The Feudal Revolution and Europe’s Rise:Institutional Divergence in the Christianand Muslim Worlds before 1500 CE

Lisa Blaydes∗ Eric Chaney†

September 1, 2011

Paper prepared for presentation at the annual meetings of theAmerican Political Science Association, Seattle WA, September 1-4, 2011

Abstract

This paper investigates the political origins of Europe’s economic rise by examiningthe emergence of increasing ruler durability in Western Europe when compared withthe Islamic world. While European rulers were less durable than their Muslim counter-parts in 800 CE, Christian kings became increasingly long-lived compared to Muslimsultans whose rule became less stable over time. The “break date” in Western Euro-pean political stability coincides with the emergence of feudal institutions, suggestinga first step in a political evolution that eventually led to medieval parliaments andthe emergence of a unique degree of constraint imposed on many Western Europeansovereigns. While feudal institutions served as the basis for military recruitment byEuropean monarchs, Muslim sultans relied on mamlukism — or the use of militaryslaves imported from non-Muslim lands. Dependence on mamluk armies limited thebargaining strength of local notables vis-a-vis the sultan, hindering the developmentof a productively adversarial relationship between ruler and local elites. We argue thatMuslim societies’ reliance on mamluks, rather than local elites, as the basis for militaryleadership, may explain why the Glorious Revolution occurred in England, not Egypt.

∗Department of Political Science, Stanford University†Department of Economics, Harvard University. This draft is preliminary and incomplete; please contact

authors for latest draft before citing. We thank David Abernathy, Phillipe Aghion, Carles Boix, GaryCox, Avner Greif, Justin Grimmer, Allen Hicken, David Laitin, Jonathan Leibowitz, Anne McCants, JimRobinson, Norman Schofield, Andrei Shleifer, Mike Tomz, Barry Weingast and seminar participants atHarvard University, Iowa State University, University of Michigan, Ohio State University, Sabancı University,Stanford University and Washington University in St. Louis for helpful conversations and comments.

1

“The kingdoms known to history have been governed in two ways: either by aprince and his servants, who, as ministers by his grade and permission, assist ingoverning the realm; or by a prince and by barons...Examples of these two kindsof government in our own time are the Turk and the King of France” (Machiavelli1903 [1532], pp. 14-15).

1 Introduction

An influential literature sees the roots of the industrial revolution in Europe’s unique in-stitutional framework.1 While it seems increasingly clear that growth-friendly, sovereign-constraining institutions — including respect for property rights and the rule of law — werekey to the emergence of sustained economic development in Europe, scholars struggle toexplain both how such institutions emerged and why they were initially limited to westernEurope.

Recent studies focusing on the evolution of European institutions generally begin theiranalysis after the year 1500 CE, while noting the peculiarity of Europe’s “initial” institu-tional framework.2 For example, in the conclusion of their seminal study of the evolution ofEnglish institutions following the Glorious Revolution, North and Weingast (1989) acknowl-edge that English institutions provided abnormal checks on the sovereign from an early (e.g.,medieval) date. Similarly, Acemoglu, Johnson and Robinson (2005) note that European po-litical institutions established prior to 1500 CE already placed “significant checks on themonarch.”

A distinguished line of scholars has stressed the feudal origins of European institutionalexceptionalism. Montesquieu (1989 [1748], p. 619) was an early proponent of this line ofthought arguing that feudalism “diminished the whole weight of lordship.”3 Recent schol-arship suggests that feudalism coincided with a rise of a powerful landed aristocracy thatproved instrumental in constraining the sovereign through the development of medieval par-liaments.4 Van Zanden, Buringh and Bosker (2010) provide an historical treatment of theemergence of European parliaments, arguing that these institutions ultimately facilitatedmedieval economic and institutional development; Stasavage (2010) describes the conditionsunder which parliamentary institutions endured in the medieval and early modern eras.

This paper uses data on ruler duration — the most significant political indicator thatis reliably available for the pre-modern period — in Western Europe and the Islamic worldto investigate the origins of European institutional exceptionalism. We begin by docu-

1For the economic importance of Western European institutional arrangements see among a large litera-ture North and Weingast (1989), De Long and Sheifer (1993), Acemoglu et al. (2005), North et al. (2009),Acemoglu and Robinson (nd).

2A notable and influential exception to this trend in the literature can be found in the work of Greif(1994) who examines the cultural determinants of institutional development in premodern societies of theMediterranean.

3Max Weber (1978, p. 1082) viewed feudalism as approximating constitutional government. Othershave claimed that “the institutional history of Europe, even of the United States, goes back to the age ofCharlemagne” (Ganshof 1968, p. ix) and that feudalism laid “the critical institutional groundwork uponwhich liberal democracy was built” (Downing 1992, p. 18).

4See Strayer (1970) and Downing (1989).

2

menting that after 1100 CE rulers in Christian Western Europe (henceforth referred tointerchangeably as Latin Europe or Europe) remained in power for significantly longer thantheir Muslim counterparts. This pattern is robust to geographic controls and restricting thesample to the Iberian Peninsula, where geography, state size and institutional heritage (e.g.,Roman/Germanic) can be held constant in a non-parametric manner. Without further inter-pretation — and in a reduced-form — this result suggests that Latin European institutionsgave these states an economic advantage over their Muslim counterparts roughly after theyear 1100 CE.5 Trend break algorithms are used to investigate the origins of the increase inEuropean political stability. Results identify a break in Latin European political stabilityin the year 790 CE. This date approximately coincides with the midpoint of Charlemagne’sreign (768-814 CE) and is consistent with an influential historical literature stressing theCarolingian origins of both feudalism and European institutional exceptionalism.

If the “feudal revolution” (Duby 1978) was the key to the divergence of Western Europefrom the rest of the world, what was it about feudalism that promoted both ruler stabil-ity and economic growth? And how did feudal institutions compare to methods of socialcontrol and organization in the Islamic world? Weakened by the Muslim invasions, Euro-pean monarchs lacked the financial resources to exclusively outsource their military needs toforeign mercenaries. The feudal relationships which evolved served as the foundation for mil-itary human resources as the landed nobility of Europe emerged as a “warrior class.” Whenmonarchical abuses emerged, European barons were able to impose forms of executive con-straint on European kings that formed the basis for more secure property rights. Sultans inthe Muslim world, by contrast, inherited more capable bureaucracies from conquered Byzan-tine and Sassanid lands and introduced mamlukism — or the use of slave soldiers importedfrom non-Muslim lands — as the primary means of elite military recruitment. Mamluks —completely segregated from the local population — swore their allegiance to the sultan alone.Local elites in the Muslim world did not serve as the source of elite military recruitment and,thus, were poorly positioned to impose the types of constraints on the executive that becameevident in Europe.6 Mamlukism — as a military-political institution —- enabled the rulerto bypass local elites in the raising of a military, leading to a concentrated, but brittle, formof power held by Muslim sovereigns compared to their European counterparts.7

The theoretical logic behind our historical narrative is straightforward; decentralizingpower lowers the payoff from successful revolt against the monarch for the aristocracy. Inother words, armed local elites in Europe were able to extract a better “soft contract” fromtheir monarch than in the Islamic world and were, therefore, less likely to overthrow thatmonarch despite their ability to do so.

5See Alesina et. al (1996) for a discussion of the link between political stability and economic development.6This pattern suggests a “reversal of fortune” though operating through a different mechanism than de-

scribed by Acemoglu et al. (2002). While Acemoglu et al. (2002) focus on the institutional reversal achievedin scarcely populated, underdeveloped areas, the reversal we propose is one where fiscal and administrativecapacity actually hindered long-term economic prosperity by providing Islamic dynasties with the means toavoid bargaining with their own elite populations.

7Mamlukism, in fact, became a defining feature of Muslim polities; indeed, the phenomenon of “slaveson horses” spread across Muslim dynasties and continued for a period of more than 800 years. Slaves onHorses: The Evolution of the Islamic Polity is the title of Patricia Crone’s influential study of mamlukismin the medieval Islamic world. She writes that rather than being a “topsy-turvy vision” slaves on horsesbecame in Islam the most “everyday of sights” (Crone 2003, p. 79).

3

Empirical analysis of ruler duration data is consistent with this mechanism and withtheories stressing the importance of political shocks as an impetus for generating Europe’sunique institutional arrangements. We demonstrate that European rulers was significantlymore stable — and European sovereigns more constrained — than rulers in the Muslimworld prior to the New World discoveries. We believe that the evolution of European polit-ical institutions provides a particularly sharp contrast to political developments in Muslimpolities, which, a priori, may have appeared to be the most likely candidates for growth anddevelopment.8 Our results are compatible with analysis by Acemoglu and Robinson (2001;2006; nd.) that suggests that economic downturns can facilitate institutional change.9

The remainder of the paper is structured as follows. The second section provides anhistorical narrative of our theory. The third section documents that rulers in the Latin Westremained in power longer than those in the Islamic world after 1100 CE. The fourth sectioninvestigates observable implications associated with the causal channel we have described.A fifth section considers alternative explanations and possible confounding factors. A sixthsection concludes.

2 Historical Narrative

In 1188 CE, Alfonso IX (1188-1230) of Leon (Spain) convened the world’s first parliament. Bythe 13th century similar institutional arrangements had spread throughout Western Europe(Stasavage 2010). One scholar has summarized the importance of this development by noting:

“Late medieval Europe had numerous political characteristics that distinguishedit from other major world civilizations. These characteristics, the most importantof which were representative assemblies, constituted a basis for liberal democracy,which provided Europe with a predisposition toward democratic political insti-tutions” (Downing 1992, p. 3).

If the emergence of representative parliamentary institutions marks the start of Europe’spolitical and economic development culminating in the industrial revolution of the 18th and19th centuries, understanding the origins of these institutions is foundational. Yet explaining

8For example, Kuran (1997) writes that by the 10th century, the Islamic world had achieved a higher levelof economic advancement compared to Christian Europe. During Islam’s “Golden Age,” public hospitals,libraries and universities were established. Mills of all types were in use and Muslim engineers createdadvanced water systems for the Islamic world’s highly urbanized population. Urbanization has been describedas a very strong indicator of economic prosperity since urban populations can only be supported by areaswith high agricultural productivity (Acemoglu and Robinson 2002). Scientific and mathematical discoveryflourished. The Islamic world, with its large urban centers and prosperous traders, appeared to exceed theWest on multiple indicators of economic prosperity. As time progressed, however, the relative success of theIslamic world became less clear and by the 19th century there was little doubt that the Muslim world hadfallen behind Europe in terms of economic development.

9We view our results as complementary to other studies of the rise of the West. For example, by the endof our sample, political stability appears to decrease in much of Western Europe. Evidence suggesting thatthe Atlantic discoveries positively affected the evolution of some Western European entities (Acemoglu et al.2005) while negatively affecting others (Drelichman and Voth 2008) is fundamental to understanding whythe sovereign-constraining institutions of the Middle Ages survived to a greater extent in some areas thanin others.

4

how Europe came to develop growth-promoting political institutions is virtually impossi-ble through an examination of Europe alone. Indeed, understanding the determinants ofsustained economic growth in Europe demands comparison with an appropriate historicalcounterfactual case or set of cases. As a result, we explore the roots of the European eco-nomic “miracle” through an examination of the political origins of institutional divergencein the Christian and Muslim worlds prior to 1500 CE.

2.1 Feudalism, Parliaments and the Rise of Europe

The usual narrative describing the birth of representative, sovereign-constraining politicalinstitutions begins with the collapse of the Western Roman Empire.10 The fiscal positionof the Germanic successor states to the Roman Empire tended to be weak. Unable to fundmilitary expenditure through tax receipts, European rulers sought other avenues for raisingarmies. The innovations introduced by Charlemagne marked a pivotal change. Lackingthe capacity to introduce a system of tax collection, Charlemagne required landholders tocontribute troops instead of funds.

This change increased the power of large landlords in two ways. First, small, independentlandowners pooled their lands with those of larger landholders to avoid having to offerthemselves up for military service. As individual landholders began to “aggregate up,”large landowners emerged who could ensure the cultivation of land while distributing theburden of military service across the larger body of peasants. Second and contemporaneously,European kings — like Charlemagne — required mounted troops, not just infantrymen,as a result of the introduction of the stirrup. The technological innovation of the stirrupmeant that “mounted shock combat” became the norm in warfare and the large investmentrequired to purchase a horse and armor for battle meant that monarchs needed to recruitindividuals with wealth to serve as the mounted military elite (White 1962). Mountedwarriors, or knights, were often compensated for their service to the king through landgrants.11 Military service and loyalty were expected in exchange for control of land (Northet al. 2009, p. 79). European barons operating in the feudal system entered battle with theirown, privately financed equipment, archers, and associated infantry. Such individuals oftenenjoyed opportunities to increase their landholdings or other forms of advancement as a resultof their fighting. Together, the methods of military recruitment that emerged in medievalEurope came to be known as the feudal system.12 The net result of these innovations wasthe creation of a landed aristocracy in Western Europe.13

10The decline of Rome as a location of centralized authority was accompanied by a depopulation ofurban centers as Roman citizens began moving to the countryside. The move to manors, and subsequentdevelopment of manorialism, was motivated by a search for basic food security. Egypt was long known to bethe breadbasket of the Roman Empire. With the empire’s decline, trade between North Africa and Europedeteriorated to the point that former urban dwellers began to gravitate to manors where they might engagein agricultural production.

11Eventually European rulers transformed feudal obligations into revenue as vassals paid to commute theirservice, allowing for the cultivation of standing and mercenary armies (Levi 1989, p. 106).

12The definition of feudalism is much debated. Here, we define feudalism as a system of military mobiliza-tion and organization distinct from manorialism, the economic system that provides the basis for feudalism.

13The process that led to the disintegration of the classical (Roman) institutional framework and the emer-gence of a decentralized “feudal” framework remains a topic of scholarly division. A competing hypothesis to

5

Strayer (1970) provides a particularly compelling discussion of the feudal system andits consequences for state development. He argues that in the weakened and cash-strappedenvironment of early medieval Europe, “standing armies or permanent officer corps were un-thinkable” (Strayer 1970, p. 27). Feudalism, characterized by its “fragmentation of politicalpower” (1970, p. 14), emerged whereby rulers would raise armies on an as-needed basis byoffering inducements of land or other privileges in exchange for support. While the systemmight appear to work against the creation of an effective state in the short run, Strayerargues that ultimately such a system “...can become a basis for state-building” (1970, p.15). Although he does not bring any systematic empirical data to bear on this question, heargues that there was a notable increase in Western European political stability following1000 CE and it was this stabilization of the political scene that allowed for an economicrevival that included higher levels of agricultural production, population growth and a re-vitalization of long-distance commerce (Strayer 1970, p. 19).14 He finds that during thisperiod of deepening political stability, the basic components of the modern state began toappear in Europe (Strayer 1970, p. 34).

The stability of European monarchs evolved hand-in-hand with both increased economicopportunities and growing constraints on the executive. Europe’s more stable political envi-ronment contributed to the rise of towns and a nascent commercial revolution that becameapparent beginning in the 12th century. Peasants seeking opportunities to sell handicraftsand agricultural surplus sought out small markets and fairs. Markets were only able arise inplaces where political stability allowed for defense from bandits and marauders. The natureof elite military recruitment under feudalism also led monarchical abuses to be self-limiting.Barons — who served as vassals to the king — had the military means by which to rebeland demand satisfaction of their grievances (Breay 2002). The independent military powerof the barons allowed for a degree of bargaining strength vis-a-vis the monarchy as baronscould either rebel against the king or support an opposition figure who might meet theirdemands in exchange for support.

English barons, for example, came to limit the power of kings in a number of meaningfulways. The promulgation of the Magna Carta in 1215 and eventual establishment of anEnglish parliament populated by knights and barons in 1265 serve as a useful example.Under feudal institutions, the king had the right to demand “military service...wherebykings expected their vassals to contribute either in men or in money to armies” (Holt 1992,p. 30). Military service was a source of “widespread and perennial acrimony” between theking and his vassals (Holt 1992, p. 78). King John’s loss of Normandy in 1204 led to a

the one put forward here was introduced in the first half of the 20th century by renowned Belgian historianHenri Pirenne. Pirenne advanced the controversial hypothesis that the Islamic invasions of the Mediter-ranean basin in the 7th and 8th centuries were the key catalyst leading to the emergence of feudalism andEurope’s unique subsequent institutional development. He claimed that these invasions cut trade betweenthe northern and southern Mediterranean and the subsequent disappearance of trade led to a sharp drop intax revenues, forcing rulers in what is today France to compensate their military with land. Pirenne saw theempowerment of the aristocracy complete by the reign of Charlemagne. He famously remarked that “[t]heEmpire of Charlemagne was the critical point of the rupture by Islam of the European equilibrium” Pirenne(1980 [1939], p. 234).

14A variety of studies show that by the late medieval period, interest rates in Western Europe had begunto decline dramatically [e.g, see Clark (1988), Clark (2007) and Epstein (2000)], perhaps also a result of theincreased political stability that we identify.

6

growing reliance on local barons for both men and money. The Magna Carta — whichreflected a greater acceptance of baronial demands than King John had hoped to make —was signed in June 1215 with a “renewal of homage and fealty” on the part of English barons(Holt 1992, p. 189). The Magna Carta was a direct product of both King John’s militaryfailures and his future needs, where “war was the compulsive urgency behind administrativeexperiment” (Holt 1992, p. 24-25).15 The Magna Carta laid the groundwork for futuredemands to limit the power of the monarch in England. Over time, a coalition of Englishelites established credible constraints on the executive with the Glorious Revolution of 1688,allowing for property rights and security from arbitrary taxation that ultimately encouragedeconomic growth (North and Weingast 1989).

In England and beyond, feudalism represented a meaningful fragmentation of politicalauthority.16 Kings — while the technical heads of government in their respective territories— had ceded considerable strength to local strongmen who enjoyed both public and privatepower, including control over public goods provision and land and rental income (Bisson1994). In the face of divided and decentralized political power, how were European sovereignsable to increase their length of rule? Our argument is that more consensual government — asit emerged in Europe, with roots in medieval feudalism — enjoyed an advantage in terms ofpolitical stability. Forced by economic weakness to bargain and negotiate with local elites,European monarchs developed forms of political organization that exhibited a flexibilitywhich trumped forms of more absolutist rule. These governmental forms contrast sharplywith political organization and military recruitment in the Islamic world during the sametime period.

2.2 The Islamic Equilibrium

“A monarchy where there is no nobility at all, is ever a pure and absolute tyranny;as that of the Turks” (Bacon 1819, p. 282).

Political development in the Islamic world provides an important comparison to institu-tional evolution in western Europe.17 Like the Latin West, Muslim states ruled over someof the wealthiest Roman provinces and had access to the institutional heritage of ancientGreece, Rome and the Germanic states. Muslim states also controlled some areas that even-tually reverted to Latin control. And, like Christian Europe, the Islamic world possessed apolitically influential “clergy.”

Yet, feudalism — with its complex system of interlocking economic and military rightsand obligations — did not emerge in the Islamic world. Despite being largely agrarian, no

15At around the same time European monarchs on the continent were also ceding liberties to vassals andbarons (Holt 1992, p. 25-26). The Golden Bull of 1222 in Hungary laid out the rights of knights and countsunder the feudal system (Holt 1992, pp. 77-78). Regional parliaments in France were established in the 13thcentury. The English parliament began to meet regularly beginning in 1295 (Bosker et al. 2010).

16See Bisson (1994) for more on this point.17Islam first emerged in the Arabian peninsula in the 7th century and within one-hundred years, Arab

Muslims came to occupy territory from the Iberian Peninsula to the Indus Valley after successful attacks onthe Sassanid, Byzantine and other empires.

7

“landed aristocracy or gentry” materialized (Crone 1999, p. 322) nor did nascent parlia-mentary institutions develop. How were Muslim rulers able to circumvent the emergence ofthe type of landed aristocracy that proved so critical to constraining monarchs in Europe?This section argues that divergence in the nature of elite military recruitment provides aconvincing explanation for why Christian Europe was ultimately able to develop growth-friendly political institutions. We argue that Muslim reliance on mamluks — or militaryslaves imported from non-Muslim lands — weakened state-society relations and hinderedthe development of impersonal political institutions. The widespread use of mamluks in theIslamic world limited the bargaining leverage enjoyed by local elites vis-a-vis the sultan, thushandicapping the development of the type of productively adversarial (and mutually depen-dent) relationship between ruler and ruled that emerged in Europe and which became thebasis for forms of executive constraint. This is because mamluks were characterized by both“cultural dissociation” as a result of their emigration from a distant locale and “personaldependence” on the sultan who served as their master (Crone 2003, p. 79). Thus, whileEuropean rulers were negotiating with local gentry to raise armies for matters of defense,Islamic rulers bypassed local elites by creating highly-skilled armies of foreigners who hadno ties to the existing gentry and swore allegiance directly to the sultan.

Historians of the medieval Islamic world have come to describe the introduction andeventual widespread adoption of mamluk institutions as a uniquely Islamic phenomenon.A mamluk is generally described as a military slave, though the term also refers to suchindividuals after their emancipation (Irwin 1986, p. 3).18 Mamluks might better be definedas elite military slaves given the fact that they were typically well-trained and generouslypaid.19 As such, mamluks were not prototypical slaves, but rather military elite who mightserve in positions like falconer, provincial governor or treasurer (Irwin 1986, p. 4).20

Who became mamluks? The most sought after mamluks were of tribal origin importedfrom areas “marginal to the settled Islamic world” (Crone 2003, p. 78) like the Caucasus(present-day Georgia, Armenia and Azerbaijan) and Transoxania (present-day Uzbekistan,Tajikistan and Kazakhstan).21 Imported as children, mamluks often underwent years oftraining which sought to both imbue them with military skills but also to encourage theirloyalty to the sultan (Pipes 1981, p. 9).

A number of factors made mamluks from the Caucuses and Transoxania (henceforthdescribed as “Turks” or Turkish mamluks) particularly valuable. Pipes suggests that oneadvantage Turks may have had over non-Turks in their recruitment as mamluks is related

18How can we think about mamluks in comparison to mercenaries that were frequently employed alongsideskilled knights and town militias in Europe during this period? While mercenaries might be hired for aparticular military campaign, offering their services to the highest bidder, mamluks were bought as slaves,often as children, and then carefully trained in the military arts to serve a particular sultan.

19While homeborn freemen were still used as foot soldiers, the “crack troops” (Crone 2003, p. 80) or“backbone” (Ayalon 1994b, p. 17) of the sultan’s army typically consisted of soldier slaves.

20Mamluks were first introduced by the Abbasids in the 9th century as a retinue of three to four thousandTurks of non-Muslim origin (Crone 1999, p. 319). This new army of crack troops became the basis forthe sultan’s strength (Kennedy 2004, p. 159). Mamluk armies were quickly adopted by numerous Muslimpolities (Ayalon 1994a, p. 25) and spread across the settled areas of the Islamic world (Crone 2003, p. 79).

21Though less common, black Africans also served as mamluks (Irwin 1986, p. 5). Fellow Muslims couldnot be enslaved and “People of the Book” — like Christians and Jews — were also protected from slaveryand, thus, not eligible to serve as mamluks (Irwin 1986, p. 9).

8

to the stirrup. He writes that the introduction of the stirrup “enhanced the power of thepeoples living where horses could be raised — primarily in the steppe lands and in deserts— and reduced the strength of peoples living in densely inhabited areas” (1981, p. 57).Hodgson also points out the “steppes formed the most outstanding source of young slaves”as a result of the “boyhood military training as horsemen” (1977, p. 399). In addition, livingin the mountains, deserts and steppes of Central Asia and the Caucasus, Turkish mamlukswere raised under conditions of extreme hardship, leading them to be both healthy and lean(Pipes 1981, p. 78).22

In order to stem the “corruption” of military slaves from the luxuries of settled life,mamluk institutions created a disconnect between the soldier slaves and local society. Todeter the mamluks from being able to develop ties to either the local population or home-borntroops, the mamluks were kept in “strict isolation” (Crone 1999, p. 319). Mamluks typicallybore their Turkish names, even after their conversion to Islam, and predominantly usedTurkish when speaking to one other with often superficial knowledge of the local language(Ayalon 1994b, pp. 16-17). Mamluks were mainly married to female slaves from theircountries of origin rather than local women (Ayalon 1994b, p. 16). The sons of mamluks(who did not enjoy mamluk status themselves) more frequently married women from the localpopulation thus offering one opportunity for assimilation into non-mamluk society (Ayalon1994b, pp. 16-7).23 A mamluk, then, was characterized by both his “personal dependence”on his master, the sultan, as well as his “cultural dissociation” (Crone 2003, p. 74) givenboth his foreign origin and the development of practices that kept him highly removed fromthe local populace.

There is no consensus in the existing literature regarding why mamlukism emerged andspread throughout the Islamic world.24 In particular, why didn’t medieval Muslim sultansuse indirect rule as we observe emerged in Europe? Patricia Crone offers perhaps the mostcompelling explanation for why mamlukism arose in the Islamic world. Crone comparesthe Abbasids explicitly to the Carolingians — their contemporaries — who also faced thechallenge of creating a polity for which their past experience offered no model. She writes,

22The medieval Arab historian Ibn Khaldun offers other ideas for why soldiers brought up in marginalareas enjoyed a huge advantage over those recruited from more settled districts. Marginal areas existedseparate from governmental authority forcing local peoples to develop a sense of group solidarity, or whatIbn Khaldun calls ‘asabiyya. To protect themselves from the harsh environmental conditions and attack,these communities developed codes of honor and social structures for defense (Pipes 1981, p. 78). In practice,it is likely that all of these factors contributed to the desirability of Turks as mamluks. Those individualsthat survived to be recruited as military slaves were not only physically powerful but also natural horsemenwho were imbued with the group solidarity that would make for ideal soldiers to serve the sultan. Soldierslaves from Greece, India, Sub-Saharan Africa and other areas on the fringe of the Islamic world did existbut were not sought after like the Turks.

23Many of the factors that favored Turkish mamluks from marginal areas of the Caucuses and Central Asiawere not transferable across generations suggesting that there were both religious and practical reasons fornot allowing mamluk status to be passed from father to son. As the qualities that made Turkish mamluks sovaluable were not innate but rather acquired characteristics (Pipes 1981, p. 81), a sultan’s stock of militaryslaves had to be constantly renewed. Maintaining military slaves was a costly proposition, then, forcing alarge percentage of state resources into a human capital investment that required constant renewal.

24The mamluk institution can be considered a “specifically Muslim institution” as it came to be nearlyubiquitous in the Islamic world and yet totally absent in both the pre-Islamic Middle East as well as thenon-Islamic world (Crone 2003, p. 80).

9

“both fell back on private ties, and in both cases, the outcome was political fragmentation.But because the fiscal and administrative machinery survived in the east, the Abbasids couldsimply buy the retainers they needed, and so they lost their power not to lords and vassalsbut to freedmen [i.e., manumitted mamluks]” (Crone 1999, p. 326, emphasis added).25 Thissuggests that the superior economic position of the Muslim rulers allowed them to importthe military support that they needed rather than to develop a system of feudalism where aking delegated land — and political power — to local lords.26

Imported military slaves were thought to be “safest to rely on” by a sultan (Marshall1977, p. 399), offering the “most efficient defense” of the ruler’s interests (Lapidus 1973).Indeed, according to one observer, the “principal deterrent to the sultan’s overthrow wasthe strength and loyalty of the royal mamluks” (Dols 1977, p. 148). If a “well-controlled”mamluk army could bring political stability to a polity, an uncontrolled one was a potentialsource of regime breakdown and disintegration (Crone 2003, p. 84). In some cases, sultansfound themselves “imprisoned” by their own “praetorian guard” (Lapidus 1973, pp. 37-38)who were successful at usurping power from within (Pipes 1981, p. 91). Purchasing slaves— who needed to be constantly replenished — was also quite expensive, leading to economicproblems for many regimes (Pipes 1981, p. 88). In some cases, military slaves came tothreaten the very dynasties that had trained them, eventually establishing their own slavesultanates (Pipes 1981, p. 23; Dale 2010, p. 16).

Perhaps more pernicious than the direct challenge mamlukism posed on ruler stabilitywas the indirect impact of mamlukism on state-society relations. Military slaves who “hadno roots in or commitments to local communities” were responsible for collecting taxes,maintaining order and controlling important resources — the result of which was a highlyexploitative system (Lapidus 1973, p. 39). Mamluks would typically hold a temporary,nonhereditary deed to land which “amounted to nothing more than a stipend” while livingin urban areas far from their agricultural holdings (Borsch 2005, pp. 26-32). Borsch arguesthat the distance between mamluk deedholders and their tenants was as “cultural and psy-chological as it was geographical” (Borsch 2005, p. 27). Sultans — reliant on their mamlukcoterie for enforcing economic and political control — found themselves “alienated from themass of their subjects” (Lapidus 1973, pp. 37-38).27

25Slaves were not permitted to exercise jurisdiction over freemen and, as such, Mamluks were typicallymanumitted prior to their first military engagement (Irwin 1986, p. 9). The practice of both convertingand freeing a mamluk prior to battle had the important consequence of barring him from passing on mam-luk status to his children (Irwin 1986, p. 9). As a result, the sons of mamluks could not belong to themamluk aristocratic caste that emerged (Ayalon 1994c, p. 205) with important consequences for issues ofintergenerational exchange.

26This perspective is largely consistent with other prominent accounts. According to Mann (1986, p. 393),Europe at this time had a “fairly primitive economy” where “no lord could generate the liquid wealth to paya large number of mercenaries. The only solution was land grants, which gave the vassal soldier a potentiallyautonomous power base.” Similarly, White (1962, p. 29) describes the Christian west in the 8th centuryas being a much less sophisticated economy than that found in the Islamic world or the empires which itconquered. According to White (1962, p. 29), “the bureaucracy of the Carolingian kingdom was so slenderthat the collection of taxes by the central government was difficult.” Given the expenses associated withraising a military force in an era of mounted shock combat, like horses and armor military service became“a matter of class” (White 1962, p. 30).

27Extractive institutions, which allow the leadership to siphon off resources from the rest of society, alsodiscourage both investment and development (Acemoglu and Robinson 2002).

10

The provision of military service in medieval Europe, then, was highly decentralized incontrast to the mamluk system where military slaves constituted a centrally-located andethnically distinct, indeed alien, caste. Mamluks were unable to transform themselves into a“hereditary landed baronage,” in part because of the “impossibility of transmitting mamlukstatus to one’s children” (Irwin 1986, p. 8).28 Thus, while Western Europe saw a strength-ening of lords who were responsible for defense of the land (Duby 1974, pp. 43, 162), theMuslim world saw a deterioration in the bargaining strength of the aristocracy as controlof the means of violence became dominated by a caste of military slaves.29 The relativebargaining strength of the gentry vis-a-vis the ruler has proven to have profound implica-tions for the development of executive constraint and the creation of impersonal economicinstitutions.

2.3 Developing Testable Predictions

The narrative above describes a process of political divergence in the Christian and Muslimworlds beginning in the 8th century following the end of Roman hegemony in the Mediter-ranean basin. Rulers in Christian Europe — operating under conditions of low fiscal andbureaucratic capacity — were forced to enter into forms of consensual rule with their localelite. Domestic elites were recruited as the backbone of the military corps and rewarded fortheir service and loyalty with land grants that might be passed down to their sons. The“feudal complex” — as this system came to be known – rolled out across continental Europeto places like England, Spain and Scandinavia along Carolingian lines; feudal institutionsexpanded less evenly to Eastern Europe where they “underwent numerous local dislocationsand torsions” (Anderson 1979, p. 411). In the Islamic world, on the other hand, relativelywealthy rulers with efficacious bureaucracies invested in the long-term training of foreignerswho drew a salary based on agricultural output but were not culturally or physically tied tothe land or even a particular locality. Intentionally separated from local elites and connectedto sultans through a master-slave relationship, mamluks were “not readily convertible into arural nobility” (Anderson 1979, p. 506). While the political power of the European landedaristocracy increased over time, leading to a gradual transfer of power (e.g., control overmonetary rents, public goods and the legal system) from the sovereign to his “vassals” (e.g.,the landed aristocracy), a similar process was not observed in the Islamic world.

Downing concisely summarizes these ideas for the European case as follows:

“[t]he key to the rough balance between crown and nobility lies in the incom-plete collapse of the Carolingian Empire in the ninth century and [...] then [the]

28According to Crone, Islam became unique among civilizations in terms of the extent to which governmentservice ceased to be associated with land ownership (Crone 2003, p. 87). While military slaves did enjoy theability to serve as tax collectors through the granting of iqta’ (Islamic land grant), Crone points out that“slave soldiers were no barons” as the iqta’ did not invest the soldiery with land in a way comparable to theEuropean fief (Crone 2003, p. 87). Interestingly, it was not until 1574 with the accession of Ottoman sultanSelim II that the janisarries (Ottoman mamluks) were able to pass on mamluk status to their sons. Andersonwrites that “a professional, skill-selected military elite was thus progressively converted into a hereditary,semi-artisanal militia.....its discipline disintegrated proportionately (Anderson 1979, p. 381).

29A related argument is put forward by Levi (1989) who finds that the relative bargaining power ofmonarchs against their resource rich constituents was the key variable in explaining divergence in politicaldevelopment in early modern France and England.

11

contestation between the prince and local centers of power. Within this dualsovereignty emerged compromises, power sharing, and a climate of partial trustand partial mistrust that formed much of medieval constitutionalism” (1989, pp.214-215).

The result was the emergence of a set of political institutions and norms in ChristianEurope which have been associated in the contemporary literature with forms of executiveconstraint. Constraint on the sovereign did not emerge without considerable local pressureand contestation. Medieval parliaments increased in importance and came to serve as a“logical extension of the traditional presentation of auxilium et consilium — aid and advice— by the vassal to his overlord” (Anderson 1979, p. 411). A sovereign’s ability to taxwithout consent diminished under the relatively strength and influence of local notables.The ability of European elites to guard against abuses of the executive increased during themedieval period while comparable developments were absent in Muslim polities. The causalchain we propose is depicted in Figure 1.

While constraint on the executive is surely associated with forms of intense debate,discussion, political pressure and contestation we contend that it is simultaneously correlatedwith longer terms of rule for sitting monarchs. The intuition behind how a decrease in thesovereign’s political control can lead to an increase in his duration in power is straightforward.Decentralizing power lowers the payoff from successful revolt against the monarch for thearistocracy. In other words, the development of a landed aristocracy with political powerdecreases the “wedge” between the payoff to the aristocracy versus the payoff to the sovereign.If the cost of overthrowing the sovereign remains constant, the smaller wedge between payoffsto the king and the aristocracy should lead to revolts in fewer states of the world. While itis possible that the cost of overthrowing the sovereign also declined as political power flowedto his vassals, this point is far from obvious based on our reading of the historical record.Indeed, decentralization of political power appears to have made coordination across noblesmore difficult and costly.

A key observable implication of this logic is that ruler duration should increase in Eu-ropean polities with the introduction and spread of feudal institutions. Furthermore, thiseffect should be most pronounced in locations close to the capital of the Carolingian empire(where feudal reforms began), radiating outward over time. Finally, the presence of parlia-ments and existence of executive constraint should be correlated with ruler duration evenwithin Europe if the logic we have described is correct.

Why focus on ruler duration? The most basic unit of political analysis for both ChristianEurope and the Islamic world during the medieval period is the monarch, whether he isknown as a king or sultan. The medieval period was characterized by the proliferation ofhundreds of dynasties for which students of history and medieval numismatics have investedtremendously scholarly effort in creating leadership chronologies. As a result, ruler durationmay be the most reliable, political significant indicator for which data is available for a wideswath of both time and territory. The following section explicitly examines changing trendsin ruler duration across Christian and Muslim dynasties from the mid-7th century to thestart of the 16th century.

12

3 Political Stability in Europe and the Islamic World

Scholars have argued that the political institutions that emerged in Western Europe in thelate Middle Ages proved to be growth-enhancing. Thus far, we have offered an historicalnarrative stressing the Carolingian origins of European institutional exceptionalism with adiscussion of Islamic political institutions as a contrasting case. In this section, we explorethe empirical implications associated with these ideas, particularly as they relate to rulerduration.

3.1 Data

We begin by defining the population of relevant political entities. Using a series of digitizedmaps of Europe, North Africa and much of the Middle East developed by Nussli (2011), weinclude all sovereign states in our sample. Using maps for the years 700, 800, 900, 1000, 1100,1200, 1300 and 1400 CE, we assign rulers who assumed power on the interval [t,t+100) to themap of year t. In other words, our sample is updated every 100 years to include the entitiesNussli (2011) denotes as sovereign states. In the data appendix we provide the number ofrulers and the average duration of each dynasty (both those that we were able to identifyand those that we were not able to identify) for each map layer.

Although Nussli’s data set has many advantages, one may be concerned that the Muslimworld is under-represented in this data set or that a data set constructed in this mannersuffers from other selection issues. To address these concerns we also use data drawn fromBosworth (1996) and Morby (1989) as robustness checks (we do not include these results dueto space constraints although these are available upon request) and we use Morby’s data topinpoint the break date in European political stability. Bosworth’s book deals exclusivelywith the Islamic world and contains data on ruler duration in 186 dynasties. Morby’s book isentitled Dynasties of the World and provides ruler durations for dynasties across the worldand across history with “an admitted emphasis on Europe and on its roots in the ancientworld” (Morby 1989, p. vii).

While both Bosworth and Morby attempt to provide as comprehensive an overview aspossible, it is more difficult to assess potential selection issues in these sources. Fortunately,the qualitative implications of the results are similar when using Nussli’s or Morby andBosworth’s data. This suggests that selection issues are not likely to be driving the resultsin either data set since these data sets were compiled independently.

Finally, we concentrate on rulers that assumed power before the year 1500. This was donein order to focus on the evolution of political stability in the period prior to the Atlanticdiscoveries.

3.2 Documenting the Divergence in Ruler Duration

We begin by documenting the divergence between ruler durations in the Islamic world andWestern Europe. To do this, we run the following regression:

durationidt = θtdt +1400∑t=700

αt ·WEi · dt +1400∑t=700

βt · EEi · dt +X ′dtγ + εidt (1)

13

where durationidt gives the duration in power of ruler i in dynasty d that assumed poweron the interval [t,t+100). The dt are century dummy variables, WEi is a variable equalto one if ruler i assumed power in a non-Muslim political entity west of Venice, EEi is adummy variable equal to one if ruler i assumed power in a non-Muslim political entity eastof Venice and Xdt is a vector of geographic controls (the area of the political entity as wellas the average agricultural suitability of the area).

The coefficients θt, αt and βt are presented in columns (1)-(8) of Table 1. Columns (1)-(3) report the results for regression (1) omitting the vector of geographic characteristics Xdt.Column (1) gives θt which are the mean values of ruler duration in the Islamic world in eachcentury. Column (2) provides the coefficients αt which give the difference between the meanvalue of ruler duration in Western Europe and the Islamic world. Column (3) provides thecoefficients βt which give the difference between the mean value of ruler duration in EasternEurope and the Islamic World. Column (4) provides the p-value for the test of the equalityof the coefficients in columns (2) and (3) in each century. Columns (5)-(8) have the sameformat as columns (1)-(4) except for these columns, geographic control variables have beenincluded in the associated regressions.

The results show that after 1100 CE rulers in Western Europe remain in power for astatistically longer period than their Muslim counterparts. Although after 1300 CE the samepattern appears to emerge in Eastern Europe, the results in column (4) demonstrate thatrulers persisted for longer reigns in Western Europe compared to their Eastern counterpartsuntil the end of the sample.

The entry in the row labeled p-value [700-1100) provides in column (1) the F-statistictesting the equality of the three θt values prior to 1100 CE. For columns (2) and (3) thisrow reports the p-value testing the hypothesis that each of the αt and βt values is equal tozero, respectively. Entries in the row labeled p-value [1100-1500) provide the same analysisfor after 1100 CE.

The overarching statistical pattern is clear. Prior to 1100 CE, one cannot reject thehypothesis that rulers in Western Europe and the Islamic world remained in power for thesame amount of time. After 1100 CE, however, this hypothesis can be rejected. This resultdoes not appear to be driven by decreasing ruler duration in the Islamic world (one cannotreject the hypothesis that average duration remained constant across the period). Rather,the divergence appears to be driven by increasingly durable rule in Western Europe. Nordoes it appear that this result is driven by issues related to selection in the data since resultsusing the Bosworth-Morby data set yield qualitatively similar results. Figure 2 details theevolution of political stability in non-Islamic Western Europe and the Islamic world basedon the Bosworth-Morby data. The moving average calculated with 100 lag years is graphedfor both Western Europe and the Islamic world starting in the year 650 CE.

3.2.1 The Iberian Peninsula

The Iberian Peninsula provides an interesting laboratory to further explore the differencesin ruler durability between Europe in the Islamic world while holding many variables —such as cultural legacy and geographic endowment — constant in a non-parametric manner.Muslim forces invaded the Iberian Peninsula in the year 711 CE and introduced mamlukismto the region shortly thereafter (Wasserstein 1985).

14

The historical evolution of the political equilibrium in the Iberian Peninsula providesadditional support that Christian Western Europe enjoyed political advantages over theMuslim world beginning somewhere around the year 1100 CE. In 1085 CE, Christian forcesconquered the city of Toledo, marking “the single most important advance made by Chris-tians since the Muslim conquest nearly four centuries before” (Wasserstein 1985, p. 256).Although Muslim Toledo was a large and wealthy city, it suffered “from serious internal divi-sions” (Wasserstein 1985, p. 253). Alfonso VI, the Castilian conquerer of Toledo, is reportedto have attributed his victory to this political instability. He is said to have remarked that“[t]he more rebellions and rivalry there are among them [the Muslims], the better it is forme! [...] the solution is to make them frightened of each other [...] then they will fall into myhands when they are weak [...] as has happened in Toledo [...thus I will be able to advancethe reconquest] without any bother” (cited in Wasserstein 1985, p. 257).

To investigate the empirical relevance of claims that Muslim entities were less stablethan their Christian counterparts beginning sometime in the 11th century we use the Iberiansubset of Bosworth and Morby’s data. This data set closely approximates the populationof rulers in the Iberian Penninsula over the period. We then run regression (1) using thissubsample. Results are presented in columns (9) and (10) of Table 1. Although these resultsare noiser than those obtained using the entire sample, they are consistent with the historicalevidence. Rulers in Christian Iberia remained in power for a statistically significant largerperiod after the year 1000 CE. This result is broadly consistent with our findings in thefull sample and suggests that the observed divergence in ruler duration are not driven bygeographic differences or cultural legacies.

4 Ruler Duration and Constraints on the Sovereign

The empirical results show that rulers assuming power in Christian Western Europe remainedin power for longer periods than their counterparts in the Islamic world after the year 1100CE. In Section 2 we presented a historical and theoretical narrative suggesting that theobserved divergence in ruler duration was driven by the emergence of a landed aristocracyin Europe beginning in the Carolingian period. We have argued that the emergence of thislanded aristocracy worked to generate increased constraint on the sovereign which in turnlowered the probability of overthrow.

Although data limitations do not allow us to empirically pin down the precise causalchannel leading to the observed increase in ruler duration, the data we have compiled doprovide evidence consistent with claims that this increase was driven (at least in part) bygreater constraint on the sovereign.

4.1 Princes, Parliaments and Ruler Duration

We expect that enhanced constraints on the sovereign will be reflected in increased rulerduration. Ideally, we would possess a metric measuring the constraints on each ruler; un-fortunately, such a metric is not available. To better understand the extent to which theincrease in ruler stability in Western Europe is indicative of constrained monarchs we be-gin using the metric developed by De Long and Shleifer (1993) to measure constraints on

15

the sovereign. We use their metric to create a dummy variable free that is equal to oneif De Long and Shleifer classify the period as “Free” and zero if they classify the periodas “Prince.” The distinction between “Free” and “Prince” corresponds to the distinctionbetween feudal and absolutist in their conceptualization.

We assign De Long and Shleifer’s metric to a political entity if that political entity’scentroid fell within the nine “countries” denoted by De Long and Shleifer (1993).30 Equippedwith this metric, we run the regression

durationidt = dt +1400∑

t=1000

αt · freei · dt +X ′dtγ + εidst (2)

freei varies by i because De Long and Shleifer do not use centuries in their classification;rather, they group their metric by the years 1050-1200 CE, 1200-1330 CE and 1330-1500CE. The estimated αt are reported in column (1) of Table 2. The results suggest that onthe interval 1050-1200 CE, rulers of free (i.e., feudal) political entities remained in powerlonger than those who ruled in non-free (i.e., absolutist) entities. Furthermore, the entry inthe row labeled p-value shows that we can reject at the 0.01 level the hypothesis that all theαt values are equal to zero. It should be noted, however, that after this date the coefficentsdecrease and are not statistically significant at conventional levels. This result may be dueto measurement error since constraints on the sovereign increased across Western Europeover the period examined rendering differences between free and non-free entities of lessimportance.

Although De Long and Shleifer’s metric has its virtues, this metric is only availablefor a limited geographic area and for a limited time period. To extend our analysis, weuse data on parliamentary activity provided by Van Zanden et al. (2011). Zanden et al.(2011) provide an index of parliamentary activities in Western European polities across thecenturies. We use their data to construct a dummy variable, parliamentdt, equal to one ifthe sovereign state d in the year t had at least one meeting of parliament on the interval[t,t+100). We then relate ruler duration to this metric using equation 2 replacing freei withparliamentdt. These results are presented in column (2) of Table 2 and show that this metricyields qualitatively similar results to the De Long and Shleifer metric.

We then construct a hybrid metric which is equal to 1 if parliamentdt or freei is equalto 1 and zero otherwise. Results using this metric are presented in column (3) of Table 2.Again, the results are qualitatively similar to those presented in columns (1) and (2). Incolumns (4)-(6) we gradually extend the sample and as we do so, the coefficients on our hybridvariable increase, becoming statistically significant in the entire sample. Finally, in column(7) we include country dummies (based on the present-day country in which the centroidof the political entity falls). The coefficient on our hybrid variable is large and statisticallysignificant, suggesting that an increase in constraints on the sovereign was associated withan approximately seven and a half year increase in ruler duration. This amounts to a fixed-effects test where, even after for controlling for cultural and geographic influences, executiveconstraint — as proxied by lack of an absolutist prince and/or existence of parliamentaryinstitutions — leads to substantively longer ruler duration.

30These regions include Southern Italy (which we define as Italy south of Rome), Northern Italy, Austria-Bohemia (which we define as Austria), Germany, Netherlands, Belgium, England, France and Spain.

16

Although these metrics of constraints on the sovereign are clearly imperfect, the corre-lations observed are consistent with our hypothesized link between increased constraint onthe sovereign and longer sovereign rule.

4.2 Constraints on the Sovereign and Intra-Dynasty Stability

The evolution of ruler duration within dynasties provides an additional opportunity to testif political stability is a reflection of increasing constraint on the sovereign. We begin bycreating a variable measuring the order in which a ruler falls in the dynastic chain. Thismetric is equal to one for the founder of the dynasty, two for the next ruler, etc.

If a ruler is unconstrained, we expect his time in power to be a strong function of his(sovereign-specific) human capital (Jones and Olken 2005). Moreover, we expect the founderof a dynasty to be highly endowed with such capital. Inasmuch as rule remains within agiven family we also expect sovereign human capital to mean-revert over time. If correct,this dynamic should produce a downward relationship between the place the ruler occupiesin a given dynastic chain and his time in power. On the other hand, if a ruler is constrainedas a result of feudal, or other, institutions this should mitigate the expected downwardrelationship between a sovereign’s place in the dynastic order and his duration in power. Theintuition is that when sovereigns are constrained as a result of the institutional framework,sovereign-specific human capital should matter less.

These predictions are consistent with the data linking ruler location in a dynastic chainto his duration. Figure 3 presents the non-parametric relationship (lowess smoother withbandwidth 0.8) between the sovereign’s order in his dynasty and his duration in power. Thedotted line includes the sample of Western European sovereigns before the year 1000 CE. Thesolid line includes sovereigns in the Islamic world. For both pre-1000 CE Western Europeand the Islamic world sample, we observe a downward-sloping curve that is consistent withthe hypothesis of an unconstrained sovereign and human capital mean reversion. The brokenline in Figure 3 presents the “constrained” cases in the Western European sample after theyear 1000 CE (in which the variable hybrid is equal to one). These results show that thenegative relationship between a ruler’s place in the dynasty and his duration in power largelydisappears. We interpret this to mean that as polities become institutionalized throughpowersharing, the characteristics of individual rulers mean less for their survival.

4.3 Carolingian Origins of European Political Stability

The historical record suggests increases in constraint on the sovereign originate with theCarolingians. To further investigate the origins of the observed increase in European politicalstability we limit the sample to non-Muslim Europe and use Morby’s data set to investigatethe origins of the increase in European ruler duration. We use Morby’s data because itcovers periods prior to the year 700 CE in detail and allows us to empirically investigatethe breakdate of ruler duration in non-Muslim Europe. Using this data set we consider thefollowing multiple linear regression with m breaks (m+1 regimes):

durationt = β0 + tδj + εt ; t = Tj−1 + 1, ..., Tj (3)

17

for j =1,...,m+1. Here the variable durationt is the mean value of ruler duration for rulerswho assumed power at year t. The data identify one break point in the year 790 CE.31 Thisbreak point has a 95% confidence interval comprising [755 CE, 855 CE]. The fitted valuesfrom Equation 3 are plotted in Figure 4. This break date is robust to the specificationchosen and implies a discrete jump in political stability in the year 790 CE of approximatelysix years, followed by a statistically significant change in the trend of duration over time.32

Figure 5 charts the F-statistic over time; the identified break date of 790 CE is indicatedwith a vertical line.

4.3.1 Geography and the Spread of Feudal Institutions

The breakdate of 790 CE is consistent with our narrative which emphasizes the Carolingianroots of increased constraint on the sovereign. We further examine the importance of Car-olingian institutional innovation using two additional variables. The first of these variables,%Carol, gives the percentage of a political entity’s land mass that belonged to the Carolin-gian Empire in the year 800 CE (the boundaries are taken from Nussli). The second variableis the distance of a political entity’s centroid from Aachen (which was the capital of theCarolingian Empire).

The logic of the empirical strategy employed in this section can be understood as follows:if the institutional reforms that led to the observed increase in ruler duration originatedwith the Carolingians, we would expect to see the increase in durations begin there andthen spread to the rest of Western Europe. We would also not expect to see any differencesbetween the two regions prior to the emergence of the Carolingian Empire.

To test this hypothesis, we use the following specification:

durationidt = θtdt +1400∑t=700

αt · Carolingiani · dt +X ′dtγ + εidt (4)

where all the variables are as defined above and Carolingiani is either the percent of thepolitical entity’s landmass that was occupied by the Carolingian Empire or the distance ofthe entities centroid from Aachen. Results in columns (1)-(3) of table 3 present the resultsusing the percentage of the entity’s landmass that was occupied by the Carolingian Empire.In column (1) we limit the sample to Western Europe. Here, the advantages to having beenin the Carolingian Empire peak in the year 1000 and gradually decrease (partly as the resultof convergence in Western Europe to the Carolingian levels of ruler duration). As we expandthe sample to include Eastern Europe (column (2)) and the Islamic world (column(3)) rulerdurations are longer in the former Carolingian lands until the end of the sample.

Results using distance to Aachen are presented in columns (4) and (5) for the non-Islamic sample and the entire sample (distance from Aachen is not significant in the Western

31We use the theoretical framework developed by Bai (1997, 1997b, 1999) and Bai and Perron (1998,2003). We use the BIC criteria to select the number of breaks using 15% trimming and setting the maximumnumber of breaks equal to 5.

32The data rarely identified more than one breakpoint in our experimentation with alternative specifica-tions. One exception occurred when we fitted a simple step function to the data, in which case the breakdates790 CE [734,852] and 1055 CE [963,1104] were identified (95% confidence intervals in brackets). Since theR2 was similar in both specifications we decided to retain the more parsimonious of the two specifications.

18

European sample). Here the results suggest a slightly different picture, with systematicdifferences not appearing until the year 1100 CE.

Although it should again be stressed data limitations do not allow us to pinpoint theextent to which Carolingian reforms generated the increase in ruler duration and constrainton the sovereign in Western Europe, the empirical evidence provided in this section is largelyconsistent with this hypothesis.

5 Potential Confounding Factors

Thus far, we have sought to document that rulers in Western Europe remained in power forlonger than their Muslim counterparts after the year 1100 CE. Through the inclusion of avariety of control variables, we have shown that these differences are not a function of statesize or geography. Instead, we have argued that the results reflect increased constraints onthe sovereign in Western Europe. In this section we investigate the extent to which alterna-tive narratives for the observed divergence in ruler duration are consistent with the empiricalevidence.

European primogeniture: If European rulers increasingly passed political power to theireldest sons through norms of primogeniture, this might explain why Christian monarchssurvived in office longer than their Muslim counterparts. This would particularly be thecase if their counterparts in the Islamic world preferred to pass power to siblings or otherrelatives. Although primogeniture clearly has its roots in medieval European, the practice ofpassing power and wealth to the first-born son only emerged in the 13th century and spreadacross Europe up to the 17th century (Bertocchi 2006), well after the break in the politicaltrend identified in the data on ruler duration. For example, disputes regarding succession inScottish and Burgundian monarchies implies that primogeniture was still being establishedas an institutionalized practice in the 13th and 14th centuries. This suggests that primogeni-ture may have emerged endogenously from changes in political stability since longer servingmonarchs would be more likely to seek to transfer power to their children versus other rel-atives. It is further clear that Islamic monarchs were passing on power to their sons asevidenced by the emergence of the vizier who would frequently serve as decisionmaker in theplace of an underage Muslim ruler.33

Differential life expectancy: One reason why rulers survived in power longer in Christian Eu-rope compared to the Islamic world could be that life expectancy in Europe was increasingrelative to trends in Muslim polities. There is no evidence to suggest this was the case. Mus-lim physicians during the medieval period were highly sophisticated, discrediting theories ofhumorism which were commonly held in Europe during this time.34 Muslim doctors set upsome of the earliest dedicated hospitals, medical schools and made tremendous advances in

33To further test this possibility, we have gathered data on the relation of each sovereign to the sovereignthat proceeded him. One cannot reject the hypothesis that the ruler’s son was as likely to succeed him inthe Islamic world as in Christian Europe (results available upon request).

34Humorism contends that the human body is composed of four basic humors — black bile, yellow bile,phlegm and blood.

19

the field of pharmacology. Borsch (2005) shows that the disease environments of the East andWest were quite similar during this period with the bubonic plague, for example, impactingboth Muslim and Christian cities. In England, medieval life expectancy at the age of 25 forthe higher ranks of English society (i.e., those who inherited land) was between 21 and 24years (Jonker 2003). Estimated life expectancy at the age of 25 for monks of Christ Churchpriory in Canterbury during the 15th century was between 21 and 29 years; monks at thistime tended to be relatively well-fed, with better medical care than the general population(Hatcher 1986). Studies of life expectancy of influential religious scholars in 11th centuryMuslim Spain show that the most prominent within this occupational group lived between69 and 75 years, on average (Shatzmiller 1994, p. 66).

Geographic size: Stasavage (2010, p. 625) offers one possible explanation for why smallEuropean polities “were able to survive despite threats from much larger neighbors.” Hefinds that geographically compact polities — with lower exogenous monitoring costs —could maintain representative parliamentary institutions to a greater extent; an implicationof this finding is that smaller polities, as appear to have been more common in Europe,enjoyed better institutions and were, thus, more likely to enjoy stability. Although state sizeis endogenous, we have included it as a control variable to show that differential state sizesdo not seem to drive our result. Indeed, the historical evidence presented above suggeststhat the small sizes of states may have been driven by the emergence of a landed aristocracyacross Western Europe.

To further investigate the the determinants of state size, we limit the sample to the years1300-1400 CE. We use this sub-sample because before this date, the maps created by Nussli(2011) attribute many areas to the Holy Roman Empire that appear to have been sovereignstates. Thus, during the period 1000-1200 CE, we believe that Nussli’s maps generallyoverestimate the size of the average European state. Using the data after 1300 CE, weestimate the following regression:

stdsizedt = θtdt + βCarolingiandt + εdt (5)

where stdsizedt is the standardized area of state d in year t (i.e., subtract the mean anddivide by the standard deviation of the area for all states in year t). The results of equation(5) are presented in Table 4. The first two columns limit the sample to Western Europe,while the last two columns use the entire sample. Results in columns (1) and (3) show thatgoing from zero percent of a state’s land mass belonging to the Carolingian Empire in 800 CEto one-hundred percent of a state’s land mass belonging to the Carolingians was associatedwith between an 0.18 and 0.35 standard deviation decrease in that entity’s land area in theyear 1300 CE and 1400 CE. The distance of a state’s centroid from Aachen yields similarresults. In sum, although a topic for future research, the evidence is consistent with theclaim that the unique political equilibrium in Europe led to small state sizes (and not theother way around).

Germanic tribes effect: One cultural explanation for why Europe developed greater degreesof executive constraint (and durable leadership) relates to the impact of proto-democratictribal institutions common among the Germanic tribes of Northwestern Europe. The most

20

significant geographic area that enjoyed both the cultural heritage of the Germanic tribesand Islamic sultanate is the Iberian peninsula. As we have shown in Section 3.2.1, this regionexhibits roughly the same trends as the broader Islamic world. This result casts doubt onthis explanation.

Muslim economic decline: Although an influential literature links economic outcomes to in-stitutional developments, it is possible that the causality runs the other way. In other words,a plausible alternative explanation is that Muslim economic decline led to shorter ruler du-ration. Determining the historical point at which the Islamic world began to fall behind theWest and the intensity with which it declined over time is difficult. Historians often cite theMongol invasion of the 13th century as the end date for the “Golden Age” of Islam. Othersargue that the divergence took place later, beginning in the 17th century (Kuran 1997) andaccelerating with the industrial revolution in Europe. Proponents of this perspective pointto the affluence of the Mughul and Ottoman empires which each produced architecture ona scale which suggested that rulers had the capacity to extract revenue in support of majorartistic endeavors (Dale 2010, pp. 130-131). There is no study that we are aware of thatsuggests Muslim economic decline began prior to the 11th century.

The Mongol invasion: A common narrative for why Islamic polities became less stable overtime relates to the Mongol conquests of the 13th century. The divergence in political stabilitybetween Christian Europe and the Islamic world predates the Mongol invasions, however.Nor can the success of the Mongols be considered strictly exogenous as weaker, less stablepolities may also be more subject to successful foreign invasion. In any case, the data do notreject the hypothesis that ruler durations remained the same in the Islamic world throughoutthe 800 years covered by the data.

6 Conclusion

The Empire of Charlemagne was the critical point of the rupture [...] of theEuropean equilibrium (Pirenne 1980 [1939], p. 234).

Western Europe was considered an economic backwater in 1000 CE; indeed, the biggestcities in the Muslim world at this time were larger then any European city would be untilthe seventeenth century (De Long and Shleifer 1993). By 1000 CE, however, the fortunesof European political leaders were already improving when compared to their peers in theIslamic world in ways that were probably imperceptible to rulers, elites and citizens of thosesocieties. This study provides the first empirical evidence that Europe’s medieval institu-tional arrangements were rooted in the Carolingian Empire. We find that political stabilityin Western Europe underwent a “structural break” in the year 790 CE — the midpoint ofthe reign of Charlemagne, a pioneer in the introduction of feudal social institutions. By theyear 1100 CE the average Western European ruler’s tenure was significantly longer than hiscounterpart’s reign in the Muslim world. This result provides the first empirical support (tothe best of our knowledge) for a distinguished line of scholarship that stresses the emergenceof feudalism as an important early step in the economic and political rise of Western Europe.

21

Although data limitations do not allow us to pinpoint the exact causal mechanism gen-erating the increase in Western European political stability, the results are consistent witha literature stressing the importance of economic and political shocks following the collapseof the Western Roman Empire in empowering a landed aristocracy. This literature stressesthat the landed aristocracy slowly gained power during centuries of economic downturn.Eventually, this aristocracy was able to place unusual constraints on the sovereign. Theseconstraints prepared the way for the emergence of parliaments and medieval Europe’s uniqueinstitutional framework. We find empirical evidence consistent with the hypothesis that Eu-ropean sovereigns faced increasing checks on their power after the introduction of feudalinstitutions.

The growth in both stability and prosperity of Western Europe closely paralleled the de-cline in the economic fortunes of the Islamic world. While both medieval European monar-chies and Islamic dynasties cultivated the types of personalistic ties typical of North etal.’s “natural state” (2009), we have argued that the interdependent military, political andeconomic relationships that developed in Europe under feudalism laid the basis for more im-personal forms of political organization down the line, including institutionalized executiveconstraint. North et al. (2009) offer some ideas for how to go from a “natural state” — likethe type of state that existed in both the medieval Islamic and Christian worlds — to anopen access order — or a state characterized by limits on violence and institutions whicheffectively constrain abuses of power.35 Muslim rulers, unlike their European counterparts,had the administrative and financial capacity to import slaves from outside of their realmsto provide military services; Muslim rulers were not, however, able to effectively disciplinethis military force through non-military means. European rulers found themselves forcedto pay their militaries through land grants, a process which eventually created a powerful,landed and independent military class. In this sense, Poggi’s observation that “the ‘feudalstate’ is one that undermines itself” (1978, p. 26) is correct; medieval kings, operating froma position of financial weakness and limited state capacity, had no choice but to offer fiefs aspayment to elites who provided rulers with military support. Feudalism led this emergent“warrior class” to be “rooted in the land” (Poggi 1978, p. 32) in a way that was very distinctfrom the nature of military recruitment and remuneration in the Islamic world. The landednobility in Europe were able to eventually extract both concessions and protections fromthe state, leading to the rise of medieval parliaments and the types of institutions that arebelieved to be growth-inducing.

This suggests that the Muslim world fell behind because of the inability of Muslim sultansto be credibly constrained. This explanation is distinct from, and complementary to, recentwork which has argued that Islamic institutions, like Muslim inheritance laws and charitableendowments, played a crucial role in the region’s economic underdevelopment (Kuran 2004;Kuran 2009; Kuran 2010a; Kuran 2010b) as well as a focus on the collectivist nature of“Eastern” societies and the negative externalities associated with informal monitoring andpunishment mechanisms (Greif 1994). According to our account, by the time of the NewWorld discoveries European rulers were already uniquely constrained compared to their Mus-lim counterparts. Although trade and colonialism may have enabled both good (Acemoglu et

35In the North et al. (2009, p. 170) account, the way that society limits and controls violence is animportant “doorstep condition” to the development of growth-producing institutions.

22

al. 2005) and bad (Drelichman and Voth 2008) institutional change in European countries,these conclusions suggest that the uniquely European emergence of checks on the sovereignpredated the discovery of the Americas.

References

Acemoglu, D. and J. A. Robinson. 2000. “Why Did the West Extend the Franchise?Democracy, Inequality, and Growth in Historical Perspective.” Quarterly Journal of Eco-nomics, 115(4) 1167-1199.Acemoglu, D. and J. A. Robinson. 2001. “A Theory of Political Transitions.” AmericanEconomic Review, 91(4) 938-963.Acemoglu, D., S. Johnson and J. A. Robinson. 2005. “The Rise of Europe: AtlanticTrade, Institutional Change, and Economic Growth.” American Economic Review, 95(3)546-579.Acemoglu, D. and J. A. Robinson. 2006. Economic Origins of Dictatorship and Democ-racy. New York: Cambridge University Press.Acemoglu, D. and J. Robinson. nd. Why Nations Fail. Book manuscript.Alesina, A., S. Ozler, N. Roubini and P. Swagel.1996. “Political Instability andEconomic Growth.” Journal of Economic Growth1(2): 189-211.Anderson, P. 1979(1974) Lineages of the Absolutist State. London: Verso.Ashtor, E. 1970. “Quelques Observations d’un Orientaliste sur la These de Pirenne.”Journal of the Economic and Social History of the Orient 13(2): 166-194.Ayalon, D. 1994a. The Military Reforms of Caliph al-Mu‘tasim: Their Background andConsequences. Islam and the Abode of War: Military Slaves and Islamic Adversaries. Alder-shot: Variorum.Ayalon, D. 1994b. Mamluk: Military Slavery in Egypt and Syria. Islam and the Abode ofWar: Military Slaves and Islamic Adversaries. Aldershot: Variorum.Ayalon, D. 1994c. Mamluk Military Aristocracy — A Non-Hereditary Nobility. Islam andthe Abode of War: Military Slaves and Islamic Adversaries. Aldershot: Variorum.Ayalon, D. 1994d. Preliminary Remarks. Islam and the Abode of War: Military Slavesand Islamic Adversaries. Aldershot: Variorum.Bacon, F.1819. The Works of Francis Bacon, Volume II. Baldwin: LondonBai, J. 1997a. “Estimation of a Change Point in Multiple Regression Models.” Review ofEconomic and Statistics 79: 551-563.Bai, J. 1997b. “Estimating Multiple Breaks One at a Time.” Econometric Theory 13:315-352.Bai, J. 1999. “Likelihood Ratio Tests for Multiple Structural Changes.” Journal of Econo-metrics 91: 299-323.Bai, J. and P. Perron 1998. “Estimating and Testing Linear Models with MultipleStructural Changes.” Econometrica 66: 47-78.Bai, J. and P. Perron 2003. “Computation and Analysis of Multiple Structural ChangeModels.” Journal of Applied Econometrics 18(1): 1-22.Bertocchi, G.2006. “The Law of Primogeniture and the Transition from Landed Aristoc-racy to Industrial Democracy.” Journal of Economic Growth11(1): 43-70.

23

Bienen, H. and N. van de Walle1991. Of Time and Power: Leadership Duration in theModern World. Stanford, CA: Stanford University Press.Bisson, T. N. 1966. “The Military Origins of Medieval Representation.” American Histor-ical Review 71(4): 1199-1218.Bisson, T. N. 1977. “The Organized Peace in Southern France and Catalonia.” AmericanHistorical Review 82(2): 290-311.Bisson, T.N. 1994. “The “Feudal Revolution”.” Past & Present 142: 6-42.Borsch, S. 2005. The Black Death in Egypt and England: A Comparative Study. Austin,TX: University of Texas Press.Bosworth, C. 1996. Islamic Dynasties. New York: Columbia University Press.Brown, E. 1974. “The Tyranny of a Construct: Feudalism and Historians of MedievalEurope.” American Historical Review 79(4): 1063-1088.Conrad, L., M. Neve, V. Nutton, R. Porter and A. Wear 1995. The WesternMedical Tradition: 800 B.C. - 1800 A.D. Cambridge: Cambridge University Press.Crone, P. 2006. Imperial Trauma: The Case of the Arabs. Common Knowledge. 12(1).Crone, P. 2004. God’s Rule: Government and Islam. New York: Columbia UniversityPress.Crone, P. 2003. Slaves on Horses: The Evolution of the Islamic Polity. Cambridge:Cambridge University Press.Crone, P. 1999. The Early Islamic World. War and Society in the Ancient and MedievalWorlds. Editors, K. Raaflaub and N. Rosenstein. Cambridge: Harvard University Press.Delong, J. B. and A. Shleifer 1993. “Princes and Merchants: European City Growthbefore the Industrial Revolution.” Journal of Law and Economics 36(2): 671-702.Dennett, D. 1948. “Pirenne and Muhammad.” Speculum 23(2): 165-190.Downing, B. 1989. “Medieval Origins of Constitutional Government in the West.” Theoryand Society 18(2): 213-247.Downing, B. 1992. The Military Revolution and Political Change: Origins of Democracyand Autocracy in Early Modern Europe. Princeton: Princeton University Press.Drelichman, M. and H. Voth 2008. “Institutions and the Resource Curse in EarlyModern Spain.” in Helpman ed. Institutions and Economic Performance: 120-147.Ehrenkreutz, A. 1972. “Another Orientalist’s Remarks Concerning the Pirenne Thesis”Journal of the Economic and Social History of the Orient, 15(1/2): 94-104.Ganshof, F. 1968. Frankish Institutions under Charlemagne. Providence: Brown Univer-sity Press.Goemans, H., K. Gleditsch and G. Chiozza2009. Introducing Archigos: A Dataset ofPolitical Leaders. Journal of Peace Research. 46(2): 269-283.Greif, A. 1994. “Cultural Beliefs and the Organization of Society”. Journal of PoliticalEconomy, 102(5): 912-950.Hatcher, J. 1986. “Mortality in the fifteenth Century: Some new Evidence.” EconomicHistory Review, 39 : 19-38.Hintze, F. 1975. The Preconditions of Representative Government in: “The HistoricalEssays of Otto Hintze” New York: Oxford University Press.Holt, J.C.1992. Magna Carta. Cambridge: Cambridge University Press.Irwin, R. 1986. The Middle East in the Middle Ages: The Early Mamluk Sultanate 1250-1382. London: Croom Helm.

24

Issawi, C. 1980. “Europe, the Middle East and the Shift in Power: Reflections on a Themeby Marshall Hodgson.” Comparative Studies in Society and History, 22(4): 487-504.Jones, B. and B. Olken 2005. “Do Leaders Matter? National Leadership and GrowthSince World War II.” Quarterly Journal of Economics, 120(3): 835-864.Jones, E. 2003. The European Miracle. New York: Cambridge University Press.Jonker, M. 2003. “Estimation of Life Expectancy in the Middle Ages.” Journal of theRoyal Statistical Society 166(1): 105-117.Kennedy, H. 2004. The Prophet and the Age of the Caliphates: The Islamic Near Eastfrom the Sixth to the Eleventh Century (2nd edition). London: Pearson Longman.Kuran, T. 1997. Islam and Underdevelopment: An Old Puzzle Revisited. Journal ofInstitutional and Theoretical Economics. 153.Kuran, T. 2004. Why the Middle East Is Economically Underdeveloped: Historical Mech-anisms of Institutional Stagnation. Journal of Economic Perspectives. 18.Kuran, T. 2009. Explaining the Economic Trajectories of Civilizations: The SystemicApproach. Journal of Economic Behavior and Organization. 71.Kuran, T. 2010a. The Long Divergence: How Islamic Law Held Back the Middle East.Princeton: Princeton University Press.Kuran, T. 2010b. “The Scale of Entrepreneurship in Middle Eastern History: InhibitiveRoles of Islamic Institutions.” The Invention of Enterprise: Entrepreneurship from AncientMesopotamia to Modern Times. Editors William Baumol, David Landes, and Joel Mokyr.Princeton: Princeton University Press.Macchiavelli, N. 1903[1532]. The Prince. Translated by L. Ricci. London: Grant Richards.Marongiu, A. 1968. Medieval Parliaments: A Comparative Study. London: Eyre & Spot-tiswoode.Mitterauer, M. 2010. Why Europe: the Medieval Origins of Its Special Path. Chicago:The University of Chicago Press.Montesquieu, C. 1989. The Spirit of the Laws. Cambridge: Cambridge University Press.Morby, J. 1989. Dynasties of the World: A Chronological and Genealogical Handbook. NewYork: Oxford University Press.Myers, A. 1975. Parliaments and Estates in Europe to 1789. London: Thames and Hudson.North, D. and R. Thomas. 1973. The Rise of the Western World: A New EconomicHistory. New York: Cambridge University Press.North, D. and B. Weingast 1989. “Constitutions and Commitment: the Evolution of In-stitutions Governing Public Choice in Seventeenth-Century England.” Journal of EconomicHistory, 49(4): 803-832.North, D., J. Wallis and B. Weingast. 2009. Violence and Social Orders: A ConceptualFramework for Interpreting Recorded Human History. Cambridge: Cambridge UniversityPress.Nussli, C. 2011. http://www.euratlas.com/about.html.Pipes, D. 1981. Slave Soldiers and Islam: The Genesis of a Military System. New Haven:Yale University Press.Pirenne, H. 1980 [1939]. Mohammed and Charlemagne. Ottowa: Barnes & Noble Books.Pirenne, H. 1946. Medieval Cities: their Origins and the Revival of Trade. Princeton:Princeton University Press.

25

Poggi, G. 1978. The Development of the Modern State: A Sociological Perspective. Stan-ford: Stanford University Press.Pomeranz, K. 2000. The Great Divergence: China, Europe and the Making of the ModernWorld Economy. Princeton: Princeton University Press.Reynolds, S. 1996. Fiefs and Vassals. Oxford: Clarendon Press.Riising, A. 1952. “The Fate of Henri Pienne’s Theses on the Consequences of IslamicExpansion.” Classica et Medievalia, 13 : 87-130.Shatzmiller, M. 1996. Labour in the Medieval Islamic World. Leiden: E. J. Brill.Stasavage, D.2010. “When Distance Mattered: Geographic Scale and the Development ofEuropean Representative Assemblies.” American Political Science Review, 104(4): 625-643.Strayer, J. 1970. On the Medieval Origins of the Modern State. Princeton: PrincetonUniversity Press.Wasserstein, D. 1985. The Rise and Fall of the Party-Kings: Politics and Society inIslamic Spain 1002-1086. Princeton: Princeton University Press.Weber, M. 1978. Economy and Society: An Outline of Interpretive Sociology. Berkeley:University of California Press.White, L. 1962. Medieval Technology and Social Change. Oxford: Oxford University Press.Wickham, C. 1984. “The Other Transition: From the Ancient World to Feudalism.” Pastand Present, 103 : 3-36.van Zanden, J.n.d. “Economic Growth in a Period of Political Fragmentation, WesternEurope 900-1300.” Mimeo.van Zanden, J., E. Buringh and M. Bosker. 2010. “The Rise and Decline of EuropeanParliaments, 1188-1789.” Mimeo.

26

Figure 1: Comparison of institutional development in Christian Europe (left) versus theIslamic world (right) following the decline of the Roman Empire.

27

Figure 2: Divergence in Non-Muslim European and Islamic World political stability from650-1500 CE.

28

Figure 3: Non-parametric plot of mean ruler duration within dynasties in Non-Muslim West-ern Europe and the Islamic World.

29

Figure 4: Political stability in Non-Muslim Western Europe. Graph details non-parametricrelationship between ruler duration and common-era years. Vertical line marks estimatedbreak date in 790 CE.

30

Figure 5: Identifying the break date in political stability for Non-Muslim Western Europe.

31

Tab

le1:

RulerDura

tionsand

theRiseofEuro

pe

Len

gth

ofR

ule

(Yea

rs)

Isla

mW

E-I

slam

EE

-Isl

amp-v

alue

Isla

mW

E-I

slam

EE

-Isl

amp-v

alue

Isla

mW

E-I

slam

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

[700,8

00)

10.5

62.

001.

440.

8013

.54

-0.8

6-1

.17

0.89

22.0

0-1

0.67

(6.2

4)(6

.39)

(6.5

0)

(6.0

7)

(6.1

5)

(6.2

6)

(6.0

6)

(7.2

3)

[800,9

00)

15.5

5-1

.89

-1.4

60.

8717

.32

-3.3

7-3

.06

0.91

22.5

0.25

(4.1

3)(4

.39)

(4.6

4)

(3.6

9)

(3.8

4)

(4.1

6)

(6.3

7)

(7.4

1)

[900,1

000)

12.8

90.

916.

450.

0013

.75

0.11

5.71

0.01

32.3

3-1

4.54

(2.9

5)(3

.39)

(3.1

0)

(3.1

7)

(3.5

1)

(3.2

8)

(8.2

6)

(8.6

9)

[100

0,11

00)

11.7

32.

57-0

.04

0.24

12.3

02.

140.

050.

3610

.97

9.62

(3.4

4)(3

.73)

(3.8

7)

(3.5

5)

(3.7

5)

(3.8

7)

(1.2

2)

(2.8

3)

[110

0,12

00)

10.8

66.

550.

230.

0111

.07

6.56

0.16

0.01

12.0

018

.58

(1.3

3)(2

.60)

(1.7

3)

(1.5

5)

(2.6

5)

(1.7

4)

(3.7

0)

(5.3

6)

[120

0,13

00)

11.5

66.

431.

170.

0212

.04

6.25

0.91

0.02

12.0

010

.65

(2.8

2)(3

.39)

(3.1

2)

(2.9

8)

(3.4

1)

(3.1

5)

(4.8

3)

(6.0

3)

[130

0,14

00)

8.98

8.95

6.73

0.25

9.47

8.44

6.30

0.27

9.64

11.4

6(1

.25)

(1.8

8)

(1.8

1)

(1.2

7)

(1.9

2)

(1.8

5)

(2.6

6)

(4.2

0)

[140

0,15

00)

7.76

11.1

75.

340.

038.

3010

.61

4.85

0.03

4.94

23.6

1(1

.34)

(1.6

9)

(2.8

3)

(1.4

5)

(1.8

2)

(2.8

8)

(1.2

6)

(4.0

2)

N27

2327

2327

2327

2327

2327

2327

2327

2327

827

8

GeographyControls?

No

No

No

No

Yes

Yes

Yes

Yes

No

No

Sample

All

All

All

All

All

All

All

All

Iber

iaIb

eria

p−value[

700−

1100

)0.

440.

740.

080.

330.

740.

080.

080.

00

p−value[

1100−

1500

)0.

480.

000.

000.

600.

000.

010.

100.

00

Not

es:

stan

dar

der

rors

clu

ster

edby

pol

itic

alen

tity

inp

are

nth

eses

.G

eogra

phy

contr

ols

incl

ude

the

are

aof

the

poli

tica

len

tity

an

dth

eav

erage

agri

cult

ura

lsu

itab

ilit

yof

the

enti

ty.

See

text

for

det

ails

.

32

Tab

le2:

Constra

ints

on

theSovereign

and

RulerDura

tion

Len

gth

ofR

ule

(Yea

rs)

Fre

eP

arl

Hybri

dH

ybri

dH

ybri

dH

ybri

dH

ybri

d(1

)(2

)(3

)(4

)(5

)(6

)(7

)

[100

0,11

00)

13.5

913

.59

13.8

414

.45

14.5

6(4

.59)

(1.7

3)

(4.4

4)

(4.3

4)

(4.3

2)

[110

0,12

00)

11.4

510

.97

11.4

515

.49

18.0

218

.49

(2.1

0)(2

.14)

(1.7

3)

(2.2

2)

(1.3

6)

(1.1

7)

[120

0,13

00)

1.55

7.25

6.25

8.71

8.56

8.94

(4.4

3)(4

.58)

(4.4

7)

(3.7

9)

(3.2

1)

(3.1

9)

[130

0,14

00)

-3.0

13.

05-0

.01

4.65

5.23

6.98

(2.3

3)(2

.16)

(2.6

5)

(0.2

5)

(1.7

2)

(1.6

3)

[140

0,15

00)

-0.5

6-0

.04

-2.5

0-0

.26

5.94

7.83

(2.9

4)(2

.88)

(2.5

2)

(1.9

4)

(2.5

7)

(2.0

7)

Hybrid

7.47

(1.4

5)

N63

563

563

511

7222

2527

2325

53

Sample

D&

SD

&S

D&

SW

EN

on-I

slam

All

All

CenturyDummies?

Yes

Yes

Yes

Yes

Yes

Yes

Yes

CountryDummies?

No

No

No

No

No

No

Yes

p−value

0.00

0.00

0.00

0.00

0.00

0.00

Not

es:

stan

dar

der

rors

clu

ster

edby

pol

itic

alen

tity

inp

are

nth

eses

.S

eete

xt

for

det

ails

.

33

Table 3: The Carolingian Origins of European Political Stability

Length of Rule (Years)

%Carol %Carol %Carol Aachen Aachen(1) (2) (3) (4) (5)

[700, 800) -1.77 -2.70 -2.64 -0.47 -0.55(3.56) (2.52) (2.81) (1.36) (1.02)

[800, 900) 8.83 2.00 3.70 1.15 0.39(4.03) (2.29) (1.47) (1.09) (1.05)

[900, 1000) 10.41 6.95 7.01 1.50 0.74(4.22) (3.91) (3.80) (0.93) (0.93)

[1000, 1100) 14.51 12.88 12.71 -0.66 -0.32(1.96) (1.93) (1.69) (1.29) (0.81)

[1100, 1200) 12.80 13.16 15.10 -2.71 -2.02(4.83) (5.19) (4.52) (0.99) (0.63)

[1200, 1300) 5.75 5.63 7.27 -2.60 -2.02(4.40) (4.01) (3.37) (0.76) (0.81)

[1300, 1400) -1.78 2.47 4.42 -2.99 -3.17(2.92) (2.05) (1.96) (0.86) (0.66)

[1400, 1500) -5.22 4.83 6.76 -2.83 -3.32(3.85) (2.68) (2.16) (1.14) (0.75)

N 1172 2225 2723 2225 2723

Geography Controls? Yes Yes Yes Yes Yes

Sample Western Europe Non-Islam All Non-Islam All

Notes: standard errors clustered by political entity in parentheses. Geography controls include the areaof the political entity and the average agricultural suitability of entity.See text for details.

Table 4: State SizeStandardized Area of Political Entity

(1) (2) (3) (4)

%Carol -0.18 -0.35(0.08) (0.13)

Aachen/1000 0.11 0.23(0.05) (0.11)

N 160 160 301 301

Sample WE WE All All

Notes: standard errors clustered by political entity in parentheses.See text for details.

34