the South (and East),they launched successive waves of attack: the Hyksos penetrated Egypt, the Hurrians and Kassitesentered Mesopotamia, the Aryans conquered India and certain tribes originating in the proto-Iranianheartland in the Altai established China’s Shang dynasty. The success of the chariot peoples wasgenerally short-lived: the Assyrian and Egyptian elites copied their war strategies and restored theircontrol over their own lands. But the horse chariot led to the consolidation of a set of ‘feudal’ stateswith a narrow aristocracy of warriors exercising full control over their subjects (McNeill 1982). TheOld Testament provides us with direct, written information on this transformation in Israel. Duringthe period of the Judges, Israel’s army was essentially a militia “made up of warriors from individualtribes who were summoned to battle only in emergency” (Yadin 1963: 255) and the internalorganization and equipment of each tribal unit remained in the hands of each clan. In response to theattacks of the peoples of the sea, David and his descendants professionalized its army, introducing aregular army and a body of heavily armed charioteers, and effectively constructing a centralizedmonarchy in the process.32

Whereas the use of bronze weapons and the horse chariot reduced θ D or the militaryeffectiveness of producers, the discovery of how to make weapons out of iron allowed a “relativelylarge proportion of the male population [to] acquire metal arms and armor” (McNeill 1982:12),mostly because there is a large supply of iron ores in the world (about 4 percent of the earth mass isiron ores). Although with some lag, this led to relatively more egalitarian political and socialstructures. After the expulsion of the Etruscan kings, iron cheapened the access to arms, madeinfantry central in war and secured the political ascendancy of Roman rural strata – to the point thatonly those who were too poor to equip themselves for military service were practically deprived ofthe vote” (Andreski 1968: 54). 29 In Greece, where the irregularity of the terrain (and the type ofcrops) made the use of chariots rather ineffective [check Victor Hanson], the spread of iron weaponsbroke the hegemony of the bronze-wielding aristocracy and equalized conditions. As pointed byAristotle in his Politics, “where the nature of the country can admit a great number of horse, there apowerful oligarchy may be easily established”, “where the troops are chiefly heavy-armed, there anoligarchy, inferior in power to the other [the calvary-dominated one] may be established” and,finally, “the light-armed and the sailors always contribute to support a democracy” (Book 6, Part 7).Thus, whereas Athens was a maritime republic, Thessaly, where the cavalry was always the mainarmy, remained wholly unaffected by democratic movements (Andreski 1968: 45).Widely employed in China by the fifth century AD, the stirrup reached Japan during thefollowing century and Eastern Europe by the seventh century. By the ninth century the stirrup and,with it, heavy cavalry were fully widespread in Europe. Given the high cost of equipping one knight,the rise of the armored horseman had to tip the balance again in favor of the specialized warrior.Under the Sassanid dynasty and after the introduction of heavy protective armor, peasants werereduced to harsh servitude. Under the T’ang dynasty the situation of the Chinese peasant alsoworsened considerably (Andreski 1968: 46-49).29 Even when the armed men had the right to participate in politics, the vote was weighted according tothe costs of the armament, that is, it was biased in favor of the cavalry and the heavy infantry.33

Whereas the impact of heavy cavalrymen operated through cohesive armies and stableimperial bureaucracies in China, in Europe it took place in a rather fragmented and conflict-riddencontinent. The geographical diversity of Europe and the fact that the effectiveness of the stirrup andthe heavy cavalry are heavily dependent on ground conditions afford us an excellent opportunity totest the effect of war technologies on state structures. In the broad plains of continental Europe,where heavy cavalry became the dominant type of military force, feudalism replaced previous formsof social and political organization: that transition was completed after Charlemagne in France andGermany; in Catalonia by the tenth century; in England with the Norman invasion; 30 in Poland afterthe gradual clearance of forests and morasses facilitated the movement of horses; in the lowlands ofthe Balkans with the invasions of the Avars, Bulgars, Magyars and Petchnegs; in Denmark by thebeginning of the lower Middle Ages (Andreski 1968: 59-61, 66-67; Rogowski and McRae 2008). Bycontrast, feudalism did not succeed in those terrains where heavy cavalry was ineffective. Infantryprevailed in Sweden due to the density of forests and in Norway, Switzerland and the highlands ofthe Balkans due to their mountainous nature. There was never serfdom in Sweden, the only countryin Europe where representatives of the peasantry sat in parliament. The inhabitants of the high Swisscantons always governed themselves through communal assemblies. And in the highlands of Serbia,Bosnia and Montenegro “the semi-nomadic mountaineers [lived] under the tribal democracytempered by patriarchy” without any ruling nobles (Andreski 1968: 67). 31 According to Ferejohnand Rosenbluth (2010), medieval Japan presented the same pattern: mountain villagers and islanderscould defend themselves well against the encroachment of horsemen and of the central state untillate in time.30 English feudalism declined after 1300 – arguably after the English learned the use of the long bow intheir wars against the Welsh (Andreski 1968: 64-65).31 In Japan feudalism crystallized with the expansion of heavy cavalry – but faced too pockets ofresistance in mountainous areas (Ferejohn and Rosenbluth 2009).34

The cannon triggered another shift in the internal distribution of political power. The need togenerate economies of scale encouraged the formation of large states with a strong, absolutist crownand led to the demise of the old class of feudal lords (Anderson 1979, Tilly 1990). ItalianRenaissance cities only resisted after their engineers discovered how to build wide earth-based wallsthat could absorb the impact of cannonballs. Such a process of power concentration was onlyreversed by the industrial production of light, one-man, cheap guns and the corresponding creationof standing armies. 324. DISTRIBUTIONAL CONSEQUENCES OF POLITICAL SETTINGSIn the model the final distribution of income is a function of the following factors: the natureof the technological shock (which could in itself lead to more or less variance in income), the finalregime in place, the extraction rate and the size of the ruling class.Technological Shock and Population StructureTo examine the effects of different political institutions on the distribution of income, let usfirst consider the impact of any technological shock on population. Following the literature on preindustrialeconomies, assume that net population growth (ΔL) tracks the rate of technological change32 I am not suggesting that the discovery of standing armies mechanically triggered the process ofpolitical liberalization that has taken place in the last two hundred years. All absolutist states blocked theintroduction of national armies for as long as it was feasible because they contributed to undermine theposition of monarchs and aristocracies. The French revolution broke down an implicit oligopolisticagreement among absolutist kings to block a new technology of war. In fact, after the defeat of Napoleon,the Austrian Prime Minister Metternich insisted on an explicit pact to abolish conscription at theCongress of Vienna. However, once the the battle of Valmy (1793) and all the Napoleonic campaigns haddemonstrated the superiority of universal conscription, most states ended up adopting some kind ofstanding army over time: Prussia immediately after the defeat at the hands of Napoleon, Russia after theCrimean war and Austria after its defeat at the hands of Germany. For recent analysis suggesting thatmodern war contributed to extension of political rights and/or to higher levels of taxation, see Levi(1997), Peacock and Wiseman (1961) ans Scheve and Stasavage (2010).35

(ΔA). That assumption implies that population adjusts so that output per person equals somesubsistence income ŷ = Y/L. This fits two key features of pre-industrial societies. First, the level ofmaterial welfare in pre-industrial societies is not very different, at least in terms of vital statistics:food consumption per capita, life expectancies and health profiles have been found to be similaracross foraging and agricultural societies (Clark 2007: 40-70, 91-111). Second, population densitiesare much higher in technologically ‘advanced’ (agrarian) societies than in hunting-gatheringterritories. In the latter, population densities fluctuate, mostly depending on climate and landconditions, between 2 (in a few cases) to less than 0.5 persons per square kilometer (Kelly 1995). Bycontrast, recent archaeological studies suggest a density of about 90 persons per square kilometer inthe first Neolithic villages in the Levant about 8,000 to 9,000 years ago (Kuijt 2000).To calculate the population growth rate (and hence the size of the population) given the levelof technology, take the previous function of ŷ, substitute the production function introduced inSection 1 for Y and solve for L:1−α1 1/ αL = ( ) AαT(12)yˆSince T and ŷ are constant, the growth rate of L is (1-α)/α times the growth rate of A.Given (12) and assuming that the share of the land factor in total output (with no otheroutput than the one from agricultural activities) was about 2/5 (α=0.4) (Clark 2007: 138), Figure 3.Adisplays the evolution of the population (in relative terms) in both foraging and agrariancommunities as the technology changes for the latter one. As soon as the technical know-howdoubles, the agricultural population becomes two and half times larger than the foraging population.For a technology ten times more advanced, the agricultural population becomes over thirty timeslarger than the hunting one (for the same unit of land).[Figure 3 here]The Consequences of Regime Type36

Before the state is established (and institutionalized peace is in place), each individualreceives an income y i and there is no inequality beyond the (relatively bounded) one determined bythe productivity A i of each individual. In a monarchical system, instead, the ruler siphons off a shareε of every individual to himself (and his servants and allies) and hence gets (ε-γ)λy i N with N denotingthe number of subjects while each subject keeps (1-ε)y i in each period. (For any minimal number ofsubjects, this process of redistribution always increases inequality.) By contrast, in a republican orself-governing polity every citizen keeps (1-1/θ D )y i and inequality does not vary with respect to thepre-shock world. (For the sake of simplicity I will assume λ=1 throughout this section).[Figure 4 here]Productivity Shock and Income Distribution. Employing the parameters of the model,Figures 4 and 5 simulate the distribution of income under different economic and politicalconditions. Figure 4.A displays the evolution of material conditions for i and j individuals as the iindividuals benefit from a growing productivity shock (represented in the horizontal axis). Figure4.A assumes that, once the shock makes the spontaneous peace equilibrium across communitiesunfeasible, the j-individuals organize and have a better military technology than producers, impose amonarchy and extract some of the producers’ income (at the highest feasible extraction rate). 33 Asthe productive shock increases and provided the monarchical regime is in place, the position of theruling class improves to the point that for A=10 is eighteen times higher than the producers’ percapita income. The Gini index goes up to around 0.29 if the productive individuals double theiroutput and the ruling class in a monarchical system imposes the corresponding highest feasibleextraction rate, that is, 48 per cent, to about 0.34 percent.Size of the Ruling Group. As noted before, holding the feasible rate of extraction constant,the distribution becomes the more unequal, the smaller the number of rulers relative to thepopulation of the country. Figure 4.B examines the effect of reducing the size of the ruling groupwhile holding the extraction rate fixed (at 0.6). The Gini index goes from less than 0.1 to around 0.6.33 The parameters assumed are δ=0.7, θ D =0.6, λ=0.5.37

Variable Extraction Rate. Figure 4.C explores the impact of a changing extraction rate onthe final distribution of income. For the same proportion of rulers, the Gini index increases from 0.17to 0.44 as the extraction rate climbs up to 70 percent.[Figure 5 here]Monarchies versus Republics. Figure 5 compares the income distribution under amonarchical and republican regime. To make the analysis richer, the baseline distribution of incomeis different than in Figures 4.A to 4.C. Instead of two types, the shock differentiates the populationinto ten segments or deciles. 34 The tenth decile does not experience any shock (y 10 = 1) – itcorresponds to the type-j individuals of Figure 4. The other deciles experience an increase inproductivity – but the extent of the raise varies across them. However, the increase is bounded sothat they have an incentive to cooperate among them. (The lowest increase, the one experienced bythe first decile, is big enough to make it impossible for those individuals that do not get any increase(those in the 10th decile) to maintain a cooperative relationship with the rest of the population.) InFigure 5 the post-shock income ranges from 2 for the first decile to 3.58 for the ninth decile (thetenth decile keeps receiving 1). A monarchical system leads to a strong redistribution of income. Thetenth decile becomes the richest one: it has a per capita income that is about nine times larger thanthe poorest decile’s and it controls almost 39 percent of all the wealth. The Gini index more thandoubles to 0.35 under a monarchy. By contrast, a republican system reduces the initial distributionslightly – as a result of the tax burden to defend the human settlement. Under an “imperial republic”(where the tenth decile is expropriated by the majority), inequality does not increase by much: evenwith the highest extraction rate, the Gini index inches up from 0.14 to 0.19.34 This differentiation is done by assumption. The assumption can be justified as follows: consider thecase in which different segments experience different technology shocks and choose the same populationgrowth rate, unlinked to the technological growth rate (in contrast to expression (12) where both werefunctionally related).38

Although I do not consider the following point in the simulation, the distribution of incomemay be also shaped by the way in which rulers extract their resources or, in other words, what kindsof alliances they form with different strata of producers. The distribution of income becomes highlyskewed when the rulers ally with the richest fraction of the population (this would be equal toimposing a lower extraction rate ε on them than on other groups) or where the ruling elite simplyreplace the existing segment of wealthy individuals. By contrast, the ruling elite may favor a moreequalized structure of income within the producers’ population. These different outcomes, resultingfrom different institutional (and taxation) choices, are particularly apparent in processes ofrepopulation and colonization. After the Christian kingdoms of Northern Spain took over most of theSouthern half of the Iberian Peninsula in the 13th century, they differed markedly in their strategiesof land redistribution. Castile and Portugal favored the formation of large landholdings. Catalanssettled in Valencia through small farms (Vicens Vives 1957). Similarly, the conquest of Americafollowed very divergent paths: farming communities settled in the Northeastern seaboard and inQuebec; the Caribbean basin and Central and South America were dominated by landowners. In allthose cases the distribution of land partly responded to economic considerations, such as the type ofnatural endowments and the supply of labor (Engerman and Sokoloff 2002). But it was stronglyaffected by the way in which the ruling elites decided to structure the distribution of assets. That wasin turn related to the prevailing political institutions and societal conditions back in the conqueringcountries: the Castilian monarch relied upon the high nobility and military orders to conquerAndalusia in the thirteenth century.Some EmpiricsThe simulated extraction rates and Gini indexes fall in line with the available historicalevidence. According to data gathered by Milanovic et al. (2007) and reproduced in Table 3, inalmost all pre-industrial regimes the Gini index fluctuated around 0.40 and in most cases the top 1039

percent of the distribution controlled more than two fifths of total income. Bourguignon andMorrisson (2002) offer similar data across the world in 1820, at the dawn of the industrialrevolution.[Table 3 here]Given the lack of reliable data on income distribution before the 19 th century, one possiblestrategy to understand the income consequences of regimes is to employ height as a proxy for accessto resources, health status and general welfare. As discussed in Steckel (1995), height and individualincome seem to be strongly correlated and, as a result, the distribution of heights should provide uswith clues about how that society distributed its resources, at least in human communities livingunder conditions of scarcity. 35 In fact, we know that insufficient nutrition was widespread and thatthe access to food was biased across social strata in pre-industrial economies: for example, inEngland in 1790 whereas the top decile consumed 4,329 kg cal per day; the bottom decile consumed1,545 kg cal per day (Fogel 1994).Table 4 summarizes a more extensive discussion of data reported in Boix and Rosenbluth(2006) on the level of variance in height across times and within different societies. Inequality is lowwithin tribal, pre-agrarian groups. With the transition to a fully agricultural society with complexpolitical institutions, heights vary strongly with the type of political regime. In ancient kingdoms andabsolute monarchies, there is considerable differentiation among individuals: the height gap betweennobles and peasants goes from 7 to 9 cm. Using fitted models between log income (in constantdollars of 1985) and height (Steckel 1995), this difference translates into an income ratio of 5 or 6 to1 between both groups.36By contrast, more democratic settings (although still predominantlyagrarian such as Ohio in middle of the 19 th century) show a much more compressed height structure.35 For a critique, mostly directed at crude cross-country or cross-ethnic group comparisons, see Eaton(2007).36 This ratio matches the ratio of the 95 th centile/50 th centile in the data Bourguignon and Morrisson(2002) offer for several (46) world areas in 1820.40

[Table 4 here]5. CONCLUSIONSBroadly speaking, the distribution of income in human societies has been characterized bythe following pattern. First, primitive (stateless, pre-agrarian) societies displayed (and still displaywherever they exist) relatively equal distributions of income and certainly very low level ofintergenerational transmissions of wealth. Second, the agricultural revolution came hand in handwith the emergence of marked inequalities of income (across individuals and over generations) andthe formation of the state. Third, the type of political institutions and the level of inequality hasvaried considerably across historical communities since the invention of agriculture.This paper develops a theoretical model to explain the transition away from relatively equalpre-agrarian societies. After the end of the last glacial period, a process of learning andexperimentation resulted in territorially differentiated technological shocks that generated inequality(within but mostly across human bands). That inequality made spontaneous cooperation unfeasible.That opened the way to both political institutions and further shifts in income distribution. Inresponse to the productivity growth, the more productive individuals responded by accepting (even ifreluctantly) the protection of their potential looters in exchange for some systematic payments.Alternatively, they could choose to devote part of their time and income to defend themselvesagainst those bandits. Both outcomes (‘monarchical’ and ‘republican’), strongly driven by theevolution of military technologies, had distinctive effects on the distribution of resources – withmonarchical regimes generating and sustaining much higher levels of inequality than republicansystems. The theoretical results match the existing descriptive work as well as broader empiricalwork compiled in Boix and Rosenbluth (2006).Understanding the sources of states institutions and of income inequality is in fact importantto explain economic development and the contemporary process of democratization. The emergenceof the state had a considerable stabilizing effect on the distribution of income over time. As41

discussed in Boix (2010), authoritarianism, economic stagnation and inequality came together in anoverwhelming majority of Ancien Régime polities -- income inequality only started to trenddownward in a systematic fashion after the industrial revolution. That was a direct outcome of theincentive structure of their ruling elites. The latter’s final income depended, directly, on the returnsto their assets and, indirectly, on the extraction rate embodied on the existing political institutions,themselves endogenous to the prevailing income distribution. Hence, those governing had a directinterest in protecting the status quo and in blocking any process of economic and politicalliberalization.42

BIBLIOGRAPHYAcemoglu, Daron and James A. Robinson. 2006. Economic Origins of Dictatorship and Democracy.New York: Cambridge University Press.Andreski. 1968.Aristotle. [1988] The Politics. New York: Cambridge University Press.Atkinson, A.B. and F. Bourguignon. 2000. “Introduction: Income Distribution and Economics.” In A.B.Atkinson and F. Bourguignon, eds. Handbook of Income Distribution. Elsevier Science. Pages 1-58.Bar-Yosef, Ofer and Richard H. Meadow. 1995. “The Origins of Agriculture in the Near East,” in T.Douglas Price and Anne B. Gebauer, eds. Last Hunters, First Farmers. Santa Fe, NM: School ofAmerican Research Press. Pages 39-94.Bates, Robert. 2001. Prosperity and Violence. The Political Economy of Development. New York:Norton.Bellwood, Peter. First Farmers. The Origins of Agricultural Societies. Malden MA: BlackwellBender. 1978.Binford. 1968.Boix, Carles. 2003. Democracy and Redistribution. New York: Cambridge University Press.Boix, Carles. 2008. “Economic Roots of Civil Wars and Revolutions in the Contemporary World,” WorldPolitics 60 (April): 390-437.Boix, Carles. 2010. “Origins and Perisistence of Inequality.” Annual Review of Political Science.Boix, Carles and Francis Rosenbluth. 2006. "Bones of Contention: The Political Economy of HeightInequality." Unpublished manuscript.Boix, Carles and Milan Svolik. 2010. “The Foundations of Limited Authoritarian Government.Institutions and Power-Sharing in Dictatorships.” Unpublished manuscript.Bollen, Kenneth A. and Pamela Paxton. 1997. “Democracy before Athens.” in Democracy andInequality, edited by Manus Midlarsky. New York: Cambridge University Press.Borgerhoff-Mulder, Sam Bowles et al. 2008. “The Inheritance of Inequality in Pre-Modern Societies.”43

Bourguignon and Morrisson. 2002.Carneiro, Robert L. 1970. “A Theory of the Origin of the State,” Science 169 (August): 733-738.Chagnon, Pierre. 1997. Yanomamö: Case Studies in Cultural Anthropology. Wadsworth Publishing.Clark, Gregory. 2007. A Farewell to Alms: A Brief Economic History of the World. Princeton UniversityPress.Clastres, Pierre. 1972. Chronique des Indiens Guayaki. Paris: Editions Plon.Clastres, Pierre. 1974. La société contre l’État. Recherches d’anthropologie politique. Paris: Les Ëditionsde Minuit.Cohen. 1977.Dal Bo and Fréchette. 2009.Deaton, Angus. 2007. “Height, health, and development,” Proceedings of the National Academy ofSciences 104 (33):13232-7Diamond, Jared. 1997. 1999. Guns, Germs and Steel. New York: Norton.Dixit, Avinash. 2008. “Elite-Led and Majoritarian Institutional Reforms.” Princeton: Unpublishedmanuscript.Dhondt, Jan. 1976. Le haut Moyen Age : VIIIe-XIe siècles. Paris: Bordas.Engerman, Stanley L. and Sokoloff, Kenneth L. 2002. “Factor Endowments, Inequality, and Paths ofDevelopment among New World Economies,” Economia 3: 41-102.Finer. 1997.Flannery 1965.Fogel, R.W. 1994. “Economic growth, population theory, and physiology: The bearing of long-termprocesses on the making of economic policy,” American Economic Review 84: 369-395.Gat, Azar. 2000. “The Human Motivational Complex.” Anthropological Quarterly.Geddes, Barbara. 2007. “Democratization.” In Carles Boix and Susan Stokes, eds. Oxford Handbook ofComparative Politics. New York: Oxford University Press.Gordon and Dew-Becker. 2008.Gottschalk and Smeeding. 1997.44

Hansen, Mogens H., ed. 2002. A Comparative Study of Six City-State Cultures. Copenhagen: The DanishRoyal Academy.Hibbs, Douglass and Ola Olsson. 2004. “Geography, Biogeography and Why Some Countries Are Richand Others Poor,” Proceedings of the National Academy of Sciences of the United States 101(10): 3715-3740.Hirschman, Albert O. 1981. Essays in Trespassing. Economics to Politics and Beyond. Cambridge:Cambridge University Press.Kahhat. 2007.Keegan. 2004.Kelly, Robert. 1995. The Foraging Spectrum: Diversity in Hunter-Gatherer Lifeways. SmithsonianInstitution Press.Kinder, Hermann and Werner Hilgemann. 1974. The Anchor atlas of world history from the Stone Age tothe eve of the French Revolution. Garden City, N.Y.: Anchor Books.Kuijt, Ian. 2000. “People and Space in Early Agricultural Villages: Exploring Daily Lives, CommunitySize, and Architecture in the Late Pre-Pottery Neolithic,” Journal of AnthropologicalArchaeology 19: 75–102.LeBlanc, Steven. 2003. Constant Battles: The Myth of the Peaceful, Noble Savage. St. Martins.Levi, Margaret. 1988. Of Rule and Revenue. Berkeley, CA: The University of California Press.Levi, Margaret. 1997. Consent, Dissent, and Patriotism. New York: Cambridge University Press.Lindert. 2000.Machiavelli, Niccolò. 1531 [1970]. The Discourses. London: Penguin.Mann. 1986.McNeill, William H. 1982. The Pursuit of Power. Chicago: The University of Chicago Press.Milanovic, Branko, Peter H. Lindert and Jeffrey G. Williamson. 2007. “Measuring Ancient Inequality,”NBER Working Papers # 13550.North, Douglass. 1981. Structure and Change in Economic History. New York: Norton.45

Olson, Mancur. 1993. "Dictatorship, Democracy, and Development," American Political Science Review87: 567-576.Olson, Mancur. 2000. Power and Prosperity. New York: Basic Books.Olsson, Ola and Douglas Hibbs. 2005.O’Rourke, Kevin H. and Jeffrey G. Williamson. 1999. Globalization and History: The Evolution of aNineteenth-Century Atlantic Economy. Cambridge, Mass: The MIT Press.Panter-Brick, Catherine, Robert H. Layton and Peter Rowley-Conwy, eds. 2001. Hunter-gatherers: aninterdisciplinary perspective. New York: Cambridge University Press.Peacock and Wisemann. 1961.Piketty, Thomas. 2000. “Theories of Persistent Inequality and Intergenerational Mobility.” In A.B.Atkinson and F. Bourguignon, eds. Handbook of Income Distribution. Chapter 8. ElsevierScience. Pages 429-476.Price, T. Douglas and Anne B. Gebauer. 1995. Last Hunters-First Farmers: New Perspectives on thePrehistoric Transition to Agriculture. Santa Fe, New Mexico: School of American ResearchPress.Rindos. 1978.Rogowski, Ronald and Duncan McRae. 2008. “Inequality and Institutions: What Theory, History and(Some) Data Tell Us.” In P. Beramendi and Christopher J Anderson, eds., Democracy, Inequalityand Representation. New York: Russell Sage Foundation.Rowley-Conwy, P. 2001. “Time, change, and the archaeology of hunter-gatherers: how original is the‘Original Affluent Society’?” In C. Panter-Brick, R. H. Layton, and P. Rowley-Conwy (eds.)Hunter-Gatherers: An Inter-Disciplinary Perspective. Pages 39-72. Cambridge: CambridgeUniversity Press.Rousseau, Jean-Jacques. 1755 [1973]. A Discourse on the Origin of Inequality. In The Social Contractand Discourses. London: Everyman’s Lybrary.Scheve, Kenneth and David Stasavage. 2007. “Political Institutions and Income Inequality, 1900-2000.”World Politics.46

Scheve, Ken and David Stasavage. 2010.Steckel, Richard H. 1995. “Stature and the Standard of Living,” Journal of Economic Literature 33(December): 1903-1940.Tilly, Charles. 1990. Coercion, Capital and European States, AD 990-1992. Blackwell.Turney-High. 19xx.Vicens Vives, Jaume. 1957. Historia económica y social de España y América. Barcelona: Teide.Wallerstein, Michael. 1999.Watson, Patty Jo. 1995. “Explaining the Transition to Agriculture,” in T. Douglas Price and Anne B.Gebauer, eds. Last Hunter, First Farmers. Santa Fe, NM: School of American Research Press.Pages 21-38.Wright, Henry T. 1977. “Recent Research on the Origin of the State,” Annual Review of Anthropology 6:379-397.Yadin. 1963.47

TABLE 1. ESTIMATING THE CAUSES OF THE AGRICULTURAL REVOLUTIONModel 1 Model 2 Model 3 Model 4 Model 5Dependent Variable Animals Plants Animals Years to YearsAnd Plants Agriculture to AgricultureConstant -3.023*** -5.214** -24.474*** 11,252.226*** 11,140.786***(0.683) (2.506) (7.273) (455.150) (461.598)Climate 0.492*** 1.540** 5.148*** -830.410*** -959.365***(0.181) (0.665) (1.9292) (120.751) (125.037)Latitude 0.008 0.0781*** 0.169*** 8.056** -27.554***(0.005) (0.020) (0.058) (3.614) (7.992)(Latitude) 2 -0.017 0.0080 -0.061 73.574*** 131.756***(0.011) (0.042) (0.120) (7.531) (13.869)EW Axis / NS Axis 3.778*** 8.690*** 33.461*** -1,479.413*** -1,169.922***(0.350) (1.283) (3.723) (233.010) (243.907)Island -3.329*** -9.946*** -33.035*** 2,299.943*** 1,688.177***(0.502) (1.842) (5.345) (334.573) (360.257)Observations 234 234 234 234 203R-squared 0.5580 0.4693 0.5303 0.5477 0.5485Standard errors in parentheses. *** p

TABLE 2. ESTIMATING THE EMERGENCE OF STATESDependent Variable Emergence of State ------------------------------------------------ Emergence of State(Original & Derivative)(Only Original)Model 1 Model 2 Model 3 Model 4 Model 5Constant 2,540.131*** 2,474.894*** 1,263.651*** 1,255.726*** 1,266.825***(211.439) (218.471) (43.309) (51.029) (173.048)Climate -260.769*** -311.078***(54.822) (57.753)Latitude 4.693*** -9.436**(1.682) (3.758)(Latitude) 2 22.853*** 45.736***(3.443) (6.455)EW Axis / NS Axis -928.002*** -792.815***(112.164) (119.894)Island 706.341*** 448.982***(155.587) (171.475)Year of Transition 0.327*** 0.324*** 0.446***To Agriculture (0.016) (0.019) (0.046)Observations 229 198 229 198 46R 2 0.449 0.454 0.638 0.607 0.684Standard errors in parentheses*** p

TABLE 3. SOME INEQUALITY STATISTICSGiniTop 10 percentRoman Empire ca. 14 AD (a) 0.36 0.44Byzantium 1000 (a) 0.41 0.46England/Wales 1688 (a) 0.45 0.38Moghul India 1750 (a) 0.39 0.35Mexico 1790 (a) 0.64 0.61Naples 1811 (a) 0.28 0.35Brazil 1872 (a) 0.39France 1780 (b) 0.56(a) Milanovic et al. (2007).(b) Morrisson 200050

TABLE 4. POLITICAL SETTINGS AND DIFFERENCES IN MALE HEIGHTSINSTITUTIONS, PLACE MEAN DIFFERENCE(IN CM.) (IN CM.)Pre-State SocietiesZuni Pueblos (before 1680 AD.)Height at 90 th centile 165.8Height at 50 th centile 161.1 4.7Monarchical/AuthoritarianAncient Egypt (New Kingdom)Royal family 174Commoners 166 8.0MycenaeRoyal 172.5Commoners 166.1 6.4Poland – 17 th cent.Noble 170.6Rural Non-Jewish 165.2Jewish 162.9 7.7DemocraticOhio Recruits, Mid 19 th CenturyLaborers 173.3Skilled Workers 174.0Farmer 174.7Professional 175.5 2.2Source: Boix and Rosenbluth (2006).51

Years to the Introduction of Agriculture0 5000 10000 15000Figure 1. Transition to Agriculture and Biogeographical ConditionsAULAUL AULJPNLIBAUL AUL SAU UAE ALGRUSRUSRUSARGUSACANUSACANARG ARG ICECAN USACHLRUSCANHAI CUB NEW SURINEWMAG MAGNAM BOT MZMUSAFCANANG ZAM MAWZAI TAZFINVEN NIG CAO CENSENCHNRUSKZKMON CHNNMARI PHI PNGINSOMNIRCHA MLIBRA COLBRA BRAIVO MAA BOL CHL HON PAR THI ALGBRU INSARGURUALGUSA MORPER PERMEXETIUSAJPNSUDINDINDIND MYACHNUSAROK LITRUS NOR SWDDENUKMEXNTHEGYCHN SUDCHNYEMRUS SPNFRNAFG IND RUM CRORUSCHN GRCIRNIRQSYR.2 .4 .6 .8 1Biogeographical Conditions (Climate, Latitude, Ratio Axis, Continental Land)Year of Emergence of State Institutions-8000 -6000 -4000 -2000 0 2000Figure 2. Transition to Agriculture and Birth of State InstitutionsSUD ALGPNGINSCENALGLIBRUSRUSRUSZAIETI CAOZAIANG ZAMMAW NAM BOT MZMMAG SAFNEWAULBRAPARBRA ARG URUARGARG CHLMEXUSAUSAUSACHL IVO COL VEN USA CANSURIDOM HAI BAH CUBCANUSACANSOMNIRNIG RUSKZKMONJPNSWDFIN TAZRUS PHI CHAYEMNOR MLI PERBOLCHNSENICEMAASAUMEX MYA JPN HONRUSNTHUKDENINS INS INSCHNCHN IND THICHNIRN AFG CHNSPNRUS RUSSUDALGFRNUAEMORSYRCRORUM CHN IND IND ROKGRC CHNCHNIND PAKIRQEGY-8000 -6000 -4000 -2000 0 2000Aprroximate transition year to agriculture. Based on Bellwood and on Parker52

35Figure 3.Technological Change and Population Growth3025Number of individuals201510501 2 3 4 5 6 7 8 9 10Level of Technology (Value of A)Population of ProducersPopulation of Bandits0.9Figure 4.A. Monarchical Rule and Productivity Shock200.8180.716Proportion / Gini Index0.60.50.40.314121086Per Capita Income0.240.1201 2 3 4 5 6 7 8 9 10Level of Technology (Value of A)0Percentage of Income Controlled by Monarchical ClassPer Capita Income of Monarchical ClassGini IndexPer Capita Income of Producer53

0.9Figure 4.B. Monarchical Regime and Change in Size of Ruling Group200.8180.716Proportion / Gini Index0.60.50.40.314121086Per Capita Income0.240.12000 0.1 0.2 0.3 0.4 0.5Proportion of Monarchical ClassPercentage of Income Controlled by Monarchical ClassPer Capita Income of Ruling GroupGini IndexPer Capita Income of Producers0.9Figure 4.C. Monarchical Rule and Varying Extraction Rate200.8180.716Proportion / Gini Index0.60.50.40.314121086Per Capita Income0.240.12000.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70Extraction RatePercentage of Income Controlled by Monarchical ClassPer Capita Income of Ruling GroupGini IndexPer Capita Income of Producers54

10Figure 5. Income Distribution Under Monarchical and Republican Regimes98Per Capita Income for Each Decile765432101 2 3 4 5 6 7 8 9 10DecileIncome after shock without state Income under Monarchy Income under Republic Income under Imperial Republic55