ANRV412-PL13-24 ARI 14 April 2010 18:9

Origins and Persistenceof Economic InequalityCarles BoixDepartment of Politics and Woodrow Wilson School of Public and International Affairs,Princeton University, Princeton, New Jersey 08540; email: cboix@princeton.edu

Annu. Rev. Polit. Sci. 2010. 13:489–516

First published online as a Review in Advance onFebruary 24, 2010

The Annual Review of Political Science is online atpolisci.annualreviews.org

This article’s doi:10.1146/annurev.polisci.12.031607.094915

Copyright c© 2010 by Annual Reviews.All rights reserved

1094-2939/10/0615-0489$20.00

Key Words

development, state, institutions, democracy

AbstractAfter reviewing the current literature on the causes of economic in-equality, the article models the historical emergence of inequality as theresult of a key technological change (i.e., the adoption of agriculture)that widened income differentials and led to the construction of stateinstitutions, which shaped (depending on their particular nature, moreor less authoritarian) the final distribution of economic assets withinand across different societies. The article then explores the evolution ofinequality in societies already endowed with state structures: A streamof biased technological shocks happens randomly and the “decisive”voter (who differs across political regimes) accepts or blocks them as afunction of their effect on her net income. The decisive voter’s responsedetermines the overall distribution of income. The model is employedto give a coherent account of some broad historical trends in the evo-lution of income inequality.

489

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

INTRODUCTION

Inequality—its causes as well as its political andsocial consequences—has been a long-standingconcern in political economy. Both Aristotleand Machiavelli associated a particular distribu-tion of wealth with the nature of political insti-tutions.1 Rousseau devoted his whole Discourseon the Origin and Basis of Inequality among Mento explore the causes (the need for social recog-nition and the exploitation of man by his ownkind) behind the destruction of the moral andpolitical equality that he ascribed to mankind inits infancy.

By comparison with those classic endeav-ors, contemporary research on the roots of in-equality remains rather limited. In recent years,economists have made considerable efforts toinvestigate the effects of inequality on growthand development (Atkinson & Bourguignon2000, Kahhat 2007). In political science, thereis now a burgeoning literature on its impact onpolitical institutions and conflict (Boix 2003,2008; Acemoglu & Robinson 2006; Geddes2007). However, very little theoretical workhas been done on the underlying causes of theevolution of inequality. Most contributions arestrict accounting exercises or limit themselvesto economic variables such as imperfect creditmarkets, which are in turn left unexplained.

Moreover, probably owing to the scarcity ofdata, those researchers who have worked onthe evolution of income distribution have ex-clusively focused on its evolution in advancedcountries in the past three decades (Gottschalk& Smeeding 1997; an exception is Rogowski& McRae 2008). Yet any reasonable inquiryabout the origins and varying persistence of in-equality should tackle a much broader temporalspan and, in particular, should account for thefollowing set of stylized facts. First, primitive,stateless communities tended to display rela-tively equal distributions of income and wealth.Although they had some internal heterogeneityin status and prestige as well as some differential

1Aristotle, Politics, IV, p. 11; Machiavelli, Discourses on the FirstTen Books of Titus Livy, Book I, Ch. 55.

access to a few goods (such as sexual partners),their overall level of social and economic strat-ification was limited and they had similar inter-personal patterns of consumption and individ-ual welfare.2 Second, that relative equality givesway to much wider distributions of income andwealth—in terms of patterns of habitation, ac-cumulation of valuable assets, and even healthand height (Boix & Rosenbluth 2006)—afterthe agricultural revolution and the formationof states. Third, the extent of social and eco-nomic inequality in agricultural societies hasvaried considerably across areas and historicalperiods. Finally, income inequality has trendeddownward, albeit with a few national exceptionsand temporal reversions, in the 150 years sincethe spread of industrial technologies.

Given the incipient state of the literature onthe causes of wealth and income inequality, thefirst part of this article reviews the existing theo-retical and empirical research on that question.The remainder of the article then offers a pos-sible analytical strategy that draws on both eco-nomic and political approaches to account forthe origins and varying persistence of inequal-ity. The second part sketches a model to explainthe transition from relatively equal primitivecommunities with no full-fledged state insti-tutions to human societies with stable politi-cal institutions and various degrees of economicinequality. According to that account, the adop-tion of agriculture both widened income differ-entials and led to the construction of state insti-tutions, which then shaped (depending on theirparticular nature, monarchical or “republican”)the final distribution of economic assets withinsociety and across different societies. The thirdpart turns to explore the effects of technolog-ical shocks in “modern” political economies,that is, those societies already endowed with

2See Clastres (1972, 1974) for a discussion of the hunter-gatherer societies of the Ache and Chagnon (1997) on theYanomamo. More generally, see Kelly (1995) and Panter-Brick et al. (2001). On hortoculturalists, see Price & Gebauer(1995). Comparing the Gini coefficients of three contempo-rary hunter-gatherer societies with classical Athens, Bollen& Paxton (1997) conclude that the former are all more egal-itarian, particularly with respect to women.

490 Boix

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

state structures. Making use of the basic build-ing blocks of the most recent literature on pos-itive political economy, it suggests an analyt-ical framework in which the “decisive” voter(who differs across political regimes) accepts orblocks a given technological shock as a func-tion of its effect on her net income. The decisivevoter’s response then determines the overall dis-tribution of income. The model is employed togive a coherent account of some broad histori-cal trends in the evolution of income inequality.

STATE OF THE LITERATURE

Accounting Models ofIncome Distribution

A first and quite extended approach to explaininequality consists in depicting any income dis-tribution as a function of both the distributionof factor endowments among individuals andthe prices paid to factors. As an accounting ex-ercise, this research strategy has considerableempirical purchase: It generates valuable de-scriptions of different “moments” of inequality.As a result, it has been directly linked to theanalysis of factor shares in national incomes.It has been employed also within the traditionof general equilibrium models to examine theimpact of technological change on the distri-bution of human capital and therefore societalincome, of population shifts from rural to ur-ban labor markets, and of international trade onrewards to different economic factors. Atkinson& Bourguignon (2000, pp. 5–13) provide an ex-cellent summary of this class of theories.

Factor-share models of income distributionrun, however, against two theoretical and em-pirical roadblocks. First, they take as given thedistribution of factor endowments among in-dividuals at each period or slice of time, andtherefore they can make sense of the evolutionof inequality only in conjunction with some ex-ogenous theory of economic change. This isunsatisfactory for those who wish to determinethe extent to which inequality affects growth(either directly or through its impact on poli-cies and institutions affecting development).

Second, factor-share models of income distri-bution assume mechanisms of price determina-tion (mostly competitive markets) that are fartoo restrictive for the purposes of explainingincome inequality. A key point of this articleis that political and economic agents imposebarriers to entry and hence distort markets tocapture rents, which then lead to more or lessinequality.

Take Kuznets’ path-breaking 1955 paper oneconomic growth and income inequality as anexample of this line of inquiry. (Kuznets’ paperis in fact simpler than standard factor-share the-ories of income distribution because it doesnot even endogenize how a shift in factor sizesmay affect the rate of return to each factorin equilibrium.) According to Kuznets (1955),the process of economic development triggersa population shift from rural to nonagriculturalsectors. Given certain assumptions about themean and variance of income in each sector,growth can lead to a temporary rise in overall in-equality. This famous Kuznets conjecture aboutthe curvilinear relationship between inequalityand development has spurred an extraordinaryamount of empirical research, albeit with ratherinconclusive results.3 But it does not contain acomplete theory of the emergence and dynam-ics of inequality—it simply relies on some kindof exogenous technological shock that gener-ates changes in factor sizes and incomes.

Theories of (Intergenerational)Persistence of Inequality

The bulk of the contemporary economic re-search on inequality has been devoted to theconstruction of models to explain the temporal(intergenerational) persistence of inequality.

In their simplest version, these models takethe following function form:

ki,t+1 = f (ki,t) − c i,t + εt, 1.

3The lack of consistent empirical results is unsurprising.Kuznets himself stressed that development does not neces-sarily lead to more inequality in all instances. For a summaryand discussion of existing evidence on the Kuznets curve, seeLindert (2000).

www.annualreviews.org • Origins and Persistence of Inequality 491

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

where k represents the assets of individual i, c isthe fraction any individual i consumes at periodt, and ε is some idiosyncratic shock that mayaffect assets k at time t.

The most general (and plausible) interpre-tation of this functional form is a world wherethere is some intergenerational transmission ofwealth among families or individuals. Individ-uals have some initial wealth k, inherited fromtheir parents, which they allocate to maximizetheir expected wealth at the end of the period.After consuming some of their wealth, they be-queath the rest to their children.

The main point of contention among re-searchers working with this class of models hasto do with the particular functional form thatEquation 1 takes—and, particularly, with thebehavior of the function f (.) and the parameterε.

The parameter ε embodies what Becker &Tomes (1979) refer to as “inequality in luck.”It may be identical for all agents, or it may bespecific to each one and correlated with someparticular trait such as income. With particu-lar values of f (.)—those that do not amplifyinequality over time—ε has been shown to gen-erate, in the long run, a unique limiting distri-bution (with perfect mobility and no povertytraps).

More important, the structure of the pro-duction function f (.) and of investment deci-sions determines the extent to which inequalitydeclines, persists, or even rises over time. Sev-eral models portray the processes of produc-tion and investment as having an equalizationeffect on the distribution of income. The sim-plest way of generating this result is, followingSolow’s aggregate neoclassical growth model,to assume a world in which each individual in-vests k and where there is a common productionfunction f (k) that is strictly increasing and con-cave. With no credit markets in place, the ex-pected marginal return on investment is higherfor poor entrepreneurs, and the average wealthof poor families grows faster than the averagewealth of rich families. Provided there are noshocks of the type embodied by ε, the wealthof every family converges to a steady state with

wealth k. Inequality disappears in the long run.(Naturally, these results depend on rather strictassumptions. If we assume that the propensityto save varies in line with income, the process ofequalization does not happen in the long run.)

Under a different scenario, the distributionof income remains unchanged (in its variance)or even widens. Imagine, in line with the mostcommon solution in the literature, that all orsome investments leading to a high-return tech-nology are by nature indivisible, and that any in-dividual needs some k larger than some thresh-old k∗ to acquire that high-return technology.Once more, assume credit markets are missing.4

Such a world may then lead to an unchangedor even to a more polarized income distribu-tion. Those with assets k < k∗ are constrainedto invest in low-return technologies and to re-main in low income brackets. Those with k > k∗

maintain or even amplify their advantage in theincome distribution. Naturally, a few changesin the underlying assumptions can overturn theprediction of higher inequality. For example,those with few assets may save and accumu-late enough wealth to overcome the thresholdk∗. The economy (and therefore average in-come) may grow over time to the point whereall individuals end up with k > k∗ and there-fore can invest in high-return technologies.Finally, technological shocks occurring withsome probability ε may rearrange the positionsof individuals in the initial income distribution.

These models overcome one of the centralweaknesses of factor-share models of incomedistribution: They offer an endogenous theoryof growth and inequality. Inequality, defined asthe distribution of wealth above or below k∗, be-comes now a direct cause of economic stagna-tion. By contrast, more equal income distribu-tions, and particularly distributions that makesure that most individuals have k > k∗, fosterthe process of development.5

4For a summary of the rather mixed evidence on the impact ofcredit constraints on intergenerational mobility, see Piketty(2000), pp. 456–59.5This discussion draws very heavily from Kahhat’s (2007) ex-cellent survey of the literature modeling the persistence of

492 Boix

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

Whereas factor-share models shy away frommodeling the underlying causes of income in-equality, the theories on the intergenerationaltransmission of wealth explicitly attempt toendogenize the evolution of inequality andits consequences for development, that is, foraverage income. They do so mainly to accountfor the growth effects of inequality—and onlysecondarily to address the causes of inequalityper se. However, their effort has two short-comings. First, they do not endogenize theinitial distribution of income. Second, they donot offer a causal explanation of the functionalstructure (production function, distributionof idiosyncratic shocks, types of investment)dictating the generation and transmission ofwealth. For example, almost all intergenera-tional models do not endogenize the presenceof imperfect credit markets. Yet we know thatinstitutional and legal frameworks vary acrosssocieties. Many economies may not have solvedinformational asymmetries that, under condi-tions of high income inequality, lead to furtherunderinvestment and income polarization. Butother economies have. To put it differently,financial-market imperfections (or, moregenerally, the presence of barriers to entry inmarkets) vary because they do not happen in adeus ex machina fashion. They are shaped by spe-cific policy choices—often due to the particulardistributional consequences of those choices.6

Institutional Models

In contrast to economic models of intergen-erational transmission of wealth, where eco-nomic agents simply maximize income withina given market structure, several researchers

inequality. For particular models linking inequality, capitalformation, and growth, see, e.g., Banerjee & Newman (1991,1993), Aghion & Bolton (1997), Piketty (1997), and Galor &Moav (2006).6For two exceptions, see Greenwood & Jovanovich (1990),who discuss the endogenous development of financial insti-tutions, and Galor & Moav (2006), who model an economywith high-income agents who, for complementarity reasons,are interested in financing the formation of skills among low-income individuals and who therefore contribute to the re-duction of variance in the distribution of wealth.

have built a third class of models predicatedon the notion that institutional and politicalchoices matter. These models come in a va-riety of forms. Some stress the role of partiesand policy choices to explain inequality (Hibbs1977, Huber & Stephens 2001, Bartels 2008).An abundant literature on labor markets hasrelated cross-national differences in the struc-ture of wage bargaining with the level of wagecompression—Wallerstein (1999) is likely to bethe best exponent of this line of thought. Fi-nally, the organization of welfare states has beenlinked to different poverty levels and incomedistributions (Esping-Andersen 1990, Goodinet al. 1999, Smeeding 2005).

Although these political-institutional mod-els have made definite contributions to the de-bate on the sources of inequality, they have sev-eral shortcomings. First, their focus is limitedto the evolution of advanced countries since the1970s. Second, government partisanship andwage institutions have been found to have littleeffect on inequality in the long run, that is, onceone looks at pre-1970 data (Scheve & Stasavage2009). Finally, they are incomplete in the fol-lowing sense: They do not integrate their the-ory of inequality determination within a modelthat explains the conditions under which a givendistribution of income is compatible (or not)with the efficient allocation of resources andhence with growth.

ORIGINS OF INEQUALITY

This section inquires into the forces underlyingthe emergence of inequality, summarizing thefull-fledged model and empirics presented byBoix (2010).

Primitive Societies andSelf-Enforcing Peace

To understand what causes the emergence ofinequality, let us start with a world populatedby agents with the same resources (i.e., laborendowments and technologies of production).These agents can be thought of as individualsliving together or as individuals or householdspopulating some territory. They can also be

www.annualreviews.org • Origins and Persistence of Inequality 493

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

thought of as representative individuals of ho-mogeneous groups, that is, groups formed by“identical” agents who have somehow solvedtheir intragroup collective action problems. (Asshown later, the possibility of several individu-als or households living together is based on thesame principles that make it possible for differ-ent groups to live in peace.)

Although those individuals still are, in linewith standard economic assumptions, strict in-come maximizers, they may choose among two(stylized) alternative strategies to advance theirgoal. On the one hand, they may devote theirresources to some direct “productive” activ-ity such as hunting. On the other hand, theymay employ them in predatory activities to grabother individuals’ production. In more complexsocieties, these two strategies should be thoughtof in broader terms. A productive strategy im-plies allocating one’s own time and endowmentsto produce goods and services, which are paidat the rate established in an arena of voluntarytransactions—a market. An extractive or expro-priatory strategy is equivalent to the appropri-ation of the assets or returns of other individu-als, either directly or through the introductionof distortionary policies that change the prices

of their assets. Extractive strategies take manyforms: they may consist in regulating the sup-ply of factors (through labor-market legislation,migration policies, capital controls, trade barri-ers to goods and services, etc.), setting prices atlevels that differ from the ones set from purelyvoluntary transactions, or directly enforcingtransfers from some agents to other individuals.

Assume that stealing is more profitable toone particular individual than producing if allthe other agents are purely engaged in a pro-ductive activity. He may still profit from someof his own production while grabbing, withoutmuch effort or opposition, the output of theother player(s). By contrast, being a producerwho is looted by the rest is the worst possiblescenario for all agents. In between those two ex-treme cases, players prefer a situation in whicheveryone produces to one in which everyoneloots (or attempts to loot) everyone else.

In this game, which has a prisoner’s dilemmastructure, players are identical and their pay-offs are symmetrical. As is well known, if in-dividuals interact only once, generalized loot-ing is the resulting equilibrium. However, ifthe game is played over time (with future pay-offs valued with discount rate δ, such as the

Figure 1An iterated prisoner’s dilemma.

494 Boix

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

ones represented in Figure 1), two equilibriamay emerge. On the one hand, the players maydirectly choose a looting strategy, which wouldthen be self-sustaining over time since everyplayer has an incentive to punish the otherplayer in response to his noncooperative re-sponse. On the other hand, they may follow atrigger strategy according to which they choosea productive response to start with and onlyshift to looting if the other agent loots them.

The latter strategy leads to a sustainableproduction-production equilibrium if the valueof choosing a production strategy is larger thanthat of looting a producer in the first period andthen facing a looting solution ever after. Em-ploying the payoffs in Figure 1, it is an equilib-rium if 3/(1−δ) > 4 + 2(δ/1−δ). Solving this in-equality shows us that peace and production arepossible whenever δ> 1/2, that is, whenever theplayers do not discount the future too heavily.

There are two key lessons in this game. Thefirst one, which is well known in the literature,is that even in a Hobbesian world, where in-dividuals have strong incentives to raid otheragents, it is possible for all players to avoid con-flict, sustain peace, and engage in productiveactivities without having to resort to any cen-tralized mechanisms of authority. Anarchy andpeace are compatible because everyone has anincentive to behave well in response to everyoneelse also behaving well.

The second lesson of the game, which iscentral to this article’s analysis, is that a self-enforcing peace outcome can take place only ifthere is some fundamental equality of condi-tions among players. To see why, let us exam-ine what would happen if individual B derived ahigher payoff from producing (6 instead of 3) asa result of experiencing a positive technologicalshock. Assume moreover that all of the increasewould be absorbed by the looter; that is, if A de-cided to raid B, A’s payoff from looting wouldgrow from, say, 4 to 7. In this new context, Awould follow a peaceful strategy only if 3/(1−δ)> 7 + 2 (δ/1−δ). Given this inequality, A wouldchoose a production strategy only if δ > 4/5,that is, if it values future payoffs very highlyindeed. (By contrast, B’s incentive to follow a

production strategy increases as the productionpayoff goes up. In this particular example, forany δ > 1/5, that is, even for cases in which thefuture is heavily discounted, B prefers peace.)In other words, as inequality increases, the in-centive to cooperate declines for the less ad-vantaged side, and systematic conflict becomesmuch more likely.

The main insights of the game match wellthe functioning characteristics of primitive,preagrarian communities. According to ar-chaeological evidence and anthropologicalresearch, foraging communities are small—thesum of a few families. Their internal level ofdifferentiation is minimal. Individuals exhibitsimilar age-specific patterns of consumption.Equality persists over time owing to the typeof technologies in use (strongly correlated withindividual ingenuity and physical strength) andto the fact that asset accumulation (and henceany intergenerational transmission of wealth)is impossible. Cooperation takes place withoutstable structures of authority or permanentleaders, at least beyond one generation. Bandor tribe chiefs act as mere referees amongdifferent individuals or families, cajoling,persuading, or mediating between their groupmembers (Clastres 1972, 1974). As emphasizedabove, non-institutionalized peace, that is,peace unenforced by any third party, is onlyone of the possible equilibria of the game. Evenunder conditions of equality, a fleeting shift indiscount rates or in the technology of lootingmay result in the collapse of cooperation. Infact, we know that generalized violence is en-demic among preagrarian societies (Gat 2000,LeBlanc 2003). But foraging bands manageinterpersonal conflict through a regular processof fission, with a section of the group secedingand migrating to a new location, where it self-organizes again along the standard patternsof relative equality and non-institutionalizedpolitics (Hirschman 1981).

Technological Shocks, Emergence ofInequality, and Political Violence

The initial equality of conditions, and hencethe possibility of self-enforced peace, breaks

www.annualreviews.org • Origins and Persistence of Inequality 495

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

down as soon as there is some kind of tech-nological shock—primarily the domesticationof plants and animals—that raises the payoffsdifferently across different individuals, house-holds, or groups.

The immediate sources of that technologicalshock, that is, of the agrarian revolution, mayhave been manifold and should not concern ushere. It may have been driven by particular en-vironmental changes; it may have been endoge-nous to population growth, which may haveresulted in a downward shift in the marginalproductivity of hunting and gathering; it mayhave been the outcome of a slow process oflearning-by-doing; or it may have been randomand unexpected (Watson 1995, Bar-Yosef &Meadow 1995). Nonetheless, as emphasized byDiamond (1997) and tested by Hibbs & Olsson(2004), independently of all these causes, theadoption of agriculture only occurred in thoseparticular regions of Earth that met some par-ticular biological and geographical conditions,such as an appropriate climate and latitude.

The agrarian revolution triggered a momen-tous transition. Formerly, all men were engagedin foraging activities, and the marginal produc-tivity of hunting was the same across the board.The new technologies of storage and plant do-mestication led to a high degree of territorialvariation in terms of land fertility and produc-tivity. In other words, the introduction of newtechnologies caused an initially homogeneousworld to acquire a differentiated resource gra-dient with a core of rich lands and a peripheryof marginal quality. Those who happened tooccupy the core now obtained a much largerincome than the rest.

As already discussed, the technology-biasedshock and the correlated increase in inequal-ity has a discriminating effect on the behaviorof individuals. Whereas the agent who benefitsfrom the shock still has an incentive to followa production strategy, the agent who does nothas a much higher propensity to loot. In otherwords, as inequality increases, violence be-comes more likely both within established com-munities (those sharing a given territory) andacross territorially differentiated communities.

Moreover, violence takes place in a particulardirection: from the disadvantaged to the advan-taged. (Naturally, violence does not only hap-pen for those motives and in that direction. Forexample, the “natural producers” may have anincentive to respond in kind to protect them-selves.) This violence may occur either withinhuman groups that share a compact territory,or between groups. The latter case matches arather strong regularity in human history: thefact that peripheral, less rich lands have tendedto breed much more warprone societies thanfertile lands. Ancient Mesopotamia and Egypt,the first centers of agriculture in the MiddleEast, withstood successive waves of invadersoriginating in the Asian steppes. The RomanEmpire fell under the pressure of Germanictribes. Medieval Europe endured the attacksof periphery populations such as the Vikings,Slavs, Turks, and Mongols.

State Formation and Political Sourcesof Inequality

The rise of economic inequality and violenceinduced, in turn, a key transformation in theexisting political structure, which then affectedthe distribution of wealth and income.

Institutional solutions to the problem ofviolence. To sustain peace and the possibil-ity of a fully productive strategy (where jointgains are certainly maximized), individuals havetwo alternatives. Both of them involve the for-mation of some organization or structure thathas the monopoly of violence over a giventerritory—that is, they involve the constructionof a “state.” But the institutional mechanismsthat underpin each solution are very different.

On the one hand, the producers—a termthat I use now to define those who benefit themost from the technological shock—can trans-fer part of their production to the potential loot-ers or bandits (those who do not benefit fromthe shock), in exchange for which the latter re-frain from violence and supply some protectionagainst other bandits. (Bandits and looters neednot be external enemies. These terms apply as

496 Boix

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

well to residents of the community who may tryto steal from, free-ride on, or otherwise take ad-vantage of richer individuals. This first path topeace, which we may refer to as a “monarchicalsolution,” follows from Olson’s (1993, 2000) in-sight that the state emerges as “roving bandits”turn into “stationary bandits” or landlords.7

On the other hand, the producers may sim-ply spend some part of their own resources to setup institutions or structures, such as a commonarmy and stable leadership, to defend their as-sets and lives against bandits. This second pathto state formation, where producers also dou-ble as defenders and rulers, implies, in a word,the construction of a “self-governing” polity.It is important to stress here that a system inwhich the community of producers sets asidesome portion of its income to hire an army orleader to protect its members (while governingthemselves through “republican” institutions)does not constitute a case of self-governmentor republicanism. Since the hired leader can-not credibly commit to preserve the terms ofthe contract once he has been appointed andtakes control of the army, hiring a protector orcondottiero ends in (is equivalent to) the subjec-tion of the producers to a leader-monarch. Inother words, a stable republican solution is onein which citizens have ultimate control over themeans of defense.

This self-governing strategy, based onspending some resources to deter looters,comes in two variants. On the one hand, theproducers may mainly focus on production andthen spend a minimal portion θD of their in-come for purely defensive purposes, such asbuilding a wall or a watch tower to observethe horizon. On the other hand, the pro-ducers may decide to increase the amount ofresources they spend on military endeavors,

7Writing from an anthropological perspective, Carneiro(1970) also offered a theory of state formation that empha-sized the interaction of violence and exit options (the latterderiving from population density). Exit strategies and densitywere, however, not endogenized in his paper. Like Olson, hefocused on the emergence of autocratic states. See Wright(1977) for a review of state-formation theories in anthropol-ogy and archaeology.

beyond what is needed to defend their terri-tory, to attack the looters. The offensive strat-egy costs an extra portion θO of their incomeand puts overall spending at θ = θO+ θD.8 If theoffensive costs θO are high, the producers will bemore likely to follow a strictly defensive strat-egy. However, as they converge to 0, the valueof looting outsiders becomes higher than thecosts of attacking, and the self-governing com-munity becomes an imperial republic in bothits nature and its behavior.

The distributive effects of different politicalregimes. Each political solution comes withvery different distributional consequences. Letus work out the main logic of the argument us-ing the payoff structure discussed in the sub-section “Primitive Societies and Self-EnforcingPeace.” Before the state was established, if peacewas an equilibrium, each individual receiveda production payoff of 3 and the distributionof income was equal. Now, after a technolog-ically biased shock takes place, the distribu-tion of income becomes unequal: IndividualsA make twice as much as individuals B. How-ever, this new income distribution is not thefinal one because it triggers a set of politicalchanges that reshape the material position ofindividuals. In a monarchical solution, predi-cated on extracting a portion ε of the assets orincome of producer A and siphoning it to thepotential bandit B such that the latter prefersnot to pillage A (and protects A), each naturalproducer keeps (1−ε)6, and the bandit-turned-ruler receives 3+6εN where N is the number ofstrict producers.9 By contrast, in a republican or

8It seems plausible that even when the main endeavor of theseagents is production, they still need to pay some minimumcosts of defense to be ready to move to the fully militarizedstrategy. If the latter were equal to 0, then government in-stitutions would appear only when bandits actually appearedon the horizon. When they were not there, everyone wouldsimply cooperate spontaneously with no permanent structuregoverning them. This does not seem realistic. For a formalproof of the condition under which this solution prevails, seeBoix (2010).9The parameter ε should be such that both the ruler and theproducers are better off than in a looting-looting equilibrium.Moreover, the monarchical settlement is only feasible if peace

www.annualreviews.org • Origins and Persistence of Inequality 497

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

Table 1 A simulation of the effects of monarchy

Percentindividuals

Incomeafter shock

Income minus/plus extraction

Percent oftotal income of

Ratioincome A

A B A B Extraction A B All A All B over B Gini indexBaseline 90 10 6 3 0 6 3 95 5 0.5 0.05Monarchy

Low extr. 90 10 6 3 1.5 4.5 16.5 71 19 3.7 0.18High extr. 90 10 6 3 3 3 30 47 53 10 0.42

self-governing polity, every citizen keeps 6−θ

and inequality does not vary (within the repub-lican community) with respect to the incomedistribution resulting from the technologicalshock.

Table 1 simulates the effects of establish-ing a monarchical world once the initial, equalworld experiences a technological shock. Indi-viduals of type A constitute 90% of the pop-ulation and double their output from 3 to 6.For a given discount rate δ = 0.5, the shockmakes the spontaneous peace equilibrium un-feasible. Assume that individuals of type B havea better military technology than type A suchthat they prevail and impose a monarchy. Giventhe structure of the model, the extraction rateε must range between 0.25 and 0.50 for themonarchical settlement to be an equilibriumfor both parties. I use this range to calculatethe postextraction per capita income of indi-viduals A and B, the proportion of total in-come in the hands of the top 10%, and the Giniindex.

Table 1 shows that monarchies have astrong effect on the distribution of income. Theratio between rulers and producers goes from0.5 after the shock but before the establishmentof a state to 3.7 under the lowest extraction rate(ε = 0.25) and 10 under the highest extractionrate (ε = 0.50).10 The Gini index goes up from

without transfers is not an equilibrium. Again, see Boix (2010)for an extended and formal discussion.10Given that the bandit-lord is unable to commit to a lowerextraction rate than the one that maximizes his income, theextraction rate will vary as a function of two factors. First,it will increase with the “administrative” efficiency of the

0.05 after the shock (yet before the introductionof a state) to 0.18 if the extraction rate is low and0.42 if the latter is high. Although not shownhere, holding the feasible rate of extraction con-stant, the smaller the number of rulers relativeto the population of the country, the more un-equal the distribution becomes. For example,halving the ruling group to just 5% of the to-tal population doubles the rulers/producers in-come ratio.

The distribution of income will also beshaped by the way in which rulers extract theirresources or, in other words, by the kinds ofalliances they form with different strata of pro-ducers. The distribution of income becomeshighly skewed when the rulers ally with a par-ticular fraction of the population (imposing alower extraction rate ε on them than on othergroups) or when the ruling elite simply re-places the existing segment of wealthy indi-viduals. By contrast, the ruling elite may favora more equalized structure of income withinthe producers’ population. These different out-comes, resulting from different institutionaland distributive choices, are particularly mani-fest in processes of repopulation and coloniza-tion. After the Christian kingdoms of northernSpain took over most of the southern half ofthe Iberian Peninsula in the thirteenth century,they differed markedly in their strategies of land

state. Second, it will vary with the tax elasticity of output andwith asset specificity (Boix 2003): The higher the mobilityof assets, the less punitive the level of extraction by the lord.Hence, predation and social and economic inequalities willbe sharper in agrarian economies than in urban settings (foridentical distributions of technologies across individuals).

498 Boix

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

redistribution: Castile and Portugal favored theformation of large landholdings; Catalans set-tled in Valencia as small farmers (Vicens Vives1957). Similarly, the conquest of the Americasfollowed very divergent paths: Farming com-munities settled in the northeastern seaboardand in Quebec; the Caribbean basin and Cen-tral and South America were dominated bylandowners (Engerman & Sokoloff 2002). Inall those cases, the distribution of land partlyresponded to economic considerations, such asthe supply of labor. But it was strongly affectedby the way in which the ruling elites decidedto structure the distribution of assets—and thatwas probably related to the prevailing politicalinstitutions and societal conditions back in theconquering countries.

Under a nonmonarchical system, finalequality within the producing, self-governingcommunity will closely mirror the technolog-ically induced inequality (provided taxation topay for defense is not progressive): The overalldistribution, including the looters, would ac-tually be lower because the producers have tospend some of their resources on deterrence.

Although a monarchical solution has a muchstronger effect on the distribution of incomeresulting from the technological shock thanthe republican solution, this does not meanthat monarchical regimes should always havemore skewed income distributions than repub-lican governments. Consider a case in whichthe technologically induced distribution is quiteunequal but still within a range that makescooperation among individuals a feasible out-come. If the extraction rate under a monar-chical regime is sufficiently low, then it maywell be that a republican regime is more un-equal than a monarchical structure, particu-larly given that in a monarchical solution re-sources are transferred from technology-rich totechnology-poor individuals.

Some empirical evidence. From an empiri-cal point of view, the simulated extraction ratesand Gini indexes do not seem farfetched. Con-sider the evidence gathered by Milanovic et al.(2007) and reproduced in Table 2. In almost all

Table 2 Some inequality statistics

Society Gini index Top 10%Roman Empire ca. 14 ADa 0.36 0.44Byzantium 1000a 0.41 0.46England/Wales 1688a 0.45 0.38Moghul India 1750a 0.39 0.35Mexico 1790a 0.64 0.61Naples 1811a 0.28 0.35Brazil 1872a 0.39 –France 1780b – 0.56

Sources: aMilanovic et al. (2007); bMorrisson 2000.

preindustrial regimes, the Gini index fluctuatesaround 0.40, and in most cases the top 10% ofthe distribution controls more than two fifths oftotal income. Bourguignon & Morrisson (2002)offer similar data across the world in 1820, atthe dawn of the industrial revolution.

Owing to the lack of reliable data on incomedistribution before the nineteenth century, andgiven that height and individual income seemto be correlated (at least within ethnically sim-ilar groups) (Steckel 1995), Boix & Rosenbluth(2006) employ height as a proxy for access tofood resources, health status, and general wel-fare. In preindustrial economies, insufficientnutrition was widespread, mostly affecting thepoorest social strata. Whereas the top decileconsumed 4329 kcal per day in England in 1790,the bottom decile consumed 1545 kcal per day(Fogel 1994). Hence the distribution of heightsshould provide us with clues about how thatsociety distributed its resources, at least in hu-man communities living under conditions ofscarcity.

As summarized in Table 3, inequality islow within tribal, preagrarian groups. Withthe transition to a fully agricultural societywith complex political institutions, heights varystrongly with the type of political regime. In an-cient kingdoms and absolute monarchies, thereis considerable differentiation among individu-als: The height gap between nobles and peas-ants ranges from 7 to 9 cm. Using fitted mod-els between log income (in constant dollars of1985) and height (Steckel 1995), this differencetranslates into an income ratio of 5 or 6 to

www.annualreviews.org • Origins and Persistence of Inequality 499

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

Table 3 Political settings and differences in male heights

Institutions and places Mean (in cm) Difference (in cm)Pre-State SocietyZuni Pueblos (before 1680 AD)

Height at 90th centile 165.8 4.7

Height at 50th centile 161.1Monarchical/Authoritarian SocietiesAncient Egypt (New Kingdom)

Royal 174 8.0

Commoners 166Mycenae

Royal 172.5 6.4

Commoners 166.1Poland—17th century

Noble 170.6 7.7

Rural non-Jewish 165.2

Jewish 162.9Democratic SocietyMembers of Ohio National Guard, mid-nineteenth century

Laborers 173.3 2.2

Skilled workers 174.0

Farmers 174.7

Professionals 175.5

Source: Boix & Rosenbluth (2006).

1 between the two groups. By contrast,more democratic settings, such as preindustrialOhio, show a much more compressed heightstructure.

Endogenizing State Formation

Although a detailed discussion of the factorsthat ultimately determine the choice of polit-ical regime is beyond the scope of this arti-cle, which regime will prevail is related to thecosts it imposes on producers. Producers willopt for a self-governing community if ε > θ.Otherwise—that is, if the costs of defense arehigher than the taxes they pay—the commu-nity of producers will end up subject to thebandit-lord.

Self-governing institutions will be morelikely when the exercise of violence and wardoes not rely on sophisticated weaponry (suchas swords, chariots, horses and so on), which

promotes specialization and leads to the forma-tion of a separate caste of warriors, and whenproducers live in territories with geographicalconditions (such as sea barriers and mountain-ous terrains) that make it easier for producersto deter bandits.

As weapons become more complex and ex-pensive and as organization (in large armies) be-comes central to military action, the productionof violence will turn into a rather specializedactivity, giving an important advantage to thoseindividuals who decide to engage in predatoryactivities. Strong and centralized states, highlyextractive fiscal systems, and rather unequal so-cieties will then prevail.

The foundation of the first big states inMesopotamia around 3500 BC was proba-bly related to the introduction of bronzeweapons. The introduction of two-wheeledchariots around 1800 BC increased the costsof war and led to the formation of a set of “feu-dal” states with a narrow aristocracy of war-riors exercising full control over their subjects(McNeill 1982). As emphasized by Rogowski &McRae (2008), medieval feudalism followed theintroduction of the stirrup and the strengthen-ing of the cavalry, and absolutism was equallyrelated to the sixteenth-century revolution as-sociated with the use of cannons and standingarmies.

Military technology has not always evolvedto trigger more centralized and hierarchicalstructures. Occasionally, there have been or-ganizational and mechanical changes that havereduced existing imbalances between warriorsand producers. Such changes have resulted in abroadening of the political base of government(to the point of “democratizing” institutions)and the re-equalization of the distributionof income. In the second half of the secondmillennium before Christ, the discovery of howto make weapons out of iron cheapened war,allowed a “relatively large proportion of themale population [to] acquire metal arms andarmor” (McNeill 1982, p. 12), and gave way tomore egalitarian political and social structures.Aristotle wrote that “when the country isadapted for cavalry, then a strong oligarchy

500 Boix

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

is likely to be established,” whereas “thelight-armed and the naval element are whollydemocratic” (Politics, Book 6, Part 7). Levi(1997) emphasizes the impact of modernwar and general military mobilization on theextension of political rights, and several authorshave linked the occurrence of modern war tohigher levels of taxation and redistribution(Peacock & Wiseman 1961).

PERSISTENCE OF INEQUALITY

After the formation of the state, the overall dis-tribution of income continues to be a functionof both technological shocks (changing the sup-ply and demand of factors and their ownershipand prices) and military and political factors(such as the domestic balance of power, the in-ternal organization of the state, and the natureof interstate competition).

Nonetheless, there is a key difference be-tween stateless communities and societies gov-erned by states. In the latter, technologicalshocks affect the distribution of income embed-ded in a stable, institutionalized political struc-ture. That is, they are mediated by politicalprocedures according to which a relatively sta-ble set of individuals systematically vote overpolicies and have enough authority to imple-ment them. In primitive or prepolitical soci-eties, technological change happens in an un-constrained manner—and often ends up wipedout by conflict unless some institutions are con-structed to protect its application and gains. Ina political society, the decisive voter (a dicta-tor, a narrow ruling elite, or the electorate ofa democracy) chooses the particular policy orregulatory framework that determines (even ifpartially) access to new technologies and, there-fore, the productivity and income of all eco-nomic agents.

The choice of a particular regulatory or pol-icy framework is equivalent to the imposition(or elimination) of entry barriers to the tech-nologies of production. The concept of pol-icy regulation should be understood here in abroad sense: It may imply the introduction (orabolition) of high tariffs, the passage of partic-

ular laws that bar the use of new transporta-tion technologies, the monopolistic (or com-petitive) regulation of a given market, and so on.As stressed above, ancien regime states tended toimpose those barriers in a discriminatory fash-ion precisely to glue together the social and po-litical coalitions that sustained them.

In a politically institutionalized society, thedecisive voter (what standard spatial modelsidentify as the “median voter”) compares herincome under different technological alterna-tives and chooses the regulatory frameworkthat maximizes her income. She will seek toblock policy reforms (such as the suppressionof guilds, the abolition of capital controls, ortrade liberalization) that would reduce her fi-nal disposable income and promote those thatwould increase it.

Notice that the choice of the policy param-eter will not be simply based on the “direct” ef-fect it has on the income of the decisive voter (orvoters). A given regulatory framework shapesthe pretax income of all individuals; that is,it generates a particular income distribution.If we accept existing optimal taxation models(Meltzer & Richards 1981) and their applica-tion to the choice of political regimes (Boix2003, Boix 2008), each income distribution willbe associated with a tax rate, and, given somecosts of excluding part of the population fromparticipating, with a particular political regime.Hence, in choosing a particular policy frame-work, the decisive voter calculates the changein her final income, taking into account notonly the direct economic effect (a change in theproduction function) but also the taxes she willhave to pay (under each political regime) andthe costs of excluding part of the population (ifany) to retain her position as a decisive voter.

In the following two subsections, I exam-ine the choices of the decisive voter. I employ asimple model in which, in the spirit of policy-reform models such as Fernandez & Rodrik’s(1991), there are two types of individuals, h andl, with their respective incomes yh > yl , andthe l individuals constitute a majority. In thefirst subsection (“Political Reform and Inequal-ity in Authoritarian Regimes”), I consider the

www.annualreviews.org • Origins and Persistence of Inequality 501

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

reaction of high-income individuals to biasedtechnological shocks when they are the decisivevoters; because they are a minority, such a casecorresponds to an authoritarian regime. In thenext subsection (“Political Reform and Inequal-ity in a Fully Enfranchised Polity”), I explorethe responses to similar shocks when the deci-sive voters are the low-income voters. I then addsome complexity to this exploration (“Amend-ments”) by considering three extensions: policymaking under uncertainty, the choice of a policyframework in the context of multiple countries,and the effect of relative gains.

One of the findings of the following dis-cussion is that underdevelopment is caused bythe reluctance of preindustrial political elitesto introduce any policy reforms. Similar resultshave been derived by Olson (1982), Justman &Gradstein (1999), Krusell & Rıos-Rull (1996),and Parente & Prescott (1999), who insist thatthe loss of economic rents leads elites to blockreforms, and Acemoglu & Robinson (2001),who have added elites’ concern over the lossof political power to explain the failure of eco-nomic liberalization. The approach I suggesthere allows us to explain also why inequality ispart and parcel of that suboptimal outcome ofauthoritarianism and underdevelopment, teaseout the political and economic sources of stag-nation, and explore the particular political en-vironment that facilitated the industrial break-through. Moreover, it offers a way to treat ina unified structure both the instances of policyblockage inherent to many ancien regime elitesand the opposition to change in many demo-cratic societies.

Political Reform and Inequality inAuthoritarian Regimes

Political reform and inequality under po-tentially falling incomes. Consider first a casein which only high-income voters decide. Toexclude the rest of the population, they need topay some exclusionary or repression cost c—butsince this cost is lower than the tax they wouldpay if everybody voted, they impose an author-itarian regime. (The cost c is lower because the

level of inequality is high enough to lead tohigher taxes under democracy. Throughout thediscussion I assume c increases with both thesize and the income of the excluded popula-tion. Poor populations are easier to repress thepoorer they are. For a formal analysis, see Boix2003.)

The decisive voter, a high-income individ-ual, will block any reform that leads to a fallof his pretax income—unless the regulatorychange leads to a distribution and type of wealthsuch that a tax under democracy is lower thanthe costs of excluding low-income voters un-der the status quo. An example of that poten-tial shock would be the introduction of newcommercial practices or inventions in a strictlyagrarian economy that reduce the value of ex-isting assets (such as land) or that divert labor tonew sectors (and therefore lower labor supply).

A very stylized representation of the generalcase of maintaining the status quo appears inFigure 2a. If high-income individuals pass thereform, they suffer an income loss. This lossis not compensated by a possible change to abroader franchise (as a result of more equal con-ditions and less threatening taxes) that wouldreduce c to 0. Hence, no change is implemented.

As I point out later, this instance of blockedreform was pervasive among ancien regime sys-tems. In preindustrialized societies, wealthy in-dividuals owned very specific assets, that is, as-sets whose value hinged on regulatory barrierssustained by the state and/or wealth (such asland) that could not be easily redeployed in new,more productive economic sectors. In fact, theirposition on the social ladder derived preciselyfrom the types of highly extractive institutionsthat accompanied the formation of the state.Because inequality and authoritarianism camehand in hand, any economic change threatenedthe political basis of the wealth of the rulingelite. Hence, those old regimes were logicallyassociated with long-run stagnation.

Political reform and inequality under po-tentially increasing incomes. By contrast,the decisive voter may support a reform oraccept a shock if the change in regulation or

502 Boix

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

0

100

200

300

400

500

600

700

800

900

1000

1100

1200(a)

1

Population

Inco

me

SQ income SQ income - C After Shock Income

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

0

100

200

300

400

500

600

700

800

900

1000

1100

1200(b)

1

Population

Inco

me

SQ Income SQ Income - C After Shock Income After Shock Income - C

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Figure 2(a) Technological shock and falling incomes among high-income voters. (b) Rich-biased technological shock.

the economic transformation boosts his pretaxincome. Still, his final decision will depend onthe overall effects of the potential reform on thewhole population.

Consider first the case in which incomegrows only among high-income voters. Giventhat the exclusion costs c do not change (by ourassumption that they are determined by the in-come level of l individuals), the decisive voteraccepts the policy change while maintainingthe exclusionary political system. The discov-ery of natural resources (such as oil) or com-modity price booms fit this case: They increaseinequality and reinforce authoritarianism, pro-

vided that those economic resources are ownedby those in control of (or in alliance with) thestate. [On the general effect of oil on politicalregimes, see Ross (2001) and Boix (2003); on theimpact of price cycles, see Friedman (2006).]

Let us examine now the case of a technologi-cal or policy shock that increases incomes acrossthe board. In those instances, the exclusion costsshift upward (given that low-income individu-als have more resources available to them tooppose their exclusion from the ballot booth).Assume that exclusion costs go up linearlywith the poor’s income, perhaps above somesubsistence wage under which low-income

www.annualreviews.org • Origins and Persistence of Inequality 503

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

voters have no strength to oppose exclusion-ary measures. Then, under any income increasethat does not change the overall level of inequal-ity (that is, whenever both types get the sameraise in relative terms), the decisive voter willapprove the reform and maintain the politicalstatus quo of restrictive franchise, since he con-tinues to have enough extra resources to containthe disenfranchised.

However, as soon as the income shock be-comes “poor-biased,” that is, as soon as it in-creases incomes among low-income individualsproportionally more than among high-incomevoters, income inequality declines, lowering thecost of taxes (under full enfranchisement) rel-ative to the cost of repression (while, as be-fore, exclusion costs rise). If the poor-biasedreform or shock is small, the decisive voterwill decide to block the reform because, al-though it would lead to higher exclusionarycosts, which would push high-income voters toprefer democracy (after the economic reformhad been passed), the after-tax after-transfer in-come in that democratic setting would proba-bly be lower than the net income under an au-thoritarian regime before the reform was passed(since inequality would still remain high andwould result in considerable taxes). If the poor-biased shock is, instead, sufficiently strong, thepostreform or postshock level of inequality willbecome moderate or low. That decline in in-equality (jointly with rising exclusionary costs)will lead high-income voters to prefer payingtaxes under democracy than funding an au-thoritarian police. However, that type of highlypoor-biased shock—leading to growth, equal-ity, and democracy—will be highly unlikely inclosed, authoritarian political economies, par-ticularly if the initial level of inequality is veryhigh. Given these considerations, it seems rea-sonable that change could only come throughsome revolutionary shock, some miscalculatedpolicy move by the governing elites, or a reac-tion to some external threat.

To summarize the discussion, permanentgrowth and considerable reductions in inequal-ity, followed by peaceful transitions to democ-racy, are extremely elusive in ancien regime

political systems. Because authoritarianregimes come into place by imposing unequaldistributions of wealth and strong entry barri-ers to favor the ruling coalition and its allies,they are remarkably robust political equilibria.Statistically they have been: Democratic or re-publican polities account for less than 1% of allthe states that emerged after the domesticationof plants and animals, and it took more than10,000 years to start the industrial revolutionand to get democracy and relative levels ofequality in a few parts of the world.

Political Reform and Inequality in aFully Enfranchised Polity

In a fully enfranchised polity—one where low-income individual l is the decisive voter—theinterests of the decisive voter are very often amirror image of the interests of a high-incomevoter (h) in the authoritarian setting we justdiscussed.

Increasing incomes for low-income individ-uals. Any poor-biased shock or reform, i.e., anyshock or reform that increases l ’s income morethan h’s income (Figure 3a), will receive thesupport of the decisive voter—unless the re-form leads to a fall in taxes and transfers (result-ing from shrinking inequality) that offsets thepositive direct impact of the reform on his in-come.11 A “rich-biased” shock, one that booststhe income of high-income individuals morethan poor, should not necessarily be opposed byl: Low-income individuals may still experiencedirect income growth (and some gains in after-transfer income through the tax-and-transfersystem). However, they will block a rich-biasedshock if the latter generates enough inequalityto turn authoritarianism into a feasible optionfor high-income voters.

Declining incomes for low-income individ-uals. Negative shocks, such as those “skill-biased” technological shocks experienced in in-dustrial economies in the past 30 years, will be

11This case (a decline in transfers bigger than an increase indirect income) becomes more likely as high incomes fall inabsolute terms.

504 Boix

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

0

100

200

300

400

500

600

700

800

900

1000

1100

1200(a)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Population

Inco

me

SQ Income After Shock Income

0

100

200

300

400

500

600

700

800

900

1000

1100

1200(b)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Population

Inco

me

SQ Income After Shock Income

Figure 3(a) Poor-biased technological shock. (b) Technological shock and falling income among low-incomeindividuals.

blocked by voters unless the direct fall in in-come is (automatically) compensated by highertransfers (Figure 3b). Since the introduction offull or quasi-full enfranchisement in the past100 years, unskilled and semiskilled voters havedelayed or impeded shocks and reforms thatwould hurt their welfare at several critical junc-tures. Examples include restrictive immigra-tion policies and probably trade protectionismamong Western nations in the 1920s and 1930s,the emergence of populism and ISI (import-substitution industrialization) in Latin Amer-ica, and the maintenance or implementation ofstrong labor regulations (leading to the forma-

tion of an insider-outsider split in labor mar-kets) in many European countries since the1970s. Antiliberal or anticapitalist revolts werecommon for example among guilds and work-ers in city-states of past ages.

The only way to pass proreform, liberaliz-ing strategies under those circumstances (lossesborne by the decisive voter) entails establishingsome transfer scheme or compensatory packagefrom the winners to the losers. This seems to ac-count for the relatively well-established corre-lation between trade openness and larger publicsectors (Katzenstein 1985, Rodrik 1998, Adsera& Boix 2002).

www.annualreviews.org • Origins and Persistence of Inequality 505

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

The likely time inconsistency of transferschemes may explain why compensatory pol-icy, and hence any reform that delivers or keepslow levels of inequality, happens mostly un-der institutions that strengthen the represen-tation of affected voters, such as encompass-ing unions or proportional-representation rules(Rogowski & Kayser 2002). However, con-trary to a widespread literature, not all com-pensation packages work optimally—only thosethat change the marginal productivity of low-income voters do (Boix 1998).

Amendments

Uncertainty. So far we have assumed a setupin which individuals have perfect informationabout the consequences (in terms of direct in-come, political regime, taxation, and transfers)of any alternative regulatory structure.12 Oncewe introduce the more realistic idea of an uncer-tain environment, the status quo becomes evenmore likely to prevail over time (Fernandez &Rodrik 1991).

To see the stabilizing nature of imperfect in-formation, consider first the case in which, al-though the decisive voter stands to gain fromthe reform, aggregate losses are larger than ag-gregate gains.13 If the decisive voter knows onlyabout aggregate results (not about her individ-ual outcome), in expectation she has losses, andtherefore she blocks that reform. Examine nowthe reverse case—with total gains larger thantotal losses but with net individual losses forthe decisive voter. Having gains in expectation,the decisive voter passes the reform. But af-ter the reform is implemented and the deci-sive voter realizes her actual losses, she will re-peal the reform—provided she still retains herpivotal position. The asymmetry between ex

12No consideration has been given to growth either, or rather,growth is already subsumed in the model with individualshaving zero discount rates—they only care about total incomein their lifespan.13Or, following Fernandez & Rodrik (1991) more strictly,aggregate losses are larger than aggregate gains for the eco-nomic sector (and not necessarily the whole economy) towhich the decisive voter belongs.

ante expected losses (leading to no reform) andex post real losses (leading to reform reversal)reinforces the stability of the status quo.

Interstate competition. Let us now abandonour Robinson Crusoe model of politics in whichour decisive voter chooses a regulatory frame-work for himself and his Friday without anyconcern for the decisions, growth rates, andincome distributions of foreign powers. (Ofcourse, we can also call it the Friday model ofpolitics, where Friday chooses policy for him-self and his master.) Consider instead what hap-pens when our decisive voter has to make deci-sions surrounded by other countries.

Interstate competition spurs technologicalchange. Assume that the decisive voter holdsthe belief that the probability of other countriessuccessfully attacking his country (or simply in-tervening in its domestic affairs) increases withthe ratio of total income of the foreign countryover total domestic income. To preempt for-eign attacks, countries may push for reformseven when they hurt the decisive voter (pro-vided they deliver a larger total income). Thatthey actually do push for such reforms, how-ever, is not very likely. In a context of uncer-tainty about how those shocks or reforms maywork, no state is likely to move unilaterally intouncharted waters.14 Still, the probability of re-forms (and therefore of shifts in the distributionof wealth) will be higher the more fragmentedthe state system is. Hence, it was by virtue ofits fragmentation that Europe had a competi-tive edge over China in the modern era ( Jones1981).

Provided no economy introduces a growth-enhancing policy reform, interstate competi-tion probably does little to alter the status quo.However, as soon as one country takes thetechnological, economic, and therefore militarylead over the rest, it plays a key role in the

14This uncertainty about the future may explain the war biasof ancien regime states: They prefer to attack other states(something they know well how to do, since it is at the basisof the formation of an authoritarian polity) than to invest ingenerating autonomous sources of growth.

506 Boix

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

process of economic transformation of every-one. When England’s take-off became apparentafter the 1830s, other nations scrambled to im-plement the liberal blueprint across Europe. Asthey did, they faced considerable internal con-flict. The conservative-liberal divide that spreadacross Europe (like the one in the Middle Easttoday) closely reflected a split between pro–status quo and proreform policy makers. But,overall, reform happened and in fact acceler-ated in response to a growing performance gap.Direct military threats or defeats made the in-come gap even more glaring and pushed somestates to embrace reform quite abruptly, e.g.,Meiji Japan. Countries that were unable or un-willing to respond either disappeared, engulfedby ever-growing colonial empires, or simplybecame marginal islands in the internationaleconomy.

Envy. So far, we have assumed the decisivevoter chooses according to a strict principleof (after-tax, after-transfer) income maximiza-tion. But voters may decide with an interestin preserving (and increasing) their positionvis-a-vis other individuals. The introductionof envy favors the status quo in polities dom-inated by high-income voters. They will op-pose any shock that is not strictly rich-biased,that is, any policy that does not preserve or in-crease the existing level of inequality. Similarly,low-income voters will fight any shock that isnot strictly poor-biased—even those that wouldlead to growing incomes in absolute terms.Hence, political economies (such as caste soci-eties) or professions (such as academics?) mostlybased on status should be particularly resistantto change.15

Contemporary Historical Trendsin Inequality

Employing the analytical tools of the previoussubsections, let us now impose some analytical

15Fehr et al. (2008) offer strong experimental evidence ofenvy in caste settings.

structure to the evolution of income inequal-ity, mostly in the past two centuries. (I focus onthis period for two reasons. First, we have rel-atively reliable data for at least a few countries.Second, the level of income inequality has gonethrough substantial changes since 1800 relativeto previous periods.)

In the long run, prosperity has bred equal-ity. Figure 4 presents cross-sectional dataof per capita incomes and Gini coefficientsacross the world after 1950. At low levels ofdevelopment there is considerable variancein the overall distribution of income. At highlevels of development variance is much smaller.Figures 5a and 5b show, in turn, the historicalevolution of the income share of the top decilein several English-speaking and continentalEuropean countries for the past two to threecenturies. Those countries with considerableinequality at an initial stage have undergone asubstantial process of equalization over time.Figure 6 reports data on the wealth share ofthe top 1% in Switzerland, the United States,and the United Kingdom.

Industrial (and even commercial) capi-talism, and therefore the breakdown of theagrarian economy, came hand in hand with ashift in the overall institutional or regulatorystructure—from the statism and mercantilismof the ancien regime to a liberal economicorder. Because the triggering event of moderneconomic development, the English industrialrevolution, was a unique event (it was thefirst industrialization process, unconditionedby any previous industrial experiment), anyexplanatory theory may never be able to pass astandard test of significance—against the nullhypothesis of pure chance. But although thecauses of the industrial transformation are stillheavily debated, two key factors, discussed inthe previous subsections, contributed to it:the fragmented nature of the European statesystem, which fostered among policy makers asearch for a successful strategy to grow fasterthan other powers; and, in that general context,the revolutionary break of 1688 in England thatdislodged the existing reactionary or blockingcoalition from power. Those two factors are

www.annualreviews.org • Origins and Persistence of Inequality 507

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

20

30

40

50

60

70

0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000

Per Capita Income in Dollars of 1985

Gin

i In

dex

Figure 4Development and inequality.

explored in Pincus’s (2009) pathbreaking studyof the Glorious Revolution.

Inequality went up with the first phase ofindustrialization in Britain. According to themodel sketched above, this should have actuallyhelped to foster the process of economic trans-formation from a political point of view: theruling elites were interested in those changesbecause they benefited them directly. To putit differently, this model turns the current ex-planation of the Kuznets curve on its head.In the standard account of the Kuznets con-jecture, inequality resulted from industrializa-tion. Instead, one may say that the emergenceof industrial capitalism was politically possibleonly because it came with more inequality atthe beginning. The relative diversification intocommercial interests (i.e., the nonspecificity) ofthe old landowning class’s wealth also facilitatedthe transition to a liberal economic order. [Forevidence on the diversity of economic invest-ments of British landowners and how it helpedto repeal the Corn Laws, see Schonhardt-Bailey(1991). For some confirmation of the Kuznetsconjecture in Britain, see Lindert (2000). Forpartial confirmation in a cross-section of coun-tries, see Barro (2008).]

Over time the new technologies of produc-tion became more widespread, either becausethey were actively promoted by capital ownersinterested in increasing the availability of com-plementary skills (Lindert 2004, pp. 87–122;Galor & Moav 2006) or because industrialistshad no desire or capacity to block the access oflow-income voters to those new technologies.In the context of the model, industrialists wereeither not the decisive voter or they were buttheir after-transfer income improved even afteraccepting the shock in place.

The spread of these new technologies ofproduction had two related consequences: In-comes grew across the board and thereforestrengthened the political capability of middle-income and low-income groups; and incomecompression increased (by the last third of thenineteenth century and at a faster pace since1910), making democracy a more attractive po-litical option than a narrow franchise (Boix2003; see also Przeworski 2009 for an empir-ical confirmation of the same hypothesis, link-ing equalizing conditions and higher chances ofdemocratization). The successive expansions ofthe franchise in Britain took place in lockstepwith the evolution of the economy (Justman &

508 Boix

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

0.25

0.3

0.35

0.4

0.45

0.5

0.55

0.6

0.65(a)

16

88

17

80

18

01

18

67

18

90

19

03

19

07

19

11

19

15

19

19

19

23

19

27

19

31

19

35

19

39

19

43

19

47

19

51

19

55

19

59

19

63

19

67

19

71

19

75

19

79

19

83

19

87

19

91

19

95

19

99

Year

Shar

e

USA -- Top 10 Percent UK - Top 20 percent UK - Top 10 Percent

0.25

0.3

0.35

0.4

0.45

0.5

0.55

0.6

0.65(b)

16

88

17

80

18

01

18

67

18

90

19

03

19

07

19

11

19

15

19

19

19

23

19

27

19

31

19

35

19

39

19

43

19

47

19

51

19

55

19

59

19

63

19

67

19

71

19

75

19

79

19

83

19

87

19

91

19

95

19

99

Year

Shar

e

France -- Top 10 Percent Germany -- Top 10 Percent NL - Top 10 percent

Switzerland - Top 10 Percent Denmark -- Top 10 Percent

Figure 5(a) Evolution of income shares in the United Kingdom and the United States. (b) Evolution of income shares in continental Europe.

www.annualreviews.org • Origins and Persistence of Inequality 509

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

0.1

0.2

0.3

0.4

0.5

0.6

0.7

1770 1780 1790 1800 1810 1820 1830 1840 1850 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

Year

Shar

e of

Wea

lth

Hel

d b

y T

op 1

Per

cen

t

England and Wales USA - Lindert 2000 USA - Piketty & Saez 2006 Switzerland

Figure 6Evolution of wealth inequality, 1770–2000.

Gradstein 1999). The expansion of the fran-chise then led to the expansion of public ed-ucation, health spending, and direct transferssuch as unemployment benefits and pensions,and to a further reduction in inequality (Boix2001; Boix 2003, ch. 5; Lindert 2004).

The fragmented structure of the interna-tional system and the effect of interstate com-petition seem to explain well the progressiveadoption of (or lack of opposition to) a mod-ern regulatory structure. Reform happened firstin North Atlantic economies with conditionssimilar to Britain’s (a “dormant” craft sectorthat could transform itself into a full-fledged in-dustrial sector). It also took place in countriesdirectly affected by an ever-growing securitydilemma (Germany) or military defeat ( Japan).However, interstate competition was not a suf-ficient condition of change. In highly unequaleconomies, that is, in economies in which elites’wealth was highly specific and barriers to entrywere central to the political-economic regime(the American South, Russia, China, most LatinAmerican republics), reform thoroughly failed.The expected gains of reforms must not have

matched the expected losses in the view of theelites. Once again, the closer any society was tothe ancien regime “ideal type,” the more robustit turned out to be as a political equilibrium. Re-actionary political economies only petered outas a result of domestic revolutions and, most of-ten, of international war and invasions. Thoseplaces that were spared from interstate violence,such as Latin America, remained stuck in theold regime outcome.

As in Britain, all those followers who en-gaged in the construction of a liberal order andthe process of economic development enjoyeda shift in their aggregate production function.Human capital became a central component ofthe economy. This resulted in a higher marginalproduct of labor, higher wages, and a gener-alized process of wage compression. This ex-plains the secular decline in inequality acrossall Western countries in the past 100 years orso—without much need to resort to any kindof institutional story such as wage bargaining,electoral systems, and so on, except to accountfor (in relative terms) small cross-country dif-ferences. The share of income earned by the

510 Boix

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

top 10% in most advanced countries was onaverage about 20 percentage points higher in1920 than in 1970. By contrast, the differencebetween maximum and minimum national val-ues in 1970 was of about 5 percentage points.(Cross-national differences have increased inthe past 20 years, however.) Moreover, with theexception of development models that stronglyrelied on state-led industrialization and the for-mation of tight-knit economic elites, such asMeiji Japan (Moore 1966), the process of eco-nomic development fostered the gradual liber-alization of politics, which in turn led to theconstruction of a welfare state and growingequality of conditions.

American exceptionalism. In almost all cases,economic development and political liberaliza-tion came together (probably with the latter—at least in the sense of broader politicalparticipation—following the former). By con-trast, the northern half of the United States hadrelatively deep democratic institutions beforeindustrialization started. Figure 5a shows howa relatively equal economy experienced grow-ing inequality in wealth distribution (peakingat the turn of the century). Democratic insti-tutions remained robust, although new fran-chise requirements were introduced and theelectorate shrank a bit in the industrial Northin the early twentieth century. In line with thedistributive effects of regime type, human cap-ital formation was stepped up, resulting in astrong leveling of wages across the labor mar-ket (see Goldin 1999 on long-run human capi-tal formation in the United States). Democraticinstitutions, in interaction with the impact ofmassive immigration on the decisive voter’s in-come, were also likely to be behind the closingof American borders in the 1920s (see Mirilovic2010 for a first cut on the impact of democ-racy on immigration policy). The ensuing fall inforeign-born population as a proportion of totalpopulation strongly correlated with a sharp de-cline in inequality (McCarty et al. 2006). A plau-sible interpretation of the historical data seemsto confirm the idea that democracy’s institu-

tions tend to generate policy responses to curbinequality-enhancing technological shocks.

The past 30 years. As is apparent in Figure 5,income dispersion has grown since the 1970sin some Western (mostly Anglo-Saxon) coun-tries. Although the sources of the skill-biasedshock are still under discussion, with the lit-erature split between trade and strict techno-logical factors, the diversity of response acrossOECD states seems to be related to their differ-ent institutional structures (mostly their wage-bargaining institutions) and partisanship.

This result is not at odds with the previousdiscussion. In the short run, certain institutionalfactors (such as wage bargaining and partisangovernments) can reduce wage dispersion—atthe price of high unemployment or subsidizedemployment through the public sector. Butthose institutions cannot explain long-run in-equality because wage dispersion broadly tracksvariance in marginal productivities in the la-bor market. Institutional conditions that createwage equality artificially lead to the misallo-cation of resources, and they end up leadingto a suboptimal outcome in the long run. Ofcourse they can sustain an artificially low levelof wage dispersion (or, in fact, excessive in-come dispersion, as in authoritarian regimes)if they are complemented with measures thatsuppress the technological shocks that lead todifferential incomes—for example, protection-ism. But those measures are precisely what putthe economy in the (old-regime) path of stag-nation. Recent empirical studies confirm thatwage-bargaining institutions explain levels ofwage inequality after the 1970s but not before(Wallerstein 1999, Scheve & Stasavage 2009).In other words, Europe’s successful reductionof inequality over most of the twentieth cen-tury must be related to the construction of edu-cated labor forces and a de facto absence of eco-nomic competitors abroad. But without thosetwo conditions it will be unable to maintaina compressed wage structure without engag-ing in heavy spending and imposing new entrybarriers.

www.annualreviews.org • Origins and Persistence of Inequality 511

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

CONCLUSIONBroadly speaking, the distribution of income inhuman societies has been characterized by thefollowing patterns. Stateless, preagrarian soci-eties displayed (and still display wherever theyexist) relatively equal distributions of incomeand certainly a very low level of intergenera-tional transmission of wealth. The agriculturalrevolution and the concomitant formation ofthe state came hand in hand with the emer-gence of marked inequalities of income acrossindividuals and over generations. The level ofinequality has varied considerably across histor-ical communities since the invention of agricul-ture. Income inequality has trended downwardsince the industrial revolution—although thedegree of inequality has differed across coun-tries and there have been some temporal rever-sals within the process of equalization.

After reviewing current economic and in-stitutional models of the origins and evolutionof inequality, this article first develops a the-oretical model to explain the transition awayfrom relatively equal preagrarian societies. Aseries of biased technological shocks gener-ated inequality (within but mostly across hu-man bands), made spontaneous cooperation un-feasible, and heightened violent conflict. Thestate of violent anarchy that followed openedthe way to both political institutions and furthershifts in income distribution. In response to theproductivity growth, less productive individu-als responded by creating protection institu-tions in exchange for systematic payments frommore productive sectors. Alternatively, the lat-ter could choose to devote part of their incometo deter bandits. Both outcomes (“monarchical”and “republican”) had distinctive and sharp ef-fects on the distribution of resources. Militarytechnologies explained much of the variance inthe prevailing political outcome.

Once political institutions were in place, in-come distribution depended on how technolog-ical shocks and/or policy regimes affected thenet income of decision makers—directly, i.e.,in the returns of their assets; and indirectly,i.e., after political institutions and tax rates,

endogenous to the changing income distribu-tion, were put into place. This framework mayexplain why inequality, authoritarianism, andeconomic stagnation go together in an ancienregime polity. It can also be employed to probethe causes of the technological breakthrough ofindustrialization and its political consequences.It is finally applied to discuss the limits of in-equality under democracy and the relationshipbetween democracy, technological shocks, andcompensatory policies.

This more complex understanding of thesources of inequality turns out to affect our in-terpretation of the consequences of inequalityfor growth and development. Most of the re-cent work on growth and inequality claims thatthe latter has deleterious effects on economicdevelopment (Alesina & Rodrik 1994, Persson& Tabellini 1994, Perotti 1996). [By contrast,work done in the 1950s and 1960s had con-cluded that inequality covaried with growth,at least in its take-off phases (Kanbur 2000).]However, the empirical evidence that inequalityis always a direct cause of underdevelopment isinconclusive. Several recent articles have shownthat inequality has no effect on development oreven raises growth rates (Li et al. 1998, Forbes2000, Voitchovsky 2005, Barro 2008). A the-oretically plausible interpretation of these re-sults, given the political-economic frameworkpresented here, is that the potential correlationbetween inequality and development is alwaysconditional on the political and institutionalcauses behind a particular income distribution(Boix 2009). In ancien regime societies, wealthwas based on the ownership of very specificassets—assets whose value relied on regulatorybarriers sustained by the state, or assets (suchas land) that could not be easily redeployedin new, more productive economic sectors.Because wealth was so specific to the sectorin which it was deployed, the wealthy strataand the governing elite (which overlapped to agreat extent) had very little interest in changingthe status quo and accepting any technologicalbreakthrough that could upset their dominanteconomic and political position. Within that

512 Boix

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

political and economic context, inequality andlong-run economic stagnation (underpinnedby a nondemocratic regime) ended up takingplace simultaneously. That same pattern ofdifferential access to the state and to regulationand wealth prevails in developing economiestoday and may well explain their high levels

of inequality and meager growth. By contrast,in other instances, inequality covaries withgrowth. For example, the process of industri-alization led, through technological changeand factor shifting, to both higher incomesand more inequality, at least during its initialphase.

DISCLOSURE STATEMENT

The author is not aware of any affiliations, memberships, funding, or financial holdings that mightbe perceived as affecting the objectivity of this review.

ACKNOWLEDGMENTS

This article draws from work presented at the University of Virginia, the New York meeting ofthe “September Group,” the Santa Fe Institute, Princeton University, Stanford University, YaleUniversity, the University of Chicago, and the London School of Economics. I am grateful forthe comments of all their participants, in particular those of Alıcia Adsera, Sam Bowles, DanGingerich, Stathis Kalyvas, Carol Mershon, and Thomas Romer. I thank Teppei Yamamoto forhis research assistance.

LITERATURE CITED

Acemoglu D, Robinson JA. 2001. Inefficient redistribution. Am. Polit. Sci. Rev. 95(3):649–61Acemoglu D, Robinson JA. 2006. Economic Origins of Dictatorship and Democracy. New York: Cambridge Univ.

PressAdsera A, Boix C. 2002. Trade, democracy and the size of the public sector. Int. Organ. 56(Spring):229–62Aghion P, Bolton P. 1997. A trickle-down theory of growth and development with debt overhang. Rev. Econ.

Stud. 64(2):151–72Alesina A, Rodrik D. 1994. Distributive politics and economic growth. Q. J. Econ. 109(2):465–90Atkinson AB, Bourguignon F. 2000. Introduction: income distribution and economics. In Handbook of Income

Distribution, ed. AB Atkinson, F Bourguignon, pp. 1–58. Amsterdam: Elsevier Sci.Banerjee AV, Newman AF. 1991. Risk-bearing and the theory of income distribution. Rev. Econ. Stud.

58(2):211–35Banerjee AV, Newman AF. 1993. Occupational choice and the process of development. J. Polit. Econ.

101(2):274–98Bar-Yosef O, Meadow RH. 1995. The origins of agriculture in the near east. See Price & Gebauer 1995,

pp. 39–94Barro RJ. 2008. Inequality and growth revisited. Work. Pap. Ser. on Regional Economic Integration, No. 11,

Asian Dev. Bank, JanuaryBartels L. 2008. Unequal Democracy: The Political Economy of the New Gilded Age. Princeton, NJ: Princeton Univ.

PressBecker G, Tomes N. 1979. An equilibrium theory of the distribution of income and intergenerational mobility.

J. Polit. Econ. 87(6):1153–89Boix C. 1998. Political Parties, Growth and Inequality. New York: Cambridge Univ. PressBoix C. 2001. Democracy, development, and the public sector. Am. J. Polit. Sci. 45(1):1–17Boix C. 2003. Democracy and Redistribution. New York: Cambridge Univ. PressBoix C. 2008. Economic roots of civil wars and revolutions in the contemporary world. World Polit.

60(Apr.):390–437

www.annualreviews.org • Origins and Persistence of Inequality 513

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

Boix C. 2009. The conditional relationship between inequality and development. PS: Polit. Sci. Polit. 42(4):645–49

Boix C. 2010. A Theory of State Formation and the Origins of Inequality. Unpublished manuscript, PrincetonUniv.

Boix C, Rosenbluth F. 2006. Bones of Contention: The Political Economy of Height Inequality. Unpublishedmanuscript, Princeton Univ.

Bollen KA, Paxton P. 1997. Democracy before Athens. In Inequality, Democracy, and Economic Development, ed.M Midlarsky, pp. 13–44. New York: Cambridge Univ. Press

Bourguignon F, Morrisson C. 2002. Inequality among world citizens: 1820–1992. Am. Econ. Rev. 92(4):727–44Carneiro RL. 1970. A theory of the origin of the state. Science 169(Aug.):733–38Chagnon P. 1997. Yanomamo: Case Studies in Cultural Anthropology. Fort Worth: Harcourt Brace College

PublishersClastres P. 1972. Chronique des Indiens Guayaki. Paris: Ed. PlonClastres P. 1974. La societe contre l’Etat. Recherches d’anthropologie politique. Paris: ed. MinuitDhondt J. 1976. Le haut Moyen Age: VIIIe-XIe siecles. Paris: BordasDiamond J. 1997. Guns, Germs and Steel: The Fates of Human Societies. New York: W.W. NortonEngerman SL, Sokoloff KL. 2002. Factor endowments, inequality, and paths of development among new

world economies. Economia 3:41–102Esping-Andersen G. 1990. The Three Worlds of Welfare Capitalism. Cambridge, UK: PolityFehr E, Hoff K, Kshetramade M. 2008. Spite and development. Am. Econ. Rev. Pap. Proc. 98(2):495–99Fernandez R, Rodrik D. 1991. Resistance to reform: status quo bias in the presence of individual-specific

uncertainty. Am. Econ. Rev. 81(Dec.):1146–55Fogel RW. 1994. Economic growth, population theory, and physiology: the bearing of long-term processes

on the making of economic policy. Am. Econ. Rev. 84:369–95Forbes KJ. 2000. A reassessment of the relationship between inequality and growth. Am. Econ. Rev.

90(Sep.):869–87Frieden JA. 1991. Debt, Development, and Democracy: Modern Political Economy and Latin America, 1965–1985.

Princeton, NJ: Princeton Univ. PressFriedman T. 2006. The first law of petropolitics. For. Policy 154:28–39Galor O, Moav O. 2006. Das human capital: a theory of the demise of the class structure. Rev. Econ. Stud.

73(1):85–117Gat A. 2000. The human motivational complex: evolutionary theory and the causes of hunter-gatherer fighting.

Part I. Primary somatic and reproductive causes. Anthropol. Q. 73(1):20–34Geddes B. 2007. What causes democratization? In Oxford Handbook of Comparative Politics, ed. C Boix, S Stokes,

pp. 317–39. New York: Oxford Univ. PressGoldin C. 1999. Egalitarianism and the returns to education during the great transformation of education in

America. J. Polit. Econ. 107(6):65–94Goodin RE, Headey B, Muffels R, Dirven HJ. 1999. The Real Worlds of Welfare Capitalism. New York:

Cambridge Univ. PressGottschalk P, Smeeding TM. 1997. Cross-national comparisons of earnings and income inequality. J. Econ.

Lit. 35(2):633–87Greenwood J, Jovanovic B. 1990. Financial development, growth and the distribution of income. J. Polit. Econ.

98(5):1076–107Hibbs DA. 1977. Political parties and macroeconomic policy. Am. Polit. Sci. Rev. 71:1467–87Hibbs D, Olsson O. 2004. Geography, biogeography and why some countries are rich and others poor. Proc.

Natl. Acad. Sci. USA 101(10):3715–40Hirschman AO. 1981. Essays in Trespassing. Economics to Politics and Beyond. Cambridge, UK: Cambridge Univ.

PressHuber E, Stephens J. 2001. Development and Crisis of the Welfare State: Parties and Policies in Global Markets.

Chicago: Univ. Chicago PressJones EL. 1981. The European Miracle. New York: Cambridge Univ. PressJustman M, Gradstein M. 1999. The industrial revolution, political transition, and the subsequent decline in

inequality in nineteenth century Britain. Explorations Econ. Hist. 36:109–27

514 Boix

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

Kahhat J. 2007. Markets and the dynamics of inequality. Backgr. pap. prepared for “Markets, the State and theDynamics of Inequality: How to Advance Inclusive Growth,” United Nations Development Program,Bur. Development Policy, Poverty Group

Kanbur R. 2000. Income distribution and development. In Handbook of Income Distribution, ed. AB Atkinson,F Bourguignon, pp. 791–842. Amsterdam: Elsevier Sci.

Katzenstein P. 1985. Small States in World Markets: Industrial Policy in Europe. Ithaca, NY: Cornell Univ. PressKelly R. 1995. The Foraging Spectrum: Diversity in Hunter-Gatherer Lifeways. Washington, DC: Smithsonian

Inst. PressKrusell P, Rıos-Rull JV. 1996. Vested interests in a positive theory of stagnation and growth. Rev. Econ. Stud.

63(2):301–29Kuznets S. 1955. Economic growth and income inequality. Am. Econ. Rev. 45(Mar.):1–28LeBlanc S. 2003. Constant Battles: The Myth of the Peaceful, Noble Savage. New York: St. Martin’sLevi M. 1997. Consent, Dissent, and Patriotism. New York: Cambridge Univ. PressLi H, Squire L, Zou HF. 1998. Explaining international and intertemporal variations in income inequality.

Econ. J. 108(446):26–43Lindert P. 2000. Three centuries of inequality in Britain and America. In Handbook of Income Distribution, ed.

AB Atkinson, F Bourguignon, pp. 167–216. Amsterdam: Elsevier Sci.Lindert P. 2004. Growing Public. Social Spending and Economic Growth Since the Eighteenth Century. New York:

Cambridge Univ. PressMcCarty N, Poole KT, Rosenthal H. 2006. Polarized America: The Dance of Ideology and Unequal Riches.

Cambridge, MA: MIT PressMcNeill WH. 1982. The Pursuit of Power. Chicago: Univ. Chicago PressMeltzer AH, Richards SH. 1981. A rational theory of the size of government. J. Polit. Econ. 89:914–27Milanovic B, Lindert PH, Williamson JG. 2007. Measuring ancient inequality. NBER Work. Pap. No. 13550Mirilovic N. 2010. Explaining the politics of immigration: dictatorship, development, and defense. Comp. Polit.

In pressMoore B. 1966. Social Origins of Democracy and Dictatorship. Boston: BeaconMorrisson C. 2000. Historical perspectives on income distribution: the case of Europe. In Handbook of Income

Distribution, ed. AB Atkinson, F Bourguignon, pp. 220–61. Amsterdam: ElsevierOlson M. 1982. The Rise and Decline of Nations:Economic Growth, Stagflation, and Social Rigidities. New Haven,

CT: Yale Univ. PressOlson M. 1993. Dictatorship, democracy, and development. Am. Polit. Sci. Rev. 87:567–76Olson M. 2000. Power and Prosperity. New York: Basic BooksO’Rourke KH, Williamson JG. 1999. Globalization and History: The Evolution of a Nineteenth-Century Atlantic

Economy. Cambridge, MA: MIT PressPanter-Brick C, Layton RH, Rowley-Conwy P, eds. 2001. Hunter-Gatherers: An Interdisciplinary Perspective.

New York: Cambridge Univ. PressParente SL, Prescott EC. 1999. Monopoly rights: a barrier to riches. Am. Econ. Rev. 89(5):1216–33Peacock AT, Wiseman J. 1961. The Growth of Public Expenditure in the United Kingdom. Princeton, NJ: Princeton

Univ. PressPerotti R. 1996. Growth, income distribution, and democracy: what the data say. J. Econ. Growth 1(2):149–87Persson T, Tabellini G. 1994. Is inequality harmful for growth? Am. Econ. Rev. 84(3):600–21Piketty T. 1997. The dynamics of the wealth distribution and the interest rate with credit rationing. Rev. Econ.

Stud. 64(2):173–89Piketty T. 2000. Theories of persistent inequality and intergenerational mobility. In Handbook of Income

Distribution, ed. AB Atkinson, F Bourguignon, pp. 429–76. Amsterdam: Elsevier Sci.Pincus S. 2009. 1688: The First Modern Revolution. New Haven, CT: Yale Univ. PressPrice TD, Gebauer AB. 1995. Last Hunters—First Farmers: New Perspectives on the Prehistoric Transition to

Agriculture. Santa Fe, NM: School Am. Res. PressPrzeworski A. 2009. Conquered or granted? A history of suffrage extensions. Br. J. Polit. Sci. 39:291–321Rogowski R, McRae D. 2008. Inequality and institutions: what theory, history and (some) data tell us. In

Democracy, Inequality and Representation, ed. P Beramendi, CJ Anderson. New York: Russell Sage Found.

www.annualreviews.org • Origins and Persistence of Inequality 515

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

ANRV412-PL13-24 ARI 14 April 2010 18:9

Rogowski R, Kayser MA. 2002. Majoritarian electoral systems and consumer power: price-level evidence fromthe OECD countries. Am. J. Polit. Sci. 46(3):526–39

Rodrik D. 1998. Why do open economies have bigger governments? J. Polit. Econ. 106:997–1032Ross ML. 2001. Does oil hinder democracy? World Polit. 53(Apr.):325–61Scheve K, Stasavage D. 2009. Political institutions and income inequality, 1900–2000. World Polit.

61(Apr.):215–53Schonhardt-Bailey C. 1991. Specific factors, capital markets, portfolio diversification, and free trade: domestic

determinants of the repeal of the Corn Laws. World Polit. 43(July):545–69Smeeding TM. 2005. Government programs and social outcomes: the United States in comparative perspective.

Luxembourg Income Study Work. Pap. No. 426Steckel RH. 1995. Stature and the standard of living. J. Econ. Lit. 33(Dec.):1903–40Vicens Vives J. 1957. Historia Economica y Social de Espana y America. Barcelona: TeideVoitchovsky S. 2005. Does the profile of income inequality matter for economic growth? Distinguishing

between the effects of inequality in different parts of the income distribution. J. Econ. Growth 10(3):273–96

Wallerstein M. 1999. Wage-setting institutions and pay inequality in advanced industrial societies. Am. J.Polit. Sci. 43(July):649–80

Watson PJ. 1995. Explaining the transition to agriculture. See Price & Gebauer 1995, pp. 21–38Wright HT. 1977. Recent research on the origin of the state. Annu. Rev. Anthropol. 6:379–97

516 Boix

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

AR412-FM ARI 5 April 2010 21:38

Annual Review ofPolitical Science

Volume 13, 2010Contents

A Long Polycentric JourneyElinor Ostrom � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 1

What Political Science Can Learn from the New Political HistoryJulian E. Zelizer � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �25

Bridging the Qualitative-Quantitative Divide: Best Practices in theDevelopment of Historically Oriented Replication DatabasesEvan S. Lieberman � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �37

The Politics of Effective Foreign AidJoseph Wright and Matthew Winters � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �61

Accountability in Coalition GovernmentsJose Marıa Maravall � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �81

Rationalist Approaches to Conflict Prevention and ResolutionAndrew H. Kydd � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 101

Political Order and One-Party RuleBeatriz Magaloni and Ruth Kricheli � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 123

RegionalismEdward D. Mansfield and Etel Solingen � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 145

The Prosecution of Human Rights ViolationsMelissa Nobles � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 165

Christian DemocracyStathis N. Kalyvas and Kees van Kersbergen � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 183

Political Theory of Empire and ImperialismJennifer Pitts � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 211

The U.S. Decennial Census: Politics and Political ScienceKenneth Prewitt � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 237

Reflections on Ethnographic Work in Political ScienceLisa Wedeen � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 255

Treaty Compliance and ViolationBeth Simmons � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 273

v

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.

AR412-FM ARI 5 April 2010 21:38

Legislative ObstructionismGregory J. Wawro and Eric Schickler � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 297

The Geographic Distribution of Political PreferencesJonathan Rodden � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 321

The Politics of Inequality in America: A Political Economy FrameworkLawrence R. Jacobs and Joe Soss � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 341

The Immutability of Categories and the Reshaping of Southern PoliticsJ. Morgan Kousser � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 365

Indigenous Peoples’ Politics in Latin AmericaDonna Lee Van Cott � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 385

Representation and Accountability in CitiesJessica Trounstine � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 407

Public Opinion on Gender Issues: The Politics of Equity and RolesNancy Burns and Katherine Gallagher � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 425

Immigration and Social Policy in the United StatesRodney E. Hero � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 445

The Rise and Routinization of Social Capital, 1988–2008Michael Woolcock � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 469

Origins and Persistence of Economic InequalityCarles Boix � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 489

Parliamentary Control of Coalition GovernmentsKaare Strøm, Wolfgang C. Muller, and Daniel Markham Smith � � � � � � � � � � � � � � � � � � � � � � � 517

Indexes

Cumulative Index of Contributing Authors, Volumes 9–13 � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 537

Cumulative Index of Chapter Titles, Volumes 9–13 � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 539

Errata

An online log of corrections to Annual Review of Political Science articles may be foundat http://polisci.annualreviews.org/

vi Contents

Ann

u. R

ev. P

olit.

Sci

. 201

0.13

:489

-516

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Uni

vers

ity o

f A

laba

ma

- B

irm

ingh

am o

n 03

/10/

13. F

or p

erso

nal u

se o

nly.