JHA TradeInstitutionsEthnic 2013

Trade, Institutions, and Ethnic Tolerance: Evidence from South Asia
Author(s): SAUMITRA JHA

Source: The American Political Science Review , November 2013, Vol. 107, No. 4
(November 2013), pp. 806-832
Published by: American Political Science Association
Stable URL: https://www.jstor.org/stable/43654035
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
American Political Science Association is collaborating with JSTOR to digitize, preserve and
extend access to The American Political Science Review
This content downloaded from

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
All use subject to https://about.jstor.org/terms
American Political Science Review Vol. 107, No. 4 November 2013
doi:10.1017/S0003055413000464 © American Political Science Association 2013
Trade, Institutions, and Ethnic Tolerance: Evidence from

SAUMITRA JHA Stanford Graduate School of Business
/provide mentary,/provide mentary, nonreplicable services and a mechanism to share the gains from exchange has resulted in
a sustained
a sustained legacy
evidence nonreplicable
of ethniclegacytolerance
that of ethnicin theSouth
servicesAsian
degree tolerance
towns.andDue
to a to
whichMuslim-specific
in mechanism South medieval
advantages
Asian to share
in Hindus towns. the Due and gains to Muslims from Muslim-specific exchange could provide has advantages resulted comple- in in
Indian Ocean shipping, interethnic complementarities were strongest in medieval trading ports , leading
to the development of institutional mechanisms that further supported interethnic exchange. Using novel
town-level data spanning South Asia's medieval and colonial history, I find that medieval ports, despite
being more ethnically mixed, were five times less prone to Hindu-Muslim riots between 1850 and 1950,
two centuries after Europeans disrupted Muslim overseas trade dominance, and remained half as prone
between 1950 and 1995. Household-level evidence suggests that these differences reflect local institutions
that emerged to support interethnic medieval trade, continue to influence modern occupational choices
and organizations, and substitute for State political incentives in supporting interethnic trust.
INTRODUCTION the mechanisms through which trade can foster lasting

peace in poor, ethnically diverse societies?
In this article, I draw upon a simple but robust the-
that a basic relationship exists between com- oretical framework to emphasize the importance of
For merce thatmerce
centuries, a andasbasic
and peace, peace, relationship
the potential loss of tradepolitical as the economists potential exists between have loss of argued trade com- two conditions for providing the basis for peaceful co-
makes violence more costly. For example, the Baron existence over long time horizons in ethnically diverse
de Montesquieu (1748) [Book XX, p. 1] proposed that societies: the presence of a nonreplicable and nonex-
"Commerce is a cure for the most destructive prej- propriable source of interethnic complementarity and
udices; for it is almost a general rule that wherever access to a nonviolent mechanism to redistribute or
the ways of man are gentle there is commerce; and share the gains from trade between groups.3 Satisfying
wherever there is commerce, there the ways of men these conditions further fosters the development of "in-
are gentle."1 Yet, from the ethnic Chinese in Indone-
stitutions" that reinforce incentives for peace, which I
sia to South Asians in East Africa to Jews in Europe, define to be "systems of complementary norms, beliefs
historical and contemporary examples abound of even and organizations."4 In contrast, societies that violate
the most commercially oriented minority groups be- these conditions- i.e., where ethnic groups compete,
coming repeated targets of ethnic violence and ex-
where the source of one group's complementarity can
propriation (Benbassa and Rodrigue 2000; Chua 2003;
be violently seized (e.g., physical capital), easily repli-
Jha 2007; Landa 1994). Some have argued that contem- cated (e.g., low skilled human capital), or that lack an
porary processes of globalization and democratization,
effective nonviolent mechanism for sharing the gains
in particular, have exacerbated ethnic conflict in the
from trade (like many commercially oriented trading
developing world.2 What, then, are the conditions and
communities)- are likely to be more prone to ethnic
conflict and less likely to develop reinforcing institu-
Saumitra Jha is Assistant Professor of Political Economy at the tions supportive of ethnic tolerance.
Stanford Graduate School of Business, 655 Knight Way, Stanford
CA 94305 (saumitra@gsb.stanford.edu). I substantiate these claims by exploiting evidence
drawn from Hindu-Muslim interaction in South Asia.
I owe particular thanks to Susan Athey and Avner Greif, as well as
Aprajit Mahajan, Kenneth Arrow and David Laitin. I am grateful to South Asia is particularly appropriate for this exercise,
Prashant Bharadwaj, Asim Khwaja, Atif Mian and Steven Wilkin- not only because it houses more than a fifth of the
son for very helpful comments and generously sharing their data
world's population and close to half of its poor, but
and for the suggestions of Ran Abramitsky, Amrita Ahuja, Roger
Bolton, Feysal Devji, Claudia Goldin, Emeric Henry, Michael Kre- its Muslim population, though a minority, is still the
mer, Prakash Kannan, Kimuli Kasara, Anjini Kochar, Timur Kuran, second largest in the world, and Hindus and Muslims
Jenny Kuan, Jessica Leino, Pedro Miranda, Monika Nalepa, Ro- have engaged in 13 centuries of economic, political and,
llini Pande, Matthias Schuendeln, TN Srinivasan, Yannay Spitzer, sadly too often, violent, interaction.5
Ashutosh Varshney, Nico Vöigtlander, Gavin Wright, Joanne Yoong
and seminar participants at Berkeley, Chicago, Clio, Davis, Harvard,
I exploit the fact that from at least the 8th century
Northwestern, Stanford, LiCEP, LSE, MIT, Princeton, NEUDC, to the 17th century, due to the coordination of over-
NYU, Rochester, UBC, USC, UWO, the all-UC economic history seas markets through Muslim pilgrimages such as the
group, and the World Bank. Sangick Jeon and Astasia Myers pro-
vided much valued research assistance.
1 In his Orations (III), the Greek philosopher Libanius (ca. 314AD)
3 Technically, two actions are complements if (1) adopting one does
was an even earlier proponent. See also Hirschman (1977) and Po-not preclude adopting the other, and (2) whenever it is possible to
lachek and Seiglie (2007) on trade between nations. implement them separately, the sum of each return cannot be greater
1 For example, in an influential book, Amy Chua (2003) [p. than 16] doing them together.
argues: "In the numerous countries around the world that have 4 This definition adapts that used by Greif (2005) to draw upon the
pervasive poverty and a market-dominant minority ... the combinedpredictions of robust comparative statics (e.g., Milgrom and Roberts
1990).
pursuit of free markets and democratization has repeatedly catalyzed
ethnic conflict in highly predictable ways, with catastrophic conse-
J Hindu-Muslim violence has resulted in more than 40,000 deaths or
quences, including genocidal violence and the subversion of markets
injuries since 1947, the overwhelming majority in towns and cities
and democracy themselves." See also Horowitz (2003). (Varshney and Wilkinson 2004).
806

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
American Political Science Review Vol. 107, No. 4
Hajj , Muslims enjoyed exogenous Porbandar in 2006-07,complementary

and a 2005 urban sample of and
nonreplicable advantages householdin levelaccessing
data on trust, conflict, and local or- Ocean
Indian
trade routes. Further, and ganization (the Indianmany
unlike Human Development
other Survey
trading
communities that have often found themselves the tar- 2009). I use these sources to rule out a number of key
gets of violence, such as the Chinese in Indonesia, oralternative explanations and to delineate the nature
the South Asians in East Africa, barriers to entry into
of the institutions that emerged to support interethnic
trade were relatively low. The fact that trade was coor- tolerance.
dinated primarily by pilgrimage routes rather than by I first provide evidence that the differences between
clan ties meant that entry into Indian Ocean tradingerstwhile ports and other towns are not the result of
networks was relatively easy for any Muslim, and the factors, such as a more mixed ethnic composition or
resultant competition between Muslim groups provided poverty, that are often suggested to explain ethnic vi-
a natural, decentralized, and large-scale mechanism to olence. Instead, I show that medieval ports are more
share the gains from trade. ethnically mixed and poorer than similar towns, both
I provide evidence that medieval overseas trad-factois that are commonly associated with greater vi-
ing ports, which served as the geographical focusesolence. In fact, a medieval trade legacy appears to re-
of these exogenous Hindu-Muslim complementarities duce violence the most in larger towns that are more
and Muslim entry into trade, were not only relatively ethnically mixed, consistent with a beneficial effect for
peaceful locations for Hindu-Muslim interaction in the peaceful coexistence when members of a greater mi-
medieval period, they were also five times less prone to nority population compete with one another in pro-
Hindu-Muslim riots and around 25 percentage points viding complementary services. The effects also do not
less likely than otherwise similar towns to experienceappear to come purely from differential human capital
any religious riot between 1850 and 1950, two cen- (Glaeser et al. 2004)- as measured by the presence of
turies after Europeans disrupted Muslim advantagesmedieval skilled crafts production in a town. In fact,
in overseas trade. Between 1850 and 1950, medieval in nonport medieval towns, where skilled artisans were
port towns were around ten times less likely to expe-more likely to compete across ethnic lines for patron-
rience their first outbreak of Hindu-Muslim rioting in age, the presence of such skills actually left a legacy of
greater violence. In contrast, in medieval ports, where
any given year. These differences in patterns of conflict
between medieval ports and other towns persist but di- skilled artisans had greater incentives to maintain in-
minish following the Partition of South Asia, in part re-terethnic complementarity in the production of these
flecting the large-scale ethnic cleansing that occurred.crafts, the reductions in future violence are greater in
Yet, between 1950 and 1995, a legacy of interethnicthose that enjoyed such skills. Thus, institutions and
complementarity still diminished the incidence of eth- human capital appear to be themselves complements.
nic rioting by more than half. The results also do not appear to be driven by the
In both democratic and colonial India, this legacy selection of medieval ports by Muslim traders based
effect of historical interethnic complementarities isupon unobserved pre-existing factors that might have
greatest in environments where the State lacked thefostered a more tolerant local population. Because of
the severity of Indian Ocean storms, sheltered harbors
political incentives to protect minorities, and thus the
need to rely on local institutions to support peacefulwere historically prized as locations for medieval ports.
co-existence was likely greater. Medieval ports withThus access to natural indentations on the historical
one century less of medieval Muslim political rule- coastline, or "medieval natural harbors,' provide an ex-
i.e., periods when Muslims were historically more likely ogenous determinant of medieval trade, one that I find
to benefit from the protection of the State and thuslater diminished in determining the location of mod-
relied less on local institutions of tolerance- exhibit ern ports. Yet, those towns that became ports because
an eight-percentage-points lowered probability of a of their location at medieval natural harbors exhibit
Hindu-Muslim riot between 1850 and 1950. Similarly,similar reductions in Hindu-Muslim violence to other
in democratic India, a legacy of interethnic comple-medieval port towns.
mentarity has its greatest effect on reducing the prob- I then evaluate the potential mechanisms through
ability of a riot in the run-up to elections in states withwhich a legacy of medieval trade might have a lasting
weak party competition, conditions under which mi-effect on ethnic violence over two centuries later. I con-
nority voters may be less likely to be politically pivotal firm that the differences in violence in medieval trading
and state politicians face lower incentives to protectports and other towns arise specifically from overseas
them (Wilkinson 2004). Thus, both in democratic Indiatrade in the medieval era- where Muslims enjoyed
and in South Asia's nondemocratic past, political in- nonreplicable nonexpropriable complementarities-
centives and local institutions appear to have acted asrather than medieval land-based trade- where Hindus
substitute means for supporting interethnic peace. could locally replicate Muslim trading networks- o.r
The article draws upon a broad range of sources, in-modern trade- where European intervention eroded
cluding a novel town-level dataset that combines hand-Muslims' overseas advantages.
collected information from medieval traveller's narra- Continued colonial-era trade in medieval ports
tives, a 16th century Mughal census- the Ain-i-Akbari , also does not explain the results. Due to heavy
colonial-era indicators of demography and develop- silting of inlets and river mouths during the mon-
ment, qualitative fieldwork conducted in Ahmadabad soon rains, a number of medieval ports have in-
and the port towns of Surat, Somnath-Veraval, and creasingly ceased to be accessible to shipping over
807

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
Trade, Institutions, and Ethnic Tolerance: Evidence from South Asia November 2013
time. Those medieval ports that subsequently became cooperation, influencing ethnic competi-
in enhancing
inactive or inaccessible to overseas shipping show
tion, and mitigating ethnic violence, in theory (Greif
a remarkably consistent legacy of reduced2005; Fearon and Laitin 1996), in the United States
modern
(Olzak a
religious violence, while modern overseas ports, 1992),
use-and in South Asia in particular. Based
ful placebo comparison, do not. on detailed fieldwork, Brass (2003) finds evidence of
"institutionalised
I next draw upon qualitative fieldwork conducted in riot systems": concerted action by
Gujarat in 2006-7, historical case studies, and evidence
local elites to maintain fissures along ethnic lines, for
from the nationally representative urban local,sample of
sometimes electoral, gain. Wilkinson (2004) fur-
the Indian Human Development Survey (conducted
ther examines the state-level incentives for fomenting
in 2005) to characterize the institutional systems violence. that
He argues that when states experience greater
appear to facilitate peaceful exchange between Hindus
electoral competition between parties, minority groups
and Muslims. While the specifics of these institutions are more likely to be pivotal in forging majorities, re-
that persist differ in different medieval ports,ducing the incentives for political leaders to instigate
they share
an economic logic and continue to deliver similar out-for political gain. Varshney (2002) argues
ethnic riots
comes. In some communities, additional ethnic insteadspecial-
for the importance of cross-religious social cap-
ization in complementary occupational roles italoccurred
or "civic engagement" in defusing religious tension.
during the medieval period- these communities "Peace appear
committees" develop from existing cultural, po-
to have maintained interethnic complementarity litical, oreven
business groups that cross religious lines/
after the decline of the original complementarity These worksin provide extremely valuable insights
trade. In other communities, organizations and into the proximate causes of ethnic violence in con-
cultural
norms emerged that appear to have mitigated the in-
temporary South Asia and beyond. This article fur-
centives for ethnic violence by allowing the gains thers from
and seeks to reconcile these studies by analyzing
interethnic trade to be shared more equitably the between
exogenous historical structure of incentives that
groups and by building interethnic trust. I led showto thethat
contemporary "institutional" environment.8
the patterns evident in the case studies are Thisnot just
article argues that contemporary interethnic civic
anecdotal, they are reflected in reduced ethnic inequal- interethnic economic competition, ethni-
engagement,
ity, in increased trust, in occupational specialization cally polarizedin elections, interethnic inequalities, and
complementary activities, and in specific organizational ethnic violence are all regularities of behavior that
membership patterns among urban Musliminand partnon-
reflect the institutional legacy of close to a
Muslims living in districts with medieval thousand port head- years of the presence or absence of exoge-
quarters even in 2005. nous complementarities between ethnic groups. The
This article follows in a rich intellectual tradition article finds, consistent with Brass (2003) and Wilkin-
evaluating the long-term effects of historical institu- son (2004), that towns experience fewer ethnic riots in
tions (e.g., Acemoglu, Johnson, and Robinson 2001;years with stronger state electoral competition. And it
Banerjee and Iyer 2005; Nunn and Wantchekon 2011).is precisely in those environments in which State po-
By stressing the central role of interethnic comple- litical incentives to protect minorities cannot be relied
mentarities in encouraging cooperation and discour- upon, such as in times of weak party competition, that
aging conflict between ethnic groups, this article intro-local institutional arrangements, including those that
duces a new dimension into an important set of studies Varshney (2002) describes, increase their effectiveness
that have sought to understand the role of ethnicity at sustaining peace.
and interethnic inequalities as a determinant of civil In the next section, I provide a theoretical framework
conflict and public goods provision (e.g., Alesina and that delineates how robust interethnic complementar-
La Ferrara 2005; Baldwin and Huber 2010; Fearon ities and a sharing mechanism may foster interethnic
and Laitin 2003; Habyarimana et al. 2009; Horowitz tolerance and the development of complementary in-
1985; Jha and Wilkinson 2012; Miguel, Satyanath, and stitutions that support interethnic trust. I describe how
Sergenti 2004). My findings suggest that if the structure these conditions were met among Hindus and Muslims
of economic incentives for exchange, mobility, and vio- in medieval ports trading to the Middle East, and pro-
lence between ethnic groups is not considered, factors vide case studies from a medieval port, Surat, and a
may be omitted that can dramatically alter the impact nearby medieval city, Ahmadabad, that illustrate these
of more proximate causes such as ethnic heterogene-
ity and interethnic inequalities on modern indices of
7 Varshney's use of pair-wise case studies is especially illuminating.
peace, public goods provision, and growth.6 Each pair consists of one town where religious riots are rare: Cali-
This article most closely builds upon important stud- cut, Lucknow, and Surat- and one where they are common: Ali-
ies that have recognized the vital role of "institutions" garh, Hyderabad, and Ahmadabad. It is reassuringly consistent with
the theory outlined here that two of the three cities that Varshney
identifies as enjoying high levels of "civic engagement"- Calicut and
6 This article also speaks to an emerging literature that explores the Surat- were once major medieval trading ports and the three cities
reasons why poor societies become and remain ethnically diverse, where civic engagement between Hindus and Muslims ultimately
despite the social costs. The role played by exogenous complementar- failed- Aligarh, Hyderabad and Ahmadabad- were centers of Mus-
ities and reinforcing institutions is common to a number of ethnically lim political patronage, where Muslim clients competed with Hindu
diverse societies, one that adds to intriguing works that stress the clients.
roles of geographical constraints to migration and misguided gov- 8 This article also adds to other work on the proximate causes of
ernmental policies in fostering ethnic diversity (Dippel 2011; Fearon Hindu-Muslim violence that stresses income shocks due to changes
and Laitin 2011). in rainfall (Bohlken and Sergenti 2010).
808

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
institutions, along with case

options. evidence
Thus rather for
than class violence, societiesthose
where that
developed in other medieval ports.
local and nonlocal After
groups compete are likely describing
to exhibit
the empirical strategy greater and the
ethnic data, I then present
violence.
the results. I begin with In town-level comparisons
contrast, when ethnic groups provide complemen- of ri-
ots across India between tary1850 and
goods or services 1950,
to one followed
another, then the incen- by
evidence from the post-Independence tive to attack nonlocals falls over long time horizons.I next
period.
discuss how State political If nonlocals leave if attacked, locals
incentives and will local
face reduced institu-
tional legacies have interacted both
supply and higher future before
relative prices for goodsand
that after
the advent of India's democracy. only nonlocals provide. I The more that nonlocal
finally provide goods rep-
resentative survey evidence increase thethat
value of local goods, the more valuable
establishes the is con-
tinued differences in trust, the presence of nonlocal suppliers andoccupational
organizations, the lower the
choice, and political behavior incentives for ethnic violence.
among urban Muslim
households in medieval ports. Thus intergroup complementarity
I conclude by can discussing
support
how the theory may shed peaceful
light co-existence
on otherover time. However,
settings even with around
the world. repeated interactions, peaceful co-existence will fail if
members of one group are able to cheaply replicate
or violently seize the resources that make members of
INCENTIVES FOR TRADE IN THE MEDIEVAL the other group desirable trading partners. Therefore,
INDIAN OCEAN to maintain intergroup complementarity over time, it
is necessary for the source of each group's comple-
A simple theoretical model can be used to illuminatementarity to be costly for others to acquire. Comple-
why Hindus and Muslim traders enjoyed an enduring mentarities emerging from expropriate assets such as
environment of peaceful co-existence in the Indian wealth, machines, or land, or even artisanal skills and
Ocean region when many other commercial ethnic other mi- forms of human capital that may be replicated,
can therefore fail to sustain peaceful co-existence in
norities have not. It is useful to provide an intuitive
the long term.
sketch of the model and its relevant predictions (please
see Jha (2009) for details of the formal theory). Furthermore, in settings where the nonlocal group
is a small minority, even such "robust" complemen-
The model focuses on settings where there are two
tarities may be insufficient to maintain peaceful co-
types of agents: local and nonlocal. Nonlocals dif-
fer from locals only in that they have better out- existence. With few competitors, members of a minor-
side options. In our motivating example, "nonlocal" ity group that provide nonreplicable complementary
traders enjoyed resources, including information services
and can enjoy high relative prices and substantial
networks, that linked them to the Middle East and profits. This has been a common feature in the histories
the rest of the Islamic world. These external resources of many ethnic minority trade networks. Particularly
made it less costly for Muslims to leave a particular in times of resource shock or crisis, when strong locals
town and go elsewhere. In contrast, the resources of discount the future more highly relative to the present,
"local" Hindus tended to be also concentrated locally.such high profits may result in a temptation for agents
In the model, individuals from either group haveto engage in violence to seize these profits even at the
the following choices every period: to stay or leavecost of losing future gains from trade.10 Thus the main-
town, to produce a good for exchange, and to attack tenance of peaceful co-existence over time will benefit
any other agent that they encounter. Such violence is from mechanisms that redistribute the surplus between
destructive, but violence may be useful for seizing the groups and thereby reduce incentives to violently ex-
victim's property and to deter or punish the victim's propriate.
actions."Strong" individuals may exist who are more From the rise of Islam to the 17th century, Muslim
likely to prevail in a violent attack against weaker op-traders involved in transoceanic commerce satisfied all
ponents. The model can be used to find conditions that the conditions outlined above for sustaining peaceful
favor "peaceful co-existence" over time: an equilib-co-existence. First, there were Islam-specific advan-
rium with a mixed population of locals and nonlocals, tages to trade across the Indian Ocean. Pilgrimages,
full production, no out-migration and no violence. particularly to Mecca, coordinated the development of
In the environment above, an important conditionthe world's largest textile market during the Hajj (Lom-
that favors peaceful co-existence is that nonlocals and bard 2000). In 1503, one of the very few non-Muslim
locals produce complementary goods or services. Toobservers to visit Mecca, Ludovico di Verthema of
Rome, described the intimate coordinating role of the
see this, consider first the alternative: that locals and
nonlocals provide substitute goods and thus are com-
petitors. Then, with repeated interactions, a strong lo- 9 An alternative theory for why ethnic violence is more likely to occur
cal will have an incentive to attack weak nonlocals, than class violence is that mobilization requires resources available
as this allows locals not only to seize the nonlocal's to the wealthy, who prefer ethnic violence to class violence (Esteban
property but also to encourage nonlocals to leave, re- and Ray 2007).
10 See Landa (1994) for a discussion of how high relative prices
ducing the future competition the local faces. In fact, and the resultant wealth led to an expropriation of the Chinese in
nonlocal competitors provide more attractive targets Southeast Asia, and Chua (2003) for many other examples around
of violence than weak locals, as local competitors are the world. This is also consistent with the findings of Baldwin and
harder to encourage to leave due to their lower outside Huber (2010) on the role of between-group inequality.
809

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
pilgrimage and trade, particularly notingMuslim competitors could improve the terms of trade
the presence
of South Asian merchants from both coasts.11 Mus-
for the local population whenever relative prices for
lims had strong preferential access to thesenonlocal goods became too high.15
pilgrimage
routes, and the markets they induced.12 The Hajjports
Trading wason South Asia's coasts were thus well-
supplemented by pilgrimages to other sites,
favoredsuch
to provide
as geographical loci for peaceful co-
Cairo, Kerbala, Basra, and Yemen, that all fostered
existence and trade between Hindus and Muslims, even
regional trade. in areas that experienced little Muslim political con-
Second, Muslim advantages in oceanic trade trol (Figure 1). The connection between the robust
stemmed from preferred access to trade networks, interethnic complementarity and incentives for toler-
which are difficult to steal or replicate. A key char- ance in these areas were not lost on contemporaries.
acteristic of trade networks is that they enjoy increas- Shaikh Zayn-ud-din al-Malibari, a Muslim cleric writ-
ing returns to scale. The remarkable scale of the Hajj ing in 1528, described the process of conversion to Is-
in particular was such that it was prohibitively costly lam in South Indian ports and the understanding that
for even a substantial number of Hindus to replicate. the long, and to his eyes, "remarkable," tolerance Mus-
Since trade networks are also intangible, they were also lims enjoyed was a direct consequence of interethnic
impossible for Hindus to steal. Oceanic trade differed complementarity and the joint surplus that Muslims
from land-based or riverine trade routes in this manner, had to bring. He wrote the following:
as most long distance land-based trade can be divided
into relays of shorter distances, each of which can be Now in all these [Malabari ports] the population became
replicated by a member of the local group. Most sea much increased and the number of buildings enlarged, by
trade routes, however, cannot be replicated in relays. means of the trade carried on by the Mahomedans, to-
wards whom the chieftains of those places abstained from
It is therefore at towns with direct access to the Indian
Ocean that Muslim advantages in Middle Eastern trade all oppression; and, notwithstanding that these rulers and
their troops were all pagans, they paid much regard to their
became most relevant and gains from exchange be-
prejudices and customs, and avoided any act of aggression
tween Hindus and Muslims were most pronounced.13
on the Mahomedans, except on some extraordinary provo-
Third, Muslims had access to a natural, decentralized cation; this amicable footing being the more remarkable,
mechanism of redistribution of the surplus from trade from the circumstance of the Mahomedans not forming a
to the local population: increased intra-Muslim compe- tenth part of the population ... (al Malibari 1528, 17). 16
tition due to the relative ease of entry by any Muslim
into Indian Ocean trade. Unlike most kin-based trade
Malabar was by no means the only region where
networks that have high barriers to entry (Rauch and similar phenomena were occurring. The northern coast
Casella 2002), entry into Islamic trade networks was interacted with Muslims both as traders and as in-
relatively cheap for all Muslims. Pilgrimages provided
vading military forces. A key event occurred in 1026,
a clear coordination device, so even nonmerchant and when the Afghan Mahmud of Ghazni sacked the major
newly converted Muslims could enter trade. Family or Hindu pilgrimage center of Somnath, killing an untold
community ties were not necessary to follow estab- number of inhabitants. The sack of Somnath has been
lished pilgrimage routes, and indeed many pilgrimagescited ever since as one of the most polarizing events
were financed through trade (Ibn Battuta 1355; Lom- in Hindu-Muslim relations, leading to "great hatred"
bard 2000). Though trading "communities" did emerge,
members of these communities often were in economic
competition either within their own communities or volved in the Middle Eastern trade and cooperated in the Karimi
convoys across the Indian Ocean (Goitein 1966). It is likely that the
with other Muslim trading communities (Penrad 2000;presence of Muslim competition made Indian ports less profitable
Subrahmanyam 2000). 14 Incipient and actual entry by
but more tolerant destinations for these groups as well.
15 There appears a systematic relationship between the strength of
different Middle Eastern empires (e.g., Ismaili Fatimids versus Sunni
11 "At Mecca, we found a prodigious multitude of strangers who wereAbbasids) and conversion to those forms of Islam on India's coasts.
. . . pilgrims; some from Syria, others from Persia, and others from For example there was a wave of conversion to Shafii Islam (more
both the Indies, that is from India on one side of the river Ganges, and common in the Arabian peninsula) during Islam's early centuries,
also from the farther India beyond that river. . . . This vast concourse
while Ismaili conversion in Indian ports coincides with the expansion
of strangers of many nations and countries resort thither from various
of the Fatimid caliphate, and the centrality of Cairo and Yemen to
causes, but chiefly for trade, and to obtain the pardon of their sins trade.
by discharging a vow of pilgrimage
16 The Roman traveller, di Verthema, visiting three decades earlier,
and beyond the Ganges , they bring for sale precious stones, pearls
concurred with the nature of ethnic specialization and complemen-
and spices; and especially from that city of the
tarity greater
even India,
among converts: which
"When I was in Calicut it was crowded
is named Bangella , they bring much gossampyne cloth [fine cottons
with merchants from almost every part of the east, especially a prodi-
or muslin], and silk." (di Verthema 1503) [pp. 58-601
gious number of Mahometans
12 Hapless di Verthema himself, exposed forsea,a Christian in the holy
the Mahometans are exclusively employe
city, was rapidly sold into slavery.
that there are not less than 15,000 Mahometans r
13 Sizeable colonies of Hindu and Jain traders
mostly existed
born inalong
thatland
place." (di Verthema 1503, 94
routes to Persia and Iraq. The French trader, Jean deDuarte
traveler Thevenot (1633- also agreed, writing of th
Barbosa
1667) noted the presence of bania moneylenders
in tradeineven
Isfahan, Basra,
among Muslim converts, as well a
and Hormuz (Mehta 1991). However, shipping
degree was dominated by
of organization: "They call these Moors M
Muslims, and the great textile mart at Mecca remained
on nearly exclusively
all the trade of the seaports; and in t
Muslim.
country they are well provided with estates an
14 Though Muslims dominated shipping, other Middle Eastern trad- people have many mosques in the country in wh
ing groups, including those of Jews and Armenians, were also in- in council. (Barbosa 1519, 146)"
810

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
FIGURE 1. Major Medieval Ports and Political Patron
Note : Muslims traded to ports across both coasts in the medieval per
control (shaded darker) and areas where such control was fleeting (lig
political control and patronage in the medieval period, with mints est
trade
of the local population for extending
Muslims beyond
(Alberuni the Ind
1030;
Thapar 2004). and even China.17
Muslim dominance
Yet, an inscription exists that shows of overseas
thattradeincontinued
1262, for the
close to a thousand
authorities of the rebuilt Somnath years. Themade
temple Portuguese
a discovery
large-
scale land grant of temple of routes to
lands totheaIndian
MuslimOcean in trader,
1498 destroyed the
Nur-
ud-din Firuz of Hormuz to commerce
settle of in
a number
the of adjacent
key trading ports, often via
trading
port of Veraval, aware ofblockade.
theThe end of Islamic trade dominance
commercial was fur-
taxation and
ther
prosperity that a colony of expedited by increased
Muslims couldcompetition
bringby the Dutch
(Sircar
1962; Thapar 2004, 84-85). Similar inscriptions sub-
stantiating mosque endowments by44) Hindu
17 K. N. Chaudhuri (1995, elites
summarizes the evidence: and
"Although
rulers have been found throughout
Hindu India and the islands ofGujarat andwere
the Indonesian archipelago thenot
west coast (Chakravartito 2000; Thapar
be brought within 2004),
the orbit of [the] as
Islamic world for well
another
four centuries, the
as on the eastern coast (Bayly commercialDasgupta
1989; expansion of Muslim merchants
2004). and
traders across the Indian Ocean to South Asia and China is his-
Tolerance towards Muslim traders operating beyond
torically recorded from as early as the eighth century. There is no
Islam's political frontiersevidence
was not
of any unique
religious to Muslims
animosity towards India but
in either India
appears to have been a common feature of oceanic or China at this time
811

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
Trade, Institutions, and Ethnic Tolerance: Evidence from South Asia , November 2013
preference
and English, and the disintegration of the Mughal em- by the community leaders suggests where
pire. Mughal ports, such as Masulipatam, Surat,
theyand
found it: in 64% of the subsequent years up until
Hughli, gave way to competition from Madras, World
Cal-
War 1, the Bohra headquarters were located
cutta, and Bombay (Dasgupta 2004). Muslim trading
in medieval ports.19 By 1899, the Bohra population in
networks continued to be important in tradeGujarat
with was estimated at 130,000 (Campbell 1899).
Southeast Asia and Zanzibar, but the expansion ofwas the most prominent of these organi-
Surat
zational
colonial rule to these regions brought competition from headquarters. Surat emerged as the major
non-Muslim traders operating under colonial Mughal
protec-pilgrimage port to Mecca, following the silting
tion (Bose 2006). By the beginning of the eighteenth
of the nearby port of Cambay, and rapidly attracted
a Bohra
century, the era of Muslim trade dominance in the In- population. East Indian Company sources
referred
dian Ocean was over, and many medieval trading portsto one of their creditors, the Bohra mer-
ceased to be commercially important. chant, Virji Vohra, as one of the richest merchants
Thus, for close to a thousand years, intergroupanywhere,
com- holding a local monopoly over Arabian
horses in
plementarities existed between Hindus and Muslims (Richards 1996). This relative affluence seems
medieval trading ports. However, even during to have
this pe-continued over time; the 19th century Bom-
bay Gazetteer
riod, peaceful co-existence could still be threatened by (Campbell 1899, 29) states "especially in
shocks. Examples include resource or politicalSurat
shocks
[Bohras] are prosperous, many of them rich and
that threatened the survival of strong locals,thesuch aswell-to-do; the poor thrifty and free from debt
bulk
andking-
emerged with the increased instability of regional the unfortunate are maintained from a common
doms and the Mughal empire, or the exogenous devel-
fund." In an environment of low average wealth, and
opment of new substitutes to Muslim shipping, such
in the absence of complementarity and peaceful orga-
as occurred with the European discovery of routes to mechanisms to facilitate the sharing of this
nizational
the Indian Ocean. Thus, higher mutual incentives wealthex-with non-Bohras, such minority affluence might
isted in medieval ports than other towns for residents
be expected to raise , not lower, ethnic violence.20
to invest in and develop complementary mechanisms However, the Bohras of Surat provided complemen-
to maintain the incentives for peaceful co-existence
tary services and possessed such organizational mech-
even in the presence of such shocks. Insofar as theseThe Bohra headquarters shifted there in 1787,
anisms.
mechanisms, once developed, were costly to reverseand Surat has remained a major center for the Bohra
clerical
by any individual agent, they can be considered part of hierarchy ever since. Despite the subsequent
the "institutional" environment that shape anshift agent's
in headquarters to Bombay in the early 20th cen-
subsequent incentives for peaceful co-existence. tury,It Bohras
is of Surti origin continue to dominate the
these institutional systems, that I argue, have survived
leadership of the Bohra community (Blank 2001, 117)
to this day.18 and Surat remains the seat of the apex Bohra institute
for training local religious administrators, the Jamea
tus-Saifiya , founded in 1809.
Examples of Institutional Persistence The Bohra community numbered 212,752 in the 1931
Among Trading Communities Indian census and is believed to be around five times
It is useful to fix ideas with a specific example of a long- that worldwide today (Blank 2001). Despite its growing
lasting institutional system that emerged due to me- size, the da'i's organization continues to wield major in-
dieval trade and continues to affect modern interethnic fluence, playing a coordinating role in many aspects of
relations. During the reign of Caliph al-Mustansir bil-life, from naming most Bohra children to determining
lah (1036-1094), the (Ismaili) Fatimid empire reached individuals' occupational choices.21
its greatest extent. It was at this time that a group
of Gujarati Hindus first converted to Ismaili Shia Is-
19 The Rauza (mausoleums) of the da'is (numbered) help trace
lam. Their historic specialization in trade was such the organizational headquarters (medieval ports are underlined).
that they became called Bohras , from the Gujarati Ahmadabad: da'is 25-33 (corresponding to the years 1567-1657),
verb vohrvun- "to trade." With the elimination of the Jamnagar: 34-36, 38 (years 1657-1711, 1719-1738) Mandvi: 37 (years
Fatimids in Egypt, Bohra trade and pilgrimage links1711-1719), Ujjain: 39^0, 47 (years 1738-1780, 1840-1885) Burhan-
pur: 41 (years 1780-1787) Surat: 42^6, 48-50 (years 1787-1840,
shifted to Yemen, the headquarters of their spiritual
1885-1915).
leader, the da 'i ul mutlaq. The organizational headquar-20 In fact, Mitra and Ray (2010) argue on related grounds that if
ters of the Bohras moved from Yemen to Gujarat in the riots follow increases in Muslim wealth, this is evidence for Hindu
16th century. Though initially attracted to the major
perpetration of such riots.
city, Ahmadabad, the community leadership left there 21 Campbell (1899, 32) states: "the head Mulla [the da'i] is treated
with the greatest respect. . . . On both religious and civil questions
around 1657 to avoid religious persecution. Revealed
his authority is final." More recently, Blank (2001, 155) quotes one
Bohra respondent: "Back in Mombasa, my elder brother was run-
ning a hardware store. Business was not good, and he wanted to sell
18 Greif and Laitin (2004) provide a general theory of how equilibriafoodstuffs instead. . . . My brother sent a letter asking Syedna [the
can be self-reinforcing and self-undermining, altering "quasiparam- da'i] for permission but His Holiness would not allow the change.
eters" that individuals take as given. I wed their work to a central in-
'Stay in hardware', Syedna said, and so my brother obeyed ... ".
sight from the theory of robust comparative statics- that the presence
On the coordination of occupational choice, an ethnographic survey
or absence of complementarities underlie most robust conclusions on conducted by Jonah Blank (2001, 203) found continued complemen-
how optimizing behavior changes in response to shocks (Milgrom tarity (italics added): "Bohras overwhelmingly tend to make their
and Roberts 1990). livings as shopkeepers . . . The most popular [contemporary] trades
812

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
Every major Bohra community

A second set ofcoordinates
mechanisms helped reduce
its
the incen-
activ-
ities through a jamaatkhana, or whether
tives for violence, community hall, each
by coordinating responses
headed by a local amil connected with
to crises or by sharing the and
gains from appointed
exchange. In Gu-
with the approval of the jaratda'i's
and Malabar,
central
merchant guildsadministration.
and inter-religious
A notable aspect of these organizations
organizationshelped organize is
both the
boycottsrole
and jointthey
play in orchestrating transfers,
petitions to political
disaster
figures to seekrelief,
redress when mem-
and char-
itable giving. Bohras not bers
only of onehave
religious group were threatened
a history ofby strong
endow-
individuals (al Malibari
ing local public goods, particularly 1528; di Verthema
clinics, that 1503). Or-
benefit
non-Bohras as well, but ganizations
also nonaffected
also emerged to encourage communities
repeated inter-
are mobilized to help Bohras and
actions between non-Bohras
members following
of different religious groups,
natural disasters and even which religious
encouraged trust and rioting
the formation of (Qutbuddin
joint ven-
tures (Dasgupta 1994).
2011). Bohras, not surprisingly then, Muslimaretraders relatively
around India so-
provided commercial
cially and residentially integrated taxes andnon-Muslim
with explicitly endowed lo- so-
cal public goods,
ciety, relative to other Muslim including water projects and A
communities.22 evenlegacy
of Indian Ocean trade, the Hindu Bohra's
temples (Bayly 1989; Risley et al. 1909). Relative
institutional system
to other
of beliefs , that others will areas, conversion
accept to Islam and immigration
the authority and coor-
dination of the da'i, normsfrom,the Middle
of East was encouraged
adopting by local popu-
complementary
roles in trade to the locallations population,
in Malabar ports (al Malibariand 1528), organiza-
reducing
tions, the Bohra clericalcosts
administration, appear
of entry into trade and further increasing within-to play
complementary roles in Muslim competition. The sharing
facilitating theoflasting
the gains fromethnic
tolerance enjoyed by Bohrastrade, whether
inthrough
Surat, increased intragroup compe-
South Asia, and
increasingly, elsewhere tition,
as well. Surat
explicit intergroup itself
transfers, has devel-
or joint ventures
oped a reputation as an between
"oasis groups, are likely
of peace"to have provided
with Hindus
respect
to Hindu-Muslim relations.23
and Muslims in medieval ports reduced incentives for
interethnic violence
Though the Bohras and their fellow in times of crisis.
Ismailis, the fol-
lowers of the Aga Khan, AsareTable 1arguably
indicates, as in the Bohra
among example, a num-
the more
organized of the Muslimber oftrading
these institutional mechanisms-
communities,that coordinate they
are but two of the community
ethnic complementarity groups
and facilitate that
the sharing emerged
of the
and persist in Indian ports.
gains from Table
trade- appear1 summarizes
to have evolved and persisted ev-
idence gleaned from both
throughfieldwork I conducted
the 19th and 20th centuries, even after the de- in
2006-7 and the historical record for the different insti- cline of Muslim advantages in trade. A natural question
tutional mechanisms that emerged in medieval ports.is whether the reduction in interethnic inequality, rise
Medieval-era organizations appear to have fulfilledin trust, and participation in organizations suggested
two distinct but complementary roles. One set of orga- by the Bohra case and by Table 1 are reflected in the
nizational mechanisms encouraged group specializa- Muslim population in medieval ports more generally.
tion and raised the costs of replicating the services I will provide household-level survey evidence from
provided by another ethnic group. Specialization in 2005 that suggests that they are.
skilled activities was encouraged through a system of In direct contrast to the robust complementarities
apprenticeships that were often exclusively limited toand the institutional systems visible at medieval trad-
members of the same ethnic group (Campbell 1899; ing ports are the incentives present in towns, like Ah-
Haynes 1991). Norms and own-group social sanctionsmadabad, less that 140 miles from Surat, that were
also emerged that raised the costs of replicating an-the centers of Muslim political authority and where
other group's activities.24 Hindus and Muslims acted as substitutes for one an-
other in competition for patronage. Ahmadabad was
founded by Muslims in 1411 to be the capital of Gu-
include many of those customarily avoided by caste Hindus , but not jarat, with its wealth based upon patronage and the
considered seriously polluting. Hardware is a specialty particularlydemand it created for "three threads"- the weaving of
associated with Bohras, and other lines of trade . . . included glass-silk, gold, and cotton. Muslims and Dheds, a Hindu
ware, metals, electrical supplies, cutlery, paper goods, printing, cloth,caste, were responsible for the weaving (Gillion 1968,
foodstuffs, cosmetics, hats, leather and dyeing ..."
27-8). Though not necessarily members of the royal
22 As one respondent from the Bohra community in Surat told the
author in 2007: "When we went to our apartment complex in Nan-household themselves, much of the city's population
pura [a predominantly non-Muslim neighbourhood], they asked uswas often tied by client relations to people who were.25
"are you 'H-Class' [Hindu] or 'M-Class' [Muslim]? When I said I am The artisanal guild structure, of which Ahmadabad
'M-Class', they refused to rent to us. But then I said I was [a] Bohra,had 40, was predominantly Hindu and Jain, with the
and they said 'in that case, you are welcome.'"
23 This term was used to describe Surat during the Gujarat riots of
Muslim guilds "a weak imitation of Hindu models"
2002 by a Times of India editorial, Feb. 1, 2007. (Gillion 1968, 23). These supervised the maintenance
24 A prominent example of this was the norm of Kaala-paani ("black
water"): that Hindus that sailed offshore would be outcaste by their
own community. This cultural norm, though common to many Indian1991). Thus it may be that these institutions were established as a
sea ports, was particularly prevalent in Calicut and other ports inresponse to growing competition between groups.
Malabar (Bouchon 2000). However, Kaala-paani proscriptions on25 The dependence on patronage of these artisans is evident from
Hindu travel overseas were not widely followed in Gujarati portsthe departure of many from the city following the occupation of the
until Muslim dominance of overseas trade began to decline (Mehtacity by Hindu Marathas (Gillion 1968, 32).
813

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
^ (0
0$5 s o) £ 13
Il 5 io lo ° ^
IsSI* 2
io £s
lo £
z
° <2
z ° ^
° S
«i ! 1t-î I» "
» S
» ř ř1 !! §? 6I ^? §|-8°S5o>Si|o>
f i1® âf 1 § E;-.
I 6 £ S 8§
-
1« <2
*->,_« o ?
«> « t - o . o Q w
I M iriifft «> iii: . U «i

!cÜ
S! =o
, güIDfflZúí^üüüü i|S|l!f s J ^3,
güIDfflZúí^üüüü
Û- 0. OC
i lì0)0-
0)0- 5á
•
geo
geo
•
^ O _ (/)"■-
•252
C ^ .2
r- CD
c'j52
m c -
ü r
0 ® g Q.IO " ~ CO C C C «T O)
1 55
1 55^ ìli"2««*
S O Ū M CŪc •£ a o o« o.
« c X -1-
c © -1-X 05
r"
co = O o -5 C .2 .2 « (§ « S
o ® ^ 5 1 g g g «« 0 a>
5 4- & i
_ >%o
"g EIW^ _ ii § ^
ìli $ "g ir.m fi §s s is §
SeO COl'I C
Se 0) ¿0) ¿ (0
Is1?2- (0 ^ ^ Cn)ra
C ra 1 ¡sn)
1^'™¡s
> c>O) c£ g>
O)Ç £
râ -sÇ<o-s
JS ^^^ ^ VL
g VL g
ifts O C £ 0) ¿ °.slil (0 ^ C ra n) 5 ¡s ils > c O) lâÎ £ Ç -s ^ ^ VL co
2 O^ ooco
^ o C0
£ 5
¿Oro
o o < -sui® ^ =
lüt ¡¡«(Oc co
C0<D
? ss. f g
I ilio !0. si|
*f I IB«" % T5^-»šl l3°.řS|| üg-
Si S.l i '9.1 I S5"« s.i| f S I? £g
čs EE¡ill ES-5> 111
ES-5>
o
Eis>o_io)Ei2>>-oIa
Eis>o_io)Ei2>>-oIa
o o 0)
8 <-
il lili IE1
®2& ®
<-
5
O O O ß O
c © "g
i s 8 S ® S g
I :Si 8 M 11 ®fi
I !| III *§ I
(0 .1 - If ° Kří ® I« o g g®
t
o
1 S -Í2
S° ® £00
o 8 CNJ
^
^c
»<
CNJ »<
0. c 5 Q.»n c 'c CD CD
js.O.ä-« oco c~ t-
Ì fill js.O.ä-« f
0) o p s c$
c ¿ w ® Icū ¿ || w $ ?
■5 '%a Is g* S >=5
0) 3 * li
2 ř ÔJ
co o o o O
c "S c » c _ c ^ 'c _ O T",
(0
"S 118$? c » c _ 8 c s ^ 'c
""5
® £ $ S 8 & $ g ¡-<o
c 5 Ewcc/)
Q- ® «r
«rc/) !E co
c c/) co co
|c J5 Q-Q-
cw |c 5 =
=
O 5? 5° f° Co Cu
c
0 I- p £ ^
0) ž> O VJ "S
o) 'E *s c • ■
c C 3 2 c 0 ,- X- V ¿O
o 2 E F P O .
ñ E a oí > o
3
8?i r i §_ §
<n
? i
£
O
?! !»i^ I .it !¡.» I!
1 ! °I i's 03 !'& f s-i| 5m
>»
E
o
¡ro |fii£ |S22 |*3 g¡
I I ¡5 . !" I s If I 11
c
o
X
fi
II (Sil .«« 3|c!c -alltf-ii
ļi (Sil Sllšii .«« 3|c!c <b|1?S i ã s J s s» £§
g1 |Û£»mI
m m
|ÒOSO
5
g 5
• •
zQ.
¿ |2 5
LU
-I
- % <0 ^
t 2 _ c ^ CD ^
m "co co c _ «fi 2 « P CM
_í OOCD
I 2
CD I lá5
O CO £S 2 O I- |1 CO
814

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
These include
of local monopolies for their polynomial controls
members as well for longitude and
as serving
to address intergroup disputes (Gillion
latitude, propensity 1968,
for natural disasters 23).
(which might These
guild structures later evolved
lower or enhance into
cooperation
the (Wade trade
1988)), proxim-
unions
present in Ahmadabad into the rivers
ities to navigable 20th century
(which may raise town (Gillion
wealth
1968; Varshney 2002). independently) as well as to the Ganges, which due
However, despite these toorganizations,
Hindu sacred geography is anthe exogenous driver of
tension in
the proportion
the old city of Ahmadabad evidentof upper castes.
today I also subset
finds the datareso-
nance in both its architecture and the historical record. to only consider towns proximate to the modern coast.
Unlike the relative integration of Surat, each Ahmad-Conditional on these factors, I can estimate the average
abadi caste and religious group lived separately in one
treatment effect of medieval trade on religious conflict
in those towns that enjoyed medieval trade.
of 356 pols - walled and gated enclosures, the lack of
trust evident in the barring of their gates each night. The experience of medieval ports can be compared
Rioting both between Hindus and Muslims and among not only to otherwise initially similar towns that were
different sects of Islam took place the city in 1646not medieval ports, but also to other medieval towns
(Watson 1886, 62), and remained an imminent threat as which were historically rich but where the theory sug-
Mughal authority declined (Gillion 1968, 31). Duringgests that robust complementarities should not exist.
These include geographically similar towns on inter-
the Gujarat riots of 2002, Ahmadabad, particularly in
nal trade routes, where Hindus could locally replicate
its medieval precincts, experienced 24 days of rioting
that took the lives of more than 324 people. DespiteMuslim trade networks, and towns that were centers of
sharing a similar proportion of Muslims to Surat (13%political patronage, where Hindus and Muslims were
vs. 12.3%) and a history as Mahatma Gandhi's head-likely competitors.26
quarters for nonviolent teaching, Ahmadabad has be- Looking at effects over long periods of history raises
come notorious for the frequency and intensity of itsa separate challenge that deviates from a canonical ex-
ethnic conflict. periment: even controlling for initial conditions, towns
In towns like Ahmadabad, it is likely that Mus- under study were subject to different external polit-
lim clients, both converts and immigrants, substituted ical influences both during and after the treatment
for and competed with Hindu clients for patronage. that might also influence subsequent religious rela-
Though these towns were historically wealthy, and tions. Some component of these influences- e.g., the
Hindu and Muslim artisans lived side by side, there expansion of Muslim or European political rule- might
was limited incentive for intere thnic exchange between partly result from a desire to occupy regions with active
these groups and thus weaker incentives to develop trade. To account for such political channels, I compare
reinforcing institutions to support such exchange. the effect of a medieval trading legacy both with and
without a rich set of controls for these political fac-
tors, including 31 fixed effects for different native states
EMPIRICAL STRATEGY and provinces, interacted with the timing of a district's
colonial annexation. I am conservative in allowing for
As described above, the "robust" complementarities arbitrary correlation within these clusters.27 As we shall
between Hindus and Muslims in India's overseas see, ports
these controls do not greatly alter the measured
were largely created by exogenous features, treatmentparticu- effects.
A second potential concern with the above approach
larly the Hajj , that were inherent to Islamic doctrine.
Such complementarities made medieval trading is that Muslim traders may have chosen to trade at
ports
conducive to interethnic exchange and favorable geographically
for similar ports for unobservable reasons;
further investment in institutional mechanisms that such as having a local population with a proclivity for
bolstered such exchange. peace independently òf trade. This historically peaceful
My empirical approach uses towns that became me- population might continue to be inclined towards peace
in modern times. To assess this hypothetical challenge
dieval trading ports as an indicator of the "treatment"
of historic incentives for interethnic trade. The ideal to the results, I relax the assumption that the selection
comparison would measure the difference in Hindu- of medieval ports was uncorrelated with subsequent
Muslim relations between a town that enjoyed such in-religious violence, and instead use the presence of nat-
centives and the same town that did not. In the absence ural harbors on the historical coastline as an instrument
of such a counterfactual, I construct a series of control for medieval port location.
functions that mimic such a counterfactual town under Given the severity of the monsoon winds, medieval
two sets of assumptions. ports- more so than their modern counterparts-
First, I assume that the selection of locations forneeded to be located in naturally protected inlets.
medieval trade was uncorrelated with subsequent re-
ligious interaction. This assumption will be violated26 A related approach would be to construct a propensity score for
if medieval ports had different initial conditions that
medieval trade based upon nearest-neighbor matching, both geo-
might also have had an effect on religious violence, for graphically and on medieval characteristics, and compare medieval
ports to towns that had similar propensities to become such ports.
example, through congenial geography that providedThis approach leads to similar results.
increased opportunities for subsequent wealth. Thus,27 The timing of annexation had a number of effects, including dif-
I add a rich set of controls for initial conditions that ferent land tenure systems (Banerjee and Iyer 2005). See also Online
might have impacted subsequent religious interactions.
Appendix Figure 1.
815

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
TABLE 2. Summary Statistics

Towns, Not Medieval Natural Harbors, Not Medieval
Ports Medieval Ports Ports
Riots, 1850-1950 Obs Mean SD Obs Mean SD Obs Mean SD
# of Hindu-Muslim Riots 476 1.116 3.416 53 0.925 5.487 59 0.136 0.472

Any H-M Riot 476 0.418 0.494 53 0.170 0.379 59 0.102 0.305
# Killed in H-M Riots 476 23.277 242.361 53 88.906 639.995 59 0.136 0.571
Total Days of H-M Riots 476 1.630 11.301 53 3.000 20.598 59 0.051 0.289
Initial Conditions
Medieval Natural Harbor 476 0.132 0.339 53 1.000 0.000 59 0.814 0.393
<10 km from Modern Coast 476 0.111 0.315 53 0.528 0.504 59 0.898 0.305
Log. Dist. Modern Coast 476 11.998 2.031 53 8.946 1.739 59 7.358 1.709
Log. Dist. Navigable River 476 12.755 1.641 53 12.684 2.358 59 13.364 1.704
Natural Disasters, 1850-1900 476 1.536 2.512 53 0.906 2.133 58 1.603 3.201
Log. Dist. R. Ganges 476 11.846 1.960 53 12.823 2.091 59 13.292 1.537
Medieval Era Characteristics
Medieval Town 476 0.592 0.492 53 0.566 0.500 59 1.000 0.000
Mughal Mint 476 0.084 0.278 53 0.000 0.000 59 0.051 0.222
Other Muslim Patronage Ctr 476 0.132 0.339 53 0.057 0.233 59 0.220 0.418
Mughal Internal Trade Route 476 0.153 0.361 53 0.038 0.192 59 0.051 0.222
Mughal Skilled Crafts in Town 476 0.048 0.215 53 0.075 0.267 59 0.169 0.378
Major Shi'a Dynasty 476 0.200 0.400 53 0.208 0.409 59 0.186 0.393
Centuries Muslim Rule 476 4.073 2.286 53 2.727 2.691 -59 2.097 2.126
Colonial Era Outcomes and
Covariates
% Muslims 1901 244 29.879 17.732 20 18.596 14.884 22 32.449 22.101
Mun. Income per Capita 316 1.805 3.092 28 2.155 2.6382 28 1.580 1.103
Colonial Overseas Port (1907) 476 0.038 0.191 53 0.170 0.379 59 0.356 0.483
Log. Population 1901 476 9.672 1.129 53 9.420 1.209 59 9.170 1.315
Harborages were typically located that

at small inlets
towns that were at harbors in the medieval pe-
formed by indentations in the medieval coastline
riod have become increasingly inaccessible to shipping
(Dasgupta 2004) (see also http://dx.doi.org/10.1017/
(Arasaratnam 1994). 28 By providing natural variation
S0003055413000464 Online Appendix Figure
in the 2).
viability of In
trade in towns over time, the silt-
fact, an overwhelming share- 81%- of ingmedieval
process allows us to assess whether it is continued
ports
in the sample were at locations that colonial era overseas
possessed likelytrade and shipping in medieval
medieval natural harbors (Table 2). Insofar, as seems
ports or other mechanisms that explain the legacy of
medievalin
plausible, that towns with an indentation trade.
their me-
dieval coastline were not any more likely than other-
wise geographically similar towns to DATA
have attracted a
more peaceful pre-existing population, this compari-
son allows us to assess the degree to The
which
datasetmedieval
on pre-Independence Hindu-Muslim vi-
traders' selection of locations in which todrawn
olence, trade may
from newspaper reports and official
have biased the results. I provide evidence that
sources, suggests
is largely based upon that compiled by Wilkin-
that medieval era coastal indentations do seem to have son (2004). In this dataset, a religious riot was defined as
little effect other than to drive medieval port location,
a violent confrontation by two communally identified
including demonstrating that these indentations groups.
are Data on historical trade in India's ports came
uncorrelated with the location of colonial era ports. from a number of sources. The Periplus Maris Erythraei
I then exploit this exogenous determinant of medieval (Casson 1989) provided the. locations of a number of
port location directly, providing instrumental variables
pre-Muslim and early Muslim ports. I then drew on
estimates that compare modern religious relations the in accounts of contemporary Muslim, Christian, and
towns that became medieval ports because of their Chinese observers, including Chau Jukua (1225), Ibn
historical natural harbors and geographically similar Battuta (1355), Ludovico di Verthema (1503), Duarte
Barbosa (1519), and Zayn ai-Din al Malibari (1528),
towns that would have become medieval ports had they
had such harbors.
Another natural process allows a further robustness
check: the coast itself has moved over time. The mas- 28 Even contemporary dredging techniques (which of course were
unavailable during the period of Muslim trading dominance) are
sive flow of water from the hills during India's mon-
unable to contend with the volumes of silt generated. Even the
soon rains regularly pushes large amounts of silt tomodern
the port of Calcutta is no longer accessible to most shipping,
mouths of rivers and inlets. Over time, silting has meant
which has been diverted to the downriver town of Haldia.
816

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
economic historians of the region

proportion (Chakravarti
of medieval 2000;
ports experiencing at least one
outbreak of religious
Chaudhuri 1995; Subrahmanyam 1990),violenceasbetween
well 1850 as
and 1950
ex-
amining every town listed was in
aroundthe10%; close to 40% of other
Imperial towns faced a
gazetteers
(1907) for evidence of contemporary
riot. The intensity of the riots and medieval
also appears to be lower:
trade. on average, five medieyal ports together experienced
a single death duetrading
I identified a town as a medieval to religious violence,
port butifin other
it
towns, of
exhibited substantive evidence religious violence overseas
direct claimed an average of nearly
trade,
23 lives independent
prior to the 18th century and per town. Natural harbors
ofthatEuropean
did not become
involvement.29 1 found a total
medievalof 68
ports alsoconfirmed medieval
have a greater incidence of reli-
trading ports in undivided gious
India,riots and
ofmore intense violence
which 59 were than medieval
dis-
tinct towns in 1901. ports. However, as indicated by municipal income, me-
These medieval ports were linked to their geograph- dieval ports were on average poorer than both harbor
ical location using ArcGIS. To categorize medieval era towns and other towns. Medieval ports also have a
"natural harbors," I used the U.S. Geological Survey more mixed religious population (see also Figure 2).
Digital Atlas of South Asia 2001 to identify water These indicators are commonly associated with higher
bodies that were within 10 km of the modern Indian rather than lower incidences of ethnic violence.
coastline, including nonperennial ponds and streamsTable 2 also shows how medieval port towns compare
and those without an outlet to the sea. If these water
along a range of geographical and medieval character-
bodies intersected the coast in the medieval period,istics. An overwhelming share, 81% of medieval ports,
they would have produced minor inlets, or sheltered were located at natural harbors. Medieval port towns
harbors. I define towns within 10 km of those water are also more likely to be near the modern coast. Yet,
bodies as having had access to a "medieval natural
apart from these distinctions, medieval port towns actu-
ally appear similar in their geographical and medieval-
harbor." This approach identifies major irregularities
era characteristics to other towns.
and inlets that are likely to have existed in the medieval
period (see Online Appendix Figure 2). The similarities between medieval ports and other
I constructed a GIS of trade routes, mints, crafts, towns
and are confirmed in Table 3, which shows the de-
terminants of medieval port location. Along a range of
towns in medieval India using the Mughal census com-
missioned by Emperor Akbar, the Ain-i-Akbari initial
and geographical conditions (column 1) medieval-
era measures of trade and human capital (column 2),
supplemental medieval sources documented in Habib
restricting the data to towns within 200 and 100 km of
(1982) (see Figure 1). I matched these towns to districts
in colonial India by manually tracing and then geo- the modern coast, and adding colonial era fixed effects
referencing pre-Independence district maps. Thisfor GISnative state, province, and timing of annexation
(columns 5-7), there are only two key robust deter-
was used to link towns across district and state changes
throughout history and across periods of British minantsand of medieval port location: coastal towns and
Muslim rule. I was able to match all towns to their medieval era natural harbors. This does not change
British district and native state.30 if we do not control for natural harbors (columns 8
and 9). In contrast, colonial-era port location seems
RESULTS to be unrelated to the presence of medieval era nat-
ural harbors (columns 10 and 11). The F tests of the
Table 2 presents summary statistics comparing strength
me- of the natural harbor relation with medieval
dieval ports to other towns that existed in 1901, portsand
are sufficiently strong to avoid weak instrument
to the subset of towns at natural harbors that did not pathologies in most specifications.
become medieval ports. Medieval ports exhibit strik-The average effect of medieval trade on the num-
ingly lower incidences of religious violence compared ber of Hindu-Muslim riots faced by medieval ports
between 1850 and 1950 is assessed in columns 1-5 of
to both these classes of other towns, as well as to towns
Table 4, which present incidence ratios from neg-
geographically close by (Figure 2). Medieval ports ex-
ative binomial regressions appropriate for count
perienced around five times fewer riots on average. The
data.31 Towns with medieval trade legacies experienced
around five times fewer religious riots than similar
29 This definition eliminates most river ports and those ports eitherbetween 1850 and 1950. This result is robust
towns
founded by Europeans (including the Presidency towns of Bombay,
Calcutta, and Madras) or those that became overseas tradingand remarkably consistent even after controlling for
ports
initial conditions that might shape port selection (col-
as a result of European establishments (e.g., Hughli, Tranquebar).
umn
M In addition, I collected data from the Imperial gazetteers on a 1), and medieval factors influencing trade and
number of different natural disasters from 1850 to 1900, including
Hindu-Muslim relations (column 2). The effect actually
droughts, earthquakes, locust infestations, floods, and cyclones. The
strengthens when we add colonial era provincial and
cross-district patterns in propensity to face these natural risks are
likely to have persisted up until India's dramatic population gains
beginning in the 20th century. The decennial censuses and Imperial
31larger
gazetteers yielded data, mainly at the district level, but also for This table uses the subsample of 248 towns for which complete
towns, on religious demography, municipal income, as well asdata exist. Using the full sample for each set of covariates yields
polit-
ical histories. Municipal income per capita provides a town-specific
similar results, but suffers from the problem of under-reporting of
measure of the average wealth of the town- this measure wasrioting based in nondescript towns, including many medieval ports. Towns
mainly upon a tax on internal commerce (octroi) and a poll tax that are well documented by official sources also tend to be those
with
minimum wealth requirements. where the religious rioting is well documented.
817

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
FIGURE 2. Medieval Legacies: Religious Composition and Hindu-Muslim Riots
Note : The pattern of modern religious demography mimics patterns of Muslim rule, medieval trade, and political patronage. Medieval
ports and major Muslim patronage centers (such as those that housed mints) continued to have greater Muslim populations relative to
nearby areas in 1931 . Medieval ports, however, experience fewer religious riots relative to towns nearby.
annexation fixed effects (column 3), and select finer dieval precincts actually appear somewhat more prone
coastal samples (columns 4 and 5). OLS also provides to violence on average. .
consistent results (columns 6 and 7). Other useful medieval comparison groups include
It may be that outlying towns that were highly riot towns where the Mughals established mints to mon-
prone are driving these results. Columns 8-11 ad- etize wealth- as the theoretical framework suggests,
dress this by instead examining the probability that these towns, despite being historically wealthy, ar-
a town experienced any religious riot between 1850 guably provided incentives for interethnic competition
and 1950. The effect is again remarkably consistent between Hindus and Muslims rather than complemen-
across specifications- medieval ports are around 25 tarity. Indeed, mint towns appear close to twice as riot
percentage points less likely to experience a religious prone. Likewise, towns on inland trade routes, where
riot. Hindus could locally replicate Muslim networks via
I can evaluate a number of alternative explanations relays, also show increased probabilities of subsequent
and additional mechanisms. Otherwise similar coastal ethnic violence. Thus, rather than historical trade per
towns do not appear more "cosmopolitan," and se, theit appears that it is the exogenous and nonrepli-
propensity for natural disasters also does not appear cable interethnic complementarities present in me-
to affect ethnic tolerance by fostering risk-mitigation dieval overseas ports that have lasting effects on ethnic
institutions (Wade 1988). Nor does it appear that tolerance.32
the
effect comes from simply "learning how to get along"
over time or other general survivorship effects: con-
32 Bayly (1985) and Prior (1993) argue that pre-Independence reli-
trolling for whether a town was mentioned in the Ain-
gious violence tended to occur when major (Shia) Muslim festival
i-Akbari or other medieval sources does not affect the
processions- Urs and Muharram- tended to coincide with Hindu
results, and towns with (often crowded and poor) me-
festivals. Since both religions follow the lunar calendar, these
818

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
O) ^ CO Õ W CVJ ^ CO CO CV1 ÏO O R O CO CO Õ Õ Ti- ÏO Tí- ^ ^ oT C3) CO = CO C30 CO

1- COt-OJIOOCOi-i-OOi-OCMCVI^COCOCOOCMOIOtJ-OOi-cÔOoOD CO = CO mScO C30 o
w ^ OI-LOT-OOOOOOOOOOOOOOOOI-OOOOO Os LL > OJ c> ^ o
■g O O O O O O Ö 0,0 O, O O Ö O O O Ò O, O (DO O O 0,0 o, Ö ö r
£ ^
co
C O TtOCOCOOCVJOOOOOO O) 1^ = u ^ o oo S- Lfì
SÔT-Tt^-LOi-CsJO>CVJCMLOr^
0 3, OOLOt-OOOOOOOO 1- CO LL Z ^
o
ou u
ddddddoddddo
ļ - - ļ - -! --- ļ
- -
ļ oo
- 2 T-
T- CŪ
* c
* 1
Ä i-CDO(X)COi-^COinO)^0)C'JCOO)WOONCVIOO^lí)CO - w oo Tt ~
^
^ Ä
io oT
1- OOCOCM-r-CMi-T-OOCOCVJCO^COinCMCOI^
O O O O O O O O O O O O O O O O O O
q3
* >
* 0
_ ^ti-^j-m^th-^tcNjcor^ = ^ co c
^^ 00
00_LOOOOOOOOOO
W0)cv|i-00r-i-00 = =J
LL5=J
^ Sh
Z co
JqLO£c o
o
0000000000 S
! ' ' ' ' ' I ' g
r- C
î „ï „ „ „i „ „ „ „ _ _ „ -I r- < x
00C0cbc0^1-CX)C0CMCX)l000OC0i-05Ttl0ls-.|^0)'i-CX)i-i-T-lO „ „ „ „ „ „ _ _ „ iS
O C0C0l0WOC0mNC'J^^C0Oi-(0T-lfìi-C0^0)Nì-0>ON CM O w o o W
^ c. COOCOrOOWOOOWOWrOrOrrrOOrOrO COO $ o > 00 H ^
^ 0000000000000^000000000000 oooqV co
c
h u
u E 1 z
o -C
¡t^ fQ
'5
¡¡_ ^ fQ CM0^-T-00000000I-0000I-0T-00I-00I- rřo 5CVJ>-T-O 0
I ddddddddddddddddddddddddddirio'nV
- - i - ! ~ - - - ! - i " i - -i - -- ~- - !- -
- "ÍM
- - -°- °22 ^^
CO
O
o £ d
-I
4-*
"£ _ CO^CO^CMO)^C01-NT-I0î-C0ir)00c0i-W00c00)(x)0) = c/) co oo g

Z
0) iii- CMO^ï-OOOOOOOOOOOOOOOOOOOOOO T-O LL d q g ®
r? in iii- _ CO CO IO 1- 1- CM O CM 1- 1- O O evi CO CVJ CO CM IO O Tř CM CO CM io CM CO O = d c/) q 3h co CO g
S 0 - 0 0 0 0 - 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1^0 ®
(0
0) g - "1 - !" 12
2 m
S2
* * * *
c
fc C
> * * * * *
> O
O _ CO 3. CMOCOOOOT-OOOT-OT-OOOOOT-T-OT-OI-OO COO CO O T- Z 00 ■ -K w
_ ^
.2 ž dddddodddddddddddddddddddd aio R V -o
c
o
o
o
S * • E ®
* * - ^ D
0 Ä boNNOCO^<O^CONi-COOCOCOCO^lí)COCOir)COTl-OT- S o O co •£
"O Ä CO OOCOCMCJO-r-COi-TtCOCMCOCOi-r^Třl^lOr^lÔlOOi-OlOCO 1^1- Wo5r-<0 O «?
T-O^tOOOOOOOOOÔOOOOOOOi-OOOO O) O co CM Z ^ Q -j
ddddodddddoodd odo d odod odod coo Q v Q "g
c
(0
1
<D
*
*
*
*
"O
(D
S _
0) ^
^ 1-OTfOOOOOOOOOOOOOOOOOOOOOOOOOLL S
3 ZJjq O
S
000 000 0000000 00 0 000000000 0 CM o ^
O Sã
* * co
0) * * .-
c ^ ř- ř- CO ^ io CO CO Tř CM T- CO = ^ CO CM i
^ P COlfìLOCOr-1-rOT-i-OO IÔzjOSkCOc = ^ CO
(0
c ^ T-OÔOOOOOOOO T-O IL ZSo ^
d oo oo do oo oo o evi o o
E
0) i_ "co I- m
o O co ^ 0) -g _ -o
Û -Q O ^ €/).>_ 5:E _ in co
c
o
xxE«"§*§
O ® 20®|¡n t-co
a, 3 t- coc 8-
©
««C-Slip ojO.S'" .2 ■§
é¡ñ
(0 ««C-Slip = o®.g
£
O) - E 2 Z O > m £ O -Q ^ S c g
0)
en - §2oogo < %
■a ^c0CDW<DÍ = ->S = -^3 2 x c/) 2
CÔ
■a Cl)-*"
Cl)-*" C C= üücoog^6^2»¿^
c/) C C ^c O
c ļi; = C. rr' CO^ ļi;-r^
= C.
O rr' -^3 CO^ ^
LU
(I)2®ÍSÍSqíS - 2®-=-=¡£¡5 5U_ ~Qļ a) ^
-¡
1 « s I ^ g rA-® 1|§ <
m
li i II? «
0|2|25°Z°22022S0 i I ¿ « ¿O0C 2
819

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
* *
* * * *
^ CO
-r- C
^ Q
O O O O O OO O Ö O O O O O O OO O Ò O O O O OO o r
S F ^
J ÏS ^00WCjT-WNOOMr)OO)Oir)C'Ji-O)NCM0N(DW(Dr- CD C'J -2 o ~ ^ *-
o 0-g5 0)î-COO^COi-Oi-0)inÔO)^^COCOCOÔMí)OCO 1-^- Wo52 ~ ^ ^
LO ^ 2 LL T-OT-T-OOI-T-OOO --O T-r-CVIi-i-CMi-CMCOi-Oi-O <N C O JO t- Z 00
Jr Q. "o ooooooòoòoooooòoòodoòooooo o o q V ^
0
oc _ : „Í „ „ _* _ g
^ -"g _ * «OÍbÔ>OOCMO>TMn^lOÎ^®COOO^ÍřCMÎOCOCMÍ-^ „ „ _* _ COCO = _ 00 5E
. o^-g 5 CMcomcoco^cococvii- ^-cocMcocococNT-i-r^'tcMcoT- coco mi- = 3 _ 00 c
1 . ^ 2 LL CVJO^-T-OOOOOOOOOOT-T-I-T-CJOOI-T-I-OOOCOLL 3 .O)
û_ "o 0000000000000000000000000000 *w
< ***
*
* *
ļT
©
•gīž5coi^com«fr^-c'icoc'i!-mco
o^-g ^-ôTcVJOOOO^Íf^^^r * *-C'l^=
=3 °
° _ 00 ~ ©
Í3-2ll <M0^-i-00000000 o O)- li. =3_Z00.2
.2 o
o
û_"o dddddddddddd " " " " 00 $
I " " " " " I " 0
c
* * * - ~ Jï ^
Ä c o cbīoiOÔOĪOOO^CŪCDCOOOO * CD CO ¡O C» * CM CD OO OÔOO) S O co ^ O) x
Ä rC _j i^LOO)<N<Ncoi-^-coc'jOr-cviCT)CMr^cococococOT-cMO^-o œ o 0 ^ S m ®
^ O _j ^wcviôr-T-r-ooNWi-wMnoi-comT-o)oifìw^ co o ;r 0 00
yâdooooodododoôdoTôôyooo Q V ¿5
*
rn bo7oí"iococoF^,N.O)cocO'«-oícoioo>rôo)coc'jcoTtcDi^ _ z
^ rn ^ CONOOOOU)WCDCO<DO)ini-CONi-mÔi-i-O^Mí)CO -= _ £ 22 3 -
o ^ $£> o COCOlfìNOT-lOCOOOCOCOWCVI0N00rJ-^COOCOCOOW P ¥cti ° -
g) o ^ o ddddddodod ddodo ddd^¿Y ood ° "g
to
¿Ç<S
E
■=-*
I
-i
00 ^CÛ ~ C0C00000^C0MO0)Ni-inif)OT-NOC0L0CDÔl00)0)C0 O CO ^ ^
LO - CO i-CvJOJCVJinr^-r-OlOCMi-COCOOOCDi-COCVJCDOCÚT-CDCOT-CO «o 0® ^ co
0)£ - CO OOCVIÔ)COOONC'JOONCÛMONCOO)<ONT-NI-OJOOO g 1- >- 00 £
^ ^ _. doihu)dor:¿doT:¿^¿^¿cM¿^¿(ú(údo^¿ Q V t
- C . ^ * * ■ E -g
- et/)* . ^ * - ^ c
■ë _ m-S o)coinoNi-ibmoocooT-cocMtx)T- o)(ûNcoTf^cocococo -2 0 ^ r-» S
o ^ _ -S i-WONCOinCOr-NNCMCOT-COW^COCOCOinCOi-0OOOW w O ^ 0 r-» r- ^
OC ^ ebrr OONWO)i-Ni-Nt-<NWt-t-(OO)Ot-CX)0O)CM^01OCM COcvj^rļl O 0 r- ^
£ 0^. docortdocdr^dodo^coôr-'odoin ¿dodo ¿V J ^
3 .
Tñ ¿
^ .Eg * ^ „ „ „ „J „ r-, r-^ 0
A ^Cû ■£= ífOi-ÔOOOCDCOCOCOCMCD
o -o co • cc rîoiôo^-N.oio^-r>.mooT-coiOi-iocoo>c'i(Mi-o)cocMco ,? ^
in -o c S?CC OOmCOOíCVI^-i-OOOO^CVllO^CM^T-CMCOOO
o>
i- if o ¿coôî^ ¿-1-' di-' ¿evi ooo^' ¿ ^
¿
U)
O * §
00 ^CÛ ^ T-mOCON0CVIN^COi-m(DONOCOÔCOi-(DONCDO = o <5
T- ^ CM .CO OCVINT-^TtNtOinÔïÔNCÛT-Ti^CMOOO^^CÛC^CO ,? ^ o 3
^ gfCC C'JT- COCOi- C'JC'JOOOOOOT- C'JT- COCOCM^ŤCMGOCOLOCOT-I- LL ^ C'J Õ)
CO
2 oocdcoi- o i- oôdo t- OWOÒO ¿T-' ¿i- o r-' o g
0
E c 'co ï ï Ï •-
^CÛ ^ o^"oo o7l^ CD O CO CO CD 00 O = o °9 05
1
(0
^ .co
ÔCOCMi-NOîi-WWOOOCOO
O CD 05 CO IO O 00 O) CD CO IO ^ F
U- ^ C'J E F
3 Z¿ ddcodôrôôdo c
2 x-v o
i 1- Q-
3
"O - ^ s
- CO 1- m ^ co
C
CO ^ 1- 0 ^ m ^
if O to > co in co
O
T. co nrooO0^ m
o nr m ^,oIs-
lo 0
I- |T
|T in
U- U-
cr d
d
c
o
E o ® ® 0 3 I- œ >, c
©ť-^Vco 0 SîOc*^ •■= O
œ
-o
O O I S " ¿ |> g ® ig o •§ I I 1
'55
«2^°°®S!:ci?«ïïî
1 ! i " i i ! g M s o •§ ! f I I
0)
0)
i_
D)
= x »
O
CC
5J§ §l§§i.ì|i«J«P = cog x » i.

gir 2.2^ s
UJ co -7 CO (j) co co _ ^ co CO fk 0 m n%
-I
CD
ž co -7 ¿QÍ52Q CO (j) co è co _ « ? ^ Ž co S CO ÏÏ fk
ilřšiřlíiilfi ^F_j_JZ_J5S05^20
^F_j_JZ_J5S05^20 (01 (0 û. O CC 2 (01 is (0
_É_
820

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
TABLE 5. Regression: Interactions with Human C
Probability of A Riot Number of Riots
OLS: Years 1850-1950 (1) (2) (3) (4) (5) (6)
Medieval Overseas Port -0.206** -0.250*** -0.237** -0.532 -0.902* -0.899**

[0.077] [0.084] [0.107] [0.454] [0.529] [0.356]
Medieval Skilled Crafts in Town -0.017 0.115 0.306* 2.015* 1.927* 2.380*
[0.113] [0.127] [0.169] [1.165] [1.108] [1.248]
Medieval Port X Skilled Town 0.002 -0.081 -0.173 -2.734*** -2.201* -3.678**
[0.269] [0.292] [0.275] [0.978] [1.263] [1.492]
R-squared 0.27 0.42 0.45 0.27 0.44 0.60
Sample Full Full Coastal, Full Full Coastal,
<200 km <200 km
Controls Medieval Medieval Medieval Medieval Medieval Medieval
Province/NS x Annexation FE No Yes Yes No Yes Yes
Observations 248 248 110 248 248 110
Notes : All regressions include quadratic polynom

Navigable Rivers and the Ganges, Coastal Town
Center, Inland Trade Route, Skilled Crafts in Tow
Robust standard errors (clustered at Native State
An important debate
where exists
there isbetween
greater c
that coordinate
'institutions'- considered narrowly and
to be su
cheaper
or executive constraints- when
or there
human is
capita
sible for beneficialInsocial
otheroutcomes (eg G
words, rather t
2004; Rodrik, Subramanian, and that
dence suggests Trebbihum 2
complements.
executive constraints were arguably high
Indian towns in the medieval period, the me
legacy effect might be capturing medieva
Placebos and Robustness
human capital rather than the community
tives and largely informal
So far the location of norms and
medieval ports has or
been assume
that I emphasise as forming medieval institu
to be exogenously determined. I now test and relax
tems. Table 4 assesses whether historical h
this assumption. First, it is possible to use the historical
accumulation- as measured byto the
natural harbor instrument medie
test for this exogeneit
of skilled craftsmen- subsequently reduce
assumption under the probit assumptions (Smith and
lence. Not only is the effect of a medieval le
Blundell 1986). Although natural harbors are a stron
to controlling for the presence of such skill
determinant of medieval port location (Table 3), I fa
that skilled towns to are more
reject the riot-prone
null hypothesis of exogeneity with(C clos
Furthermore, Table to 90% confidence (Table 4). In other the
5 shows that leg
words, medieva
dieval skills differs dramatically depending
ports that were chosen for trade due to their natura
they were accumulated in do
harbor locations medieval port
not appear significantly t
differen
skills accumulation
from otherby Hindus
medieval and ways
ports in unobserved Mu tha
more likely to be might
institutionally organiz
be relevant for modern religious conflict. Thu
tain complementarity- and
it seems reasonable in other
to interpret the coefficienttow
of me
medieval port towns that accumulated
dieval port as the average treatment effect of medieval
sk
enced between two and four fewer religio
trade on religious violence in those towns that enjoye
otherwise similar towns,
medieval trade. in contrast to no
which experienced Even
two additional
though the exogeneity tests suggestriots.
that se- T
are consistent with a more general inter
lection does not appear important in this context, for
that the returns to human capital rise in en
completeness, Table 6 subjects the main results to a
battery of placebo checks as well as providing local
average treatment comparisons. As I have argued, in-
processions coincide every three decades. According t
terethnic complementarities between Hindus and Mus-
such religious festivals were used as a display of we
by an emergent Hindulims in overseas tradeclass
middle were largely
indisrupted
the during the
wake of
Muslim political power. colonial
If it period.
is If my interpretation
the case that is true, the
theeffect coinci
of processions played of colonial
an era port status on role
important ethnic tolerance should
in pre-Inde
then it is likely that riots should
be weaker than the occur more
effect of medieval often
ports. If, in- in ar
term Shiite traditions. However, as Table 4 reveals, th
stead, the effect of medieval
effect is robust to matching by towns' Shia histories,
ports is coming from some
unobserved
long histories of Shia rule characteristicappear
actually of port townsto that less
make riot
likely to experience anythem more tolerant, such
religious as simply increased wealth or
rioting.
821

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
TABLE 6. Regression: Placebos and Robustness
A) Coe
# H-M Riots, 1 850-1 950 0.326 0.085 0.201 0.306 0.591 0.690
[0.549] [0.616] [0.573] [0.919] [0.807] [0.770]
R-squared 0.26 0.36 0.48 0.43 0.46 0.57
Any H-M Riot, 1850-1950 0.084 0.034 0.057 0.148 0.082 0.095
[0.117] [0.119] [0.116] [0.142] [0.120] [0.101]
R-squared 0.26 0.30 0.39 0.41 0.43 0.57
B) Coefficient on Medieval Port Silted by 1901 (O
# H-M Riots, 1850-1950 -1.308*** -1.245** -1.298** -1.187* -1.439** -1.375*
[0.417] [0.533] [0.564] [0.635] [0.671] [0.716]
R-squared 0.26
0.39 0.51 0.44 0.49 0.59
Any H-M Riot, 1850-1950 -0.272** -0.233* -0.198** -0.203 -0.201 -0.096
[0.111] [0.111] [0.086] [0.126] [0.144] [0.150]
R-squared 0.27 0.31 0.40 0.41 0.44 0.57
C) Coefficient on Medieval Port (2SLS)
# H-M Riots, 1850-1950 -3.938 -3.550* -2.056 -3.363* -2.374** -2.118**
[2.531] [2.005] [1.421] [1.979] [1.034] [0.966]
Any H-M Riot, 1850-1950 -0.253 -0.657 -0.359 -0.240 -0.637 -0.648*
[0.543] [0.526] [0.298] [0.370] [0.415] [0.333]
Sample Full Coastal, Coastal, Full Coastal, Coastal,
<200 km < 1 00 km <200 km < 1 00 km
Controls Medieval Medieval Medieval Medieval Medieval Medieval
Province/NS x Annex FE No No No Yes Yes Yes
Observations 248 110 89 248 110 89
Notes : Each cell represents a regression. A

Distances from the Modern Coast, Naviga
Mughal Mint, Other Patronage Center, Inlan
Robust standard errors (clustered at Native
human capital accumulation, we should expect Persistence and Change, 1850-1995

colonial-era overseas ports to have lower colonial-era
violence. As panel A suggests, however, status as a I now examine the relationship between a medieval
colonial era port seems to have no relationship with trade legacy and the time series variation in ethnic riots.
the propensity for colonial-era violence. Figure 3 compares the timing of the first failure of in-
It still might be the case that medieval ports enjoy tere thnic peace among the same sample of towns, using
unobserved differences in wealth stemming from their data on religious violence through 1995 from Varshney
location at harbors that are not captured by colonial and Wilkinson (2004). Notice first that most towns in
port records, such as profits from smuggling. Panel the sample analyzed above experienced at least one
B compares the sample of medieval ports that subse- riot over the 145-year period, including a majority of
quently silted up and became inaccessible to overseas medieval trading ports. Though tragic, the latter finding
shipping. Though there are only 13 silted medieval adds confidence that the sample of medieval ports were
ports, the effect of silted ports on the probability of
religious violence is remarkably consistent in magni-
tude with that of all medieval ports. Thus, the transmis- log. municipal income per capita. Medieval ports, like other me-
sion mechanism that links a medieval trading legacy to dieval towns and Mughal mint towns, also continue to have a greater
contemporary religious relations does not appear to be proportion of Muslims. However, despite being poorer and more
ethnically mixed, medieval ports exhibit lower violence. Once again,
through unobserved modern trade.
as the placebo checks reveal, these effects are specific to medieval,
Panel C presents instrumental variable results, com- not colonial, ports. Despite the endogeneity of ethnic demography
paring towns that became medieval ports because they and population, assessing the residual effect of historical robust com-
were close to indentations on the medieval coastline to plementarity controlling for these factors may be of interest. Online
Appendix Table 2 adds controls for population in 1901, polynomials
towns that would have been medieval ports had they
in the percentage Muslim in 1901, and their interaction with medieval
enjoyed such proximity. This comparison yields results
port. The table reveals that more larger, ethnically mixed medieval
that are broadly consistent with the average treatment
ports actually are less prone to religious violence. These results are
effect in magnitude and direction.33 consistent with the persistence of interethnic complementarity in me-
dieval ports: when a minority population specializes in a complemen-
tary service, increases in its population tend to increase intraminority
competition and improve the terms of trade for the members of the
majority, reducing incentives for expropriative violence. The size
33 Online Appendix Table 1 estimates the effect of a medieval trading
legacy on indices of income and religious demography. As the table
interaction also reassures that our results are not driven by possible
demonstrates, a medieval trade legacy has a negative effect on under-reporting
the in small medieval port towns.
822

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
FIGURE 3. Timing of the first failure of religious
Note : This Kaplan-Meier curve compares the relative survival rate of to
also at risk for violence earlier

migration ofin
their time as well,
Muslim populations, and not
even if this re-
different in unobserved waysflected exthat
ante worsesimply
interethnic relations,
made are likely
conflict
impossible or unreported.to However,
face lower ethnic violence
at after
allthe Partition.
times Simi-ethnic
tolerance in medieval ports have
larly, those enjoyed
that attracted a better
larger inflows of Hindu andsur-
vival probability than nonports,
Sikh refugees forwith a of
a given level medieval
Muslim out-migrationtrade
legacy increasing the survival
might face of greaterpeaceful co-existence
interethnic competition and ethnic
by more than 10 times prior
violence. to
Though1950 when
these flows controlling
are arguably endogenous,
for geographical and historical characteristics
it is interesting to examine how these processes(Onlinemediate
our results. Columns 5is
Appendix Table 3). The difference and 6most
find evidence consistent
remarkable
in such periods as the emergence
with both these patterns,of mass
and reveal religious
that medieval ports
have around
politics in the wake of the three fewer riots
Khilafat relative to other townsin the
agitations
with similar pre-Partition
1920s and the turmoil of Partition in 1947 populations
thatand Partition-
steadily
resulted in the failure oferareligious
population flows. tolerance in other
towns.34 The effects above may be influenced by outlier towns
Table 7 uses an expanded sample of Indian towns with large degrees of rioting. Columns 7-9 exploit the
existing in 1950 to trace the changes before and after panel variation between 1950 and 1995 explicitly, as-
1950, comparing towns within modern Indian states. sessing the probability that a medieval port experi-
While medieval ports do have more riots than prior to enced any Hindu-Muslim riot in a given year, rela-
Independence, they continue to be half as riot prone as tive to otherwise similar towns in the same year and
otherwise similar towns (columns 1 and 2) and, just as state, with and without controlling for Partition flows
before 1950, have between two and three fewer riots and pre-Independence population and demographics.
between 1950 and 1995 (columns 3-6). One key dif- We allow for arbitrary correlation across years, within
ference between the two periods is the forced migra- states. Notice that a legacy of medieval interethnic com-
tion of 17.9 million people, mainly minorities, during plementarities reduces the probability of any riot in a
the Partition (Bharadwaj, Khwaja, and Mian 2008; Jha town in any given year by around three percentage
and Wilkinson 2012). Somewhat paradoxically, towns points. Once again these effects are larger in larger
that experienced the most ethnic cleansing and out- towns and in towns with greater Partition-era losses in
their Muslim population.
India underwent key economic changes between
34 One important shock also experienced just prior to Partition 1950
was and 1995 as well, with increased market inte-
the shock to private organizational skills gained by India's com-
gration in commodity and product markets, as well
bat troops during the Second World War. Yet, Jha and Wilkinson
as to a lesser extent in the
(2012) find that though districts that raised troops that acquired labor market (Online Ap-
more combat experience tended to have greater ethnic cleansingpendix
of Figure 3). Insofar as the convergence in agricul-
tural
their ethnic minorities, districts with medieval ports that experienced harvest prices to the national minimum reflects a
district's access to national markets, we can examine
such organizational shocks reveal significantly less ethnic cleansing
during the Partition. This is once again consistent with the presence
whether differences in product market integration also
of persistent inter-ethnic complementarities, as both members of the
majority and minority would lose from ethnic cleansing in suchalter an the effects of local institutions. Similarly, towns
environment. surrounded by districts with relatively low agricultural
823

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
* * * *
***** *
!_
!_ «õ oT-J
0 >¿>0 OOOOOOOOOOOOOOOOOOOOOOOOOOOOOCDo CT) 1- ^ b -o"
¿dddòoddddddòdòdddodoòdodc)òodoo.2)r" f/c 1/5
i- - -i- - m <8=ř f/c
S
POPO c . CDS ï
-
0 (Ö W '0
-
-
c
*
CD
*
.9
h:
o
21
0 * *
*r SS'Q ~j WrC
LL r? 0
LL
^ =
000000000000000000
^ ' £ S' O ©
C .9
1 ?wf
2 ř- co CM ^
pij 5 8
~o oo o "■ ï: « s«
> CO CO
S>2>
Z . D)
E - c .g
* ** ^
Tt
co ^ cMoot-0500^Noo^comoocqiniq^ t o CM
ro 00^ NOSO)î-i-^0O)^t0SCOC3)mT-OÇ2 ^
u
Tt Tt CM
r^ciò cvi^coòdd^òdò^cModdòòò o O v
cm O V ^ ^ ^ Oc O
* Ä T3 >.
*
ín J cr> lo r» ifi^CNjcDwôœifiœ^wor-r-cococû co - p >- __ Z o ^ c

ÄC0 OJ LO
iü-Q en CO TtCMCMÔOCMOT-OÔOOOOCOr^CM COLL E .9

cviöö CO i- Ç'Í Ò O Ò Ò Ò O Ò C'Í -r- Ò O. Ò O. O O. O P 0" =3
I 'I III M- -0
w o 5;
0 rr Œ 0
0 o rr Œ DC
* i - i i - i CO -D E G CO c
m 0 c
CO O C'J CO CO m (On ^ w 10 ffl
s ^Í3 RS88 O C
SE O'l Tdö
O'l O ci£ (O
°Ov " ¿in
n* (O '
¡5
_Q
-lë T3
O „
n*
. „
c c *- ^
c | „ c CO j» ^ 0
Z ^c/) S o co En cr>=
co - i co co i- ej) r ,? ^ ^zo ^qj
10
o> - co O
I
- r-'dd
' I
' i-
co
- O ro)ü5
° ^ ^ | 0 2
o>
T- -1 <
¿ r- ^ * * 5 « CO
C c ¿ o CT>
2
■O
IO
J-Soj^íb;:. ^8>z®
00
T-
(0 -§>£
Z - O
%ņ
o o o
-go
ti?
0
"■5
c
® +sV)
c 3
c ,Ç= tfT ï ï w -= w
o
cû e?)
9 e?)
^ ^ co
co ÍL
-j =
s-m^co
° rro t.
t.
-§>£ SS e?) co 58S ¿ -j s- ^ ° il t. S
"O
c
0) Z öö öo
Q. 1-1 1-1 - CO
d) o C -o
"O Q. ^ C
o o i2 iS
ç ■¿= ^ ^ ■ CO
o ? Ž I ^ fi fts
0) > ¿vE11 co .9? E
c
s «
2 iOS
ê -9
05X Q)
I EEE Ç
> O-0ÍT
0^-ri
e «Ic^l^ ScSoI^g. Ií2c
c
0)
"C "C Ï=~i5 Ï=~i5
û- |||S05«|ag¿|
jyj p E Ö -= û-S*
jyj^ ^œgg|
co E
E
sI s S|csi il- §S oId. U^£5«= !jy1

o £ "C £ "C s <5 Ï=~i5 ! ^ I i i
ir
g = sfig»sai£gļg^|
g ¿§.2sss°ssssssg5ix<0. -g£î
LU ÉO g î - ! X X X X £ ! .
J
CD «1 |ül
_É_
řl gœ
řl gœct^S
<?2œ œ S 'S ! S S ï ?ÍS|Î|Í 55
55?sSsS2S22x<CL222a:w 55^ Ô2|2<
2|2<
824

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
TABLE 8. Regression: Interactions with Mediev
Probability of A Riot Number of Riots
OLS: Years 1850-1950 (1) (2) (3) (4)
Medieval Overseas Port -0.235*** -0.053 -1.330** -1.478

[0.058] [0.097] [0.518] [0.894]
Centuries Muslim Rule (to 1707) 0.041 0.044 0.169 0.166
[0.033] [0.032] [0.123] [0.126]
Town under Muslim Ruler (1850-1947) -0.130 -0.127 -1.095* -1.098
[0.145] [0.147] [0.628] [0.629]
Town under Hindu/Sikh Ruler (1 850-1 947) -0.097 -0.096 -0.848** -0.850**
[0.081] [0.077] [0.403] [0.407]
Medieval Port x Centuries Muslim Rule 0.078** -0.063
[0.031] [0.250]
R-squared 0.27 0.28 0.27 0.27
Sample Full Full Full Full
Controls Medieval Medieval Medieval Medieval
Province/NS x Annexation FE No No No No
Observations 248 248 248 248
Notes : All regressions include quadratic

Coast, Navigable Rivers and the Gang
Other Patronage Center, Inland Trade R
interactions are demeaned. Robust stan
10%; **5%; ***1%.
wages in a given non-Muslim

year- rule rather suggesting
than in places where minor-
mand, high labor ity traders
supply, could rely on theor backing of a minority-
greater
migrant labor in controlled thestate or onagricultural
British third-party enforcement.
see greater in-migration, Table 8 assesses the cross-native state which and province co
local institutional variation differences.
in riots between 1850 and 1950, adding con-
Indi
ductivity was also trols for whether a town was under a Hindu
deeply or Muslim
affected
olution. As column native ruler (relative to direct British rule) between whi
9 suggests,
years that are 1858 more and 1947, as wellintegrated
as an interaction between the fa
probabilities of medieval port effect and the duration of Muslim rule
experiencing viole
fect of medieval interethnic trade remains robust to prior to the fracturing of the Mughal empire (1707).
Though British towns appear as likely as otherwise
accounting for variation in these factors. Furthermore,
similar native state towns to experience any rioting,
the reductions in the probability of medieval port vio-
native-state-controlled towns appear less riot prone.
lence appear greater relative to other towns in districts
and years characterized by low agricultural wagesFurther,
and consistent with the substitute role of state en-
less diffusion of Green Revolution crop varieties. forcement
It is and the incentives to build local institutions,
in hard economic times, and periods of lower develop- having one century less of Muslim rule strengthens the
medieval port legacy, lowering the probability that a
ment, that local institutions seem to be relatively more
important. medieval port will experience any riot by close to eight
percentage points.
Similarly, in democratic India, the panel variation in
Political Incentives as Substitutes for Local the yearly probability of a riot allows us to explicitly
Institutions of Peace consider whether the medieval port legacy changes in
its effect depending on the electoral incentives that
Beyond economic conditions that might accentuate state-level
the politicians face to protect minorities, as em-
importance of local institutions of peace, it isphasized
likely by Brass (2003) and Wilkinson (2004). In
particular, Wilkinson (2004) argues that minority vote
that the incentives to develop informal, local institu-
tional systems that support tolerance between blocksHindus will be more likely to be pivotal in states with
and Muslims in medieval ports would be greater largeindegrees of party competition, reducing incentives
environments where the political incentives forfor politicians to foment ethnic violence for electoral
rulers
gain. While party fractionalization could be endoge-
and the State to protect the minority population them-
selves were also weaker. In particular, we should nousex- (and in fact, states with medieval ports have
tended
pect that the institutional legacy should be greatest for to have had more effective parties- see Online
ports where the need for them was greater- for Appendix
ex- Figure 4), if we assume that individual towns
ample, where Muslim traders existed for longerhave little impact on state-level party competition, we
under
825

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
TABLE 9. Regression: Interactions with Democratic Political Incentives, 1950-95
Any H-M Riot in Year? (OLS) (1) (2) (3) (4)
Medieval Overseas Port -0.034** -0.034** -0.034** -0.035**

[0.016] [0.015] [0.015] [0.015]
# Effective Parties -0.002* -0.002*** -0.001 -0.002
[0.001] [0.001] [0.002] [0.002]
Log. Days to Next Election 0.005 0.005 0.007 0.005
[0.044] [0.043] [0.059] [0.059]
Med Port X # Effective Parties -0.003 -0.002 -0.010 -0.009
[0.002] [0.002] [0.009] [0.009]
Med Port x Log. Days to Election 0.104** 0.104** 0.165** 0.165**
[0.046] [0.045] [0.071] [0.071]
Sample Independent India Effective Parties < 3
1950-1995
Effective Parties incl. Independents? N Y N Y
State (1991) and Year Fixed Effects? Y Y Y Y
Observations 16252 16252 13267 13061
R-squared 0.08 0.08 0.07 0.08
Notes: All regressions include quadratic p

Navigable Rivers and the Ganges, Coasta
Center, Inland Trade Route, Skilled Crafts
yearly averages. All regressions also includ
de-meaned. Robust standard errors (clust
can examine how the

religious medieval
institutions that promote a version of port
Islam
environments when politicians
less adapted to domestic norms and traditions.36 have
Within
to protect their India, this period saw the emergence
minority of the Hindu na-
populations
gests, consistent tionalist
with Bharatiya Janata Party (BJP) (Hansen(2004
Wilkinson 1999)
with a greater average number
and the playing of the "ethnic" of e
card by Indira Gandhi
the state level are
(Kohli 2012).less likely
These factors to
actually coincided with afac
However, controlling for
drop in party competition in states these
with medieval ports reg
(Online Appendix
dieval ports continue to Figure 4), increasing statethe
reduce incentivespr
in any year by around
for riots that may havethree
overwhelmed localpercen
institutions.
so even more inOnline the Appendixrun-up
Table 4 breaks down to the riotelect
effect
These results continue to hold if we include the vote over 1950-1995 into those that occurred before and af-
shares of independents- which could be more easily ter 1980. While this reduces the variation in the data-
influenced by a single town- in our measure of party there are naturally fewer riots- the point estimates are
competition (column 2). In fact, the relative reduction
fairly consistent before and after 1980: medieval ports
continue to be half as riot prone as other towns after
in violence in medieval ports closer in time to the next
elections increases from 10.4 to 16.5 percentage points
1980, and experience around 1.44 fewer riots, particu-
if we only consider towns in state-years with lower larly when comparing towns within the same state and
party competition (columns 3 and 4). with the same levels of Partition flows.37
Year fixed effects allow us to compare towns within During the outbreak of rioting in Gujarat in Febru-
the same year, but may also be absorbing potentially ary, 2002, Gujarat was also experiencing a period of low
interesting political trend breaks. 1980, in particular,
party competition and faced impending elections (On-
is sometimes seen as a watershed moment in Indian line Appendix Figure 4). In Jha (2013), I exploit data
from
politics, as the end of the 1970s brought a confluence of the two months of religious violence in Gujarati
exogenous and potentially endogenous political andfollowing the Godhra burning. I show that while
towns
economic changes. Indeed, as Figure 3 suggests, atowns num-that were historic focuses of interethnic compe-
ber of medieval ports that had never experienced a like Ahmadabad, were marginal constituencies
tition,
Hindu-Muslim riot before did so following 1980. Out-
prior to the riots and more affected by violence during
side the country, the OPEC oil shocks fueled a rel-
ative increase in funding by certain Gulf states for
36 Indeed, such new externally funded schools were among the only
new buildings in relatively economically stagnant medieval port
towns like Veraval as well playing an increasing role even in economi-
35 The number of effective parties is calculated at the state level ac-
cally vibrant port towns like Surat as well [author's field observations,
cording to EP = , where v, is the vote share of the ith party. This
2006-7].
37 A more comprehensive assessment of all the many relevant
measure weights parties with a higher vote share more heavily than
those with a very low vote share. I am grateful to Steven Wilkinson
changes that occurred in this period is beyond the scope of the article
for sharing these data. but is a focus of continuing research.
826

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
FIGURE 4. Muslim Occupational Shares and Relat
specialization
those two months, Gujarati and that supported
medieval ports sharing
inmechanisms
general
between
acted as "oases of peace"- groups. Household
these towns level evidence25
were can shed
per
centage points less likely than
important otherwise
light similar
on whether the institutional town
differences
in occupational
to experience ethnic rioting in that choices, interethnic
period. In inequality,
medieval organi-
port towns, like Surat, zational
thatstructures, and interethnic trustsome
did experience that we noted
vio-
lence, the riots themselves
in our were also
case studies less ports
of medieval extensive
like Surat and and
other medieval towns
severe. The electoral consequences forlike Ahmadabad
the BJP are general
state
government, seen as complicit
phenomena, andin the
to shed riots,
further were
light on how also
they inter-
very different in medievalact with political incentives.
ports: whileThe theurbanvote
sample of the
shar
Indian Human
for the BJP rose significantly Development
from Survey (2005)
43.43% in 1998provides t
48.92% in the elections immediately following
information on 14,820 households the
in 989 primary 200
sam-
riots, sufficient to win the
pling BJP an
units in 27 additional
Indian 13 towns
states sampled from seatswithin
probability proportionate
nonmedieval port constituencies, medieval to their population.
port voters Since
the townby
reduced the BJP vote share identifiers
6.77 arepercentage
embargoed but district iden-
points
compared to other towns within
tifiers their
are not, I compare urbanown districts,
households in districts
enough to cause the BJP with medieval
a loss of towns as their seats,
three headquarters to other
and ren
districts, as well
der medieval ports the focuses ofas electoral
doing a "difference-in-difference"
competitio
in subsequent elections. comparison
Thus Gujarat-of Muslim and non-Muslim urban house-
and democratic
India more generally- appear
holds withinto mimic
medieval medieval
port and nonport India
districts.39
The Duncan
it is when State political agents Occupational
have lower Segregation Index ofto
incentives
protect minorities that Muslims
local and non-Muslim workers
institutions in medieval ports
become mos
relevant.38 (0.243) and nonports (0.225) suggests that urban
Muslims are occupationally segregated in general,
Household Evidence on the Institutional and those in medieval ports are only slightly more
Mechanisms specialized.40 However, as Figure 4(a) reveals, urban
Muslims in medieval and nonmedieval port districts
The theory and cases suggest two commonalities specialize
in me- differently, with Muslim residents of me-
dieval
dieval ports- that institutional systems emerged ports concentrating in a manner consistent with
that
coordinated new forms of complementary economic
39 Insofar as I associate nonmedieval port PSUs with medieval ports,
38 In fact, Jha (2013) argues that Gujarat's experience is and miss urban households in medieval ports that are not district
consistent
with electoral competition creating state-level contagion ofheadquarters,
local in- I will be underestimating the effects of a medieval
port legacy.
centives for both conflict and peace: when towns with interethnic
competition become marginal constituencies and thus more 40 The polit-
Index is I = ' II mi - fell» where ra, is the proportion of
Muslimstowns
ically influential, state-level politicians seeking to win these in occupational category i, and kt is the proportion of non-
for their parties have greater incentives to adopt platformsMuslims. Complete segregation would yield a DOSI of 1 while in-
closer
to local populist positions that may lead to enhanced ethnictegration
mobi- would yield a DOSI of 0. Using the Colonial Gazetteers
lization and reduced opposition to ethnic violence throughout theconstruct equivalent numbers for the nonagricultural
(1899), we can
state. In contrast, when towns with intercomplementarity become
populations of British Gujarati districts. Interestingly, the DOSI for
marginal constituencies, their preferences for tolerance may
both alsoport districts (0.174) and nonport districts (0.136) was
medieval
be transmitted through the state government's platform, reducing
actually lower in 1899. The increase may partly reflect the population
violence elsewhere as well. transfers that would occur during the Partition.
827

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
complementary mechanisms that continue tosocial organizations (columns B7-9). The remarkable
support
trade. In fact, despite making up around 18% of theof the Bohras, born from medieval trade,
organization
population in medieval ports, Muslims comprise close
does not appear to be unique to them but rather finds
to 40% of small traders in medieval ports, and 26%among communities across India.
resonance
of organized businesspeople. In the absence of Panel strong C compares trust and political participation
legal enforcement, such professional choicesamong canMuslim
be households in 2004 in states that had ex-
seen as a behavioral choices that also reflectperienced
greater weaker party competition between 1992 and
trust. In contrast, Muslims in other urban areas 2002. Notearethat urban Muslims were as likely to vote
disproportionately artisans and laborers. as Hindus in states that had sustained high party com-
The development and persistence of organizations petition, but their participation is significantly lower
and mechanisms to facilitate between-groupinsharing states with low party competition (columns Cl-3).
should also be reflected in reductions in between- Along with voting more, Muslims in high competition
states
group inequality. Figure 4(b) shows the distributions of are also less likely to express "no confidence" in
household wealth by religion. Relative to other the urban
state government (columns C4-6), consistent with
Muslims, Muslims in medieval ports show strikingly the argument by Wilkinson (2004) that party competi-
less within- and between-group inequality. Table tion 10
leads Muslims to be more likely to be politically
(Al-2) confirms that these relationships survive pivotal
theand thus to gain state protection. However,
inclusion of state and district fixed effects. Though
despite being extremely distrustful of the state gov-
ernment, Muslims in medieval ports in states with low
households in medieval ports are not richer on average,
party
and Muslims in general possess fewer assets, Muslims incompetition continue to be more likely to take
the polio vaccine (column C7-9). Once again, local
medieval ports possess similar wealth to non-Muslims.
It is good to reiterate that absent complementarity
institutions appear to support interethnic trust particu-
and sharing mechanisms, relative wealth accumulation
larly when political incentives to protect minorities are
weak.
by a weak minority group may be likely to raise, not
lower ethnic violence (Mitra and Ray 2010). However,
despite being richer, Muslim households in medieval
ports were seven percentage points less likely DISCUSSION
than
other urban Muslims to report "a lot of conflict" in
their neighborhoods in 2005 (columns A4-6). This article has sought to establish that interethnic me-
Survey measures of trust naturally face drawbacks
dieval trade has left a lasting legacy on violence. In the
ports of the medieval Indian Ocean, Islam, by making
from relying on self-reports and hypothetical scenar-
trade accessible to all Muslims, satisfied two conditions
ios. Beyond occupational choice, I examine a further
behavioral measure of trust: the decision to vaccinate that support peaceful co-existence over time: the provi-
against polio. A surprisingly resilient meme that po-sion of a nonreplicable, nonexpropriable complemen-
lio vaccinations are a vehicle for minority male steril-
tary service and a means to more equitably share the
ization appears to have led to a rise in polio deathssurplus from trade. These conditions appear to have
in India in the 2000s. In fact, Jha and Mahajan laid
(in the basis for an enduring legacy of ethnic tolerance
in South Asia.
progress) find that deaths due to polio fall disparately
on one subgroup: Muslim male infants in urban ar- The existence of robust complementarities and a
eas. Muslims were twice as likely than Hindus to cite sharing mechanism are not, however, unique to the
a "lack of faith" as their reason to refuse the vac- Indian Ocean. Sephardic Jews benefited from valu-
cine. However, while Muslims in other urban areas able do
trading networks in the 15th and 16th centuries
appear less likely to vaccinate their children inthat ourrendered them welcome arrivals in Ottoman ports
sample, Muslims in medieval ports appear as likely in the Mediterranean. Like Muslim traders in South
as non-Muslims to vaccinate their children (columns Asia, their complementarity stemmed from nonexpro-
A7-9). priable, nonreplicable advantages in trade- links to the
A further piece of the puzzle lies in whether or- emergent Atlantic economy otherwise not available to
ganizations in medieval ports do in fact differ. Panel Muslims in the Ottoman empire. Their systematic ex-
B provides household comparisons of organizational pulsion from Spain meant that they also naturally sat-
membership. Note that medieval port households are isfied the second criterion- they came in large enough
close to nine percentage points more likely to be mem- numbers to foster within-group competition, with their
bers of business groups and trade unions, and seven immigration actively encouraged by local Ottoman au-
percentage points more likely to join credit and savings thorities. The city of Salónica in particular attracted a
groups. Though, Muslims in general are three to four large number of Jewish refugees (Mazower 2005). A
percentage points less likely to join business groups, combination of ethnic occupational specialization in
Muslims in medieval ports remain four percentage complementary activities and permissive immigration
points more likely than nonmedieval port households that facilitated within-group competition appears to
to participate in these organizations (columns Bl-6). have resulted in a long history of peaceful ethnic co-
Muslims in medieval ports also are around 11 percent- existence (Benbassa and Rodrigue 2000). For the next
age points more likely than non-Muslims in medieval four centuries, Ottoman Salónica, sometimes referred
ports- and around nine percentage points more likely to as the "Mother of Israel" was seen as the most tol-
than Muslims elsewhere- to participate in religious and erant place for Jews in Europe, with Jews specialized
828

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
* * *
** * * ** , . u
_ . O) ^Fo fôTcbo' .2 bô^ ^ -J2
oT _ . OO-f-Ort C'1 O 00 CD y ^ i-Oi-O CM ¿5 ^ S3 73
qqqoq qqqow^ y ^ w 0000 o .ço V ç
o 00000 1 00000 " - ' b 0 öoöö " " ó Q "co
ç I 1 O) " - ' c " I " C
8 o g 8
-§ CO •§ co
** "0** O)
O *^ , ,- ,, ,*,, (/)
- , OO)
o * i i- -1 1* WO)
1-1 "ft
1-1y "ft
o * y
, -'■=
,* , C'J
- , "t"ť
* , -I ,* CO , 00 - , .y_0
Û. i3-
Û. P ^0000 O ï-^ OOOOCO^W
00000 O CO ^ (/)2!CM
-M O
ŪLO) CO O
OOOO (Ow2!
A -M
<P P i- O i- O ¿ ^
"o 00000 1 1 2 òdòdo Q p 0000 ob in^
0 1 - 1 1 i 1 - 1 O 1 - 1 1 - 1 0 • - > i ■ - ■ CO
§ 0 1 - 1 1 i 1 - 1 g> O 1 - 1 1 -
"55 ® •$ <no
^ Œ „i „ œ i „i „i „ „ 17
_ b^-^-ûoLO loouo^tcviio „ $> oocm^-ôdlot-t- „ „ o | ^
íC ^ _ i-Oi-OÔi-
^ qqqqqqo o o oT-Wlí)i-OinO)7^
o 1- o 1- ~ OO1-O1-OOOO
oooooooooO<53o m
ò o, o o, o p,o dc^dc^dcDO ^ ò 0,0 0.0 0.0 0.0 . m v
"8 c
* ® O)
«Ü
* * E m
0)0000 CO LO 00 1- N CO 00 O •- CD ^
J5 co O T- CO CO IO i-OCMlOCOyi: > O t- 00 CO ■oTt CD
15 co o o o o co ooooï-^œ o o o i- o i- w V ¿
-c ddddd _ ^ . 00000 Q CD 0000 où £
O _ , ^ I-- <0 i 1-1 1-1
5 O S cņ
R
.sO)
to*
55
~1 O
S o
C'J
"£s
^ O)
i
O 00 O > CO LO O T- Tf O CO CT) _ CD C
*î= ^ LO O 1- CO CO LO CŪ > i-0(0lfic07¿ S CO CM LO Tf C'J ^3 °7 _ -pco
*î= § LO O O O O C0 </) CŪ O O O O co ^ OOOO 1- .co A ~ cg
U n ddddo _ ^ ^ doddd ^ _ û <d dodo ob <o> ^2
U0n _œ
, ^ ^ c
^ ^ _ s <o> c
S?
^2
-I ° *** 0 . 5 v r|
-I *** v
< _ CM ^ O 00 " IO CO CO CM CM jļ> -S CT) CO CO LO CM CM CO 2

in 3 ^ _ O CM O T- co CO O r^CMÔh-LOCO-T-nJ t-CMOOt-COIÔOO£ = d- -Q
o 3 qqqqoor ooooooo^-^ 0/3 oooooooooO< ļp ^
o 0000000 0000,000 0/3 o 00 0,0 0,0 0.0 *^g)
11 1 o|
CM
uT
^ CO
O
* "ort ™ C
> * ^ ™ O
(0 ONCOO CON-CMCO
-C
0) Vco
1 zSzS OOOOt-^ÇO
CO O t- O O) yy¿ OOOO
CO T- CM CT) V
T-W 0 1:^
g ^ g o
o
CD 2- g dddddb dodo ob eco
"5
o 2
8I 3 3 ??
S eco
-è
" „ -fi
•M
£ s>^ T- f^- T- CM £ f^r^.N.O) C ÔOON „

■O
Í8 ^^ CÑT
^ LOO^tCMh-l-
OOOOCM"^ CO O O CM 00
OOOOt-^ÇO 7 i: -
7 "Ö Oi-OCMi-ÇOA
OOOO 00 °? ->cO
^
c 2: ddddo w ddddo bB dodo ob q.11
(0 2O i ^ ©c
i ' >
i P co
i " i " >~ë
(0
c 0 ^ "8 cg
o 3 „í „i „ m i „i „ „
*5 0 N CO T- 00 Ò „ CM ^ CM „ 00 T- „ LO © lOOCOOOOOOCOCO © ^ 3 to
X ^ r- t- N O N CM CM COCM-Ôi-COCM^cO O CM LO i- CO ^ i- O O 3
S ^ O O O O O O 1- OOOOOOT-^-h~ 0000000000< Û O)
c 0000000 0000000 0/5 ddddddddo2 «5
(0
ff
~ " " |l
0 c "o S
T30 2 ^CmB^-O
0) ä O % o oP-Sc®
3
° ä O 'S .S- O o _ tó O
È
- S o r -g
o =2|i _ tó ^ s -si
o c 1^ 5 ^ E ž
1
in co p co Eôw E0 co^
3cq2 in co = p 75QP co = ôQigJir E0 £. co^ « ü
(0
i - co _c 3 Q co 3 "ČŪ J CO 3 ^ ïrû.®
O S£oSx 2 O 3 X 5 « > O s X > £ $ S0-
LU
m I ļ i r ^ Č I S r 'S?! 8 ^ Ž t Ì "S|< if
-I
m
1 m « ļ ! i E s r ^ S Č I I ! S E s r s|S| ^ I I Ž fi t I ļSā
< CO "O CO -O 5¡"0 ~0 CO "O £" =3 0 û. TD CO "D ^ 5" 0 E K "P
< _ _l 0 D 0 <P 0 3 0 Ô^.. =3 0 © © ^ S ^XmÛd) 0 "P
cr X il o S 2 2 -í cr il co īž
_é_
829

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
in commerce (Mazo wer 2005). Complementary or organization

Jewish also exists to equitably share the gains
trading networks appear to have facilitated from trade with local communities.
tolerance
While
elsewhere as well. Consistent with and building the trading networks of the Chinese in mod-
in part
ern Indonesia
on an earlier version of this article, Voigtländer andand South Asians in modern East Africa
also made
Voth (2011) find that German overseas trading portsthem valuable to the local population, these
that were part of the Hanseatic League were groups have tended to lack such a sharing mecha-
also less
likely to show persistent anti-Semitism. nism. Chinese and South Asian ethnic trading net-
works,
The logic underlying peaceful coexistence based upon personal and community ties, were
between
Hindus and Muslims in medieval ports and theclosed to competitors, and thus relatively small minor-
support-
ing institutions that emerged can be readily applied
ity to able to capture much of the gains from
groups were
tradenon-
other historical and contemporary settings where (Rauch and Casella 2002). This arguably ren-
local and local ethnic groups co-exist, both dered these minorities increasingly attractive targets
to under-
for ethnic
stand why ethnic tolerance fails, and how tolerance may violence and susceptible to expropriation by
"strong"
be fostered. The theory and evidence described locals.
here
In fact,
suggests that ethnic violence is more likely when ethnica perverse "ethnic cronyism" equilibrium
groups compete, when the source of interethnic may develop.
com- Commercial minorities in Kenya and In-
plementarity is easy for one group to expropriatedonesia arguably
or possessed incentives to buy "protec-
tion" from local rulers. These transfers not only may
replicate, or when no mechanism exists to redistribute
the gains from trade. Competition betweenhave potentially
locals and undermined democratic institutions
immigrant groups for jobs has often beenand cited as a
consolidated dictatorships, but also perversely pro-
reason for ethnic tension in the United States vided(Olzak
local leaders incentives to intermittently facili-
1992). The theory above suggests that these tate tensions
pogroms to demonstrate the need for their pro-
tection (Jha 2009). Yet, even in East Africa, one South
are most likely to arise in jobs that are unspecialized
and require either few or generally available Asian community
skills or stands out in the degree of ethnic
tolerance
inputs, since these are the least costly for locals to enter.it enjoys. The Ismaili trading community in
Yet, even nonlocal minorities who do not compete,East Africa,butmany of whose members can trace their
enjoy complementarities that stem from tangible backgrounds as- to Indian medieval ports, have developed
sets, such as land, machines, or other forms of organizations
physical that foreswear political bribery and in-
capital, may be targets of violence. stead share the gains from trade through provision of
Being impossible to violently expropriate, localspecial-
public goods. Long after the decline of Muslim ad-
ized skills do provide a better basis for interethnic
vantages in overseas trade in the Indian Ocean, it may
complementarity and tolerance, but even be that both
these can the institutions and the ideas medieval
be replicated in the longer term. Minoritiescommunities
that have developed to support ethnic tolerance
specialized skills can become increasingly attractive
may continue to yield dividends today.
targets of violence if locals become able to duplicate
those skills. The forced expulsion of Jews from Spain at
Supplementary materials
the end of the 15th century was precipitated in part by
prior conversions, both forced and voluntary, of Jews
To view supplementary material for this article, please
to Christianity. These "new" Christians provided the
visit http://dx.doi.org/10.1017/S0003055413000464
administrative skills to Spanish rulers for which they
previously depended on the better-educated Jewish
REFERENCES
population (Benbassa and Rodrigue 2000; Botticini
and Eckstein 2012). It is possible that the expansion
Acemoglu, D., S. Johnson, and J. A. Robinson. 2001. "The Colonial
of public education in Western Europe and the United
Origins of Comparative Development: An Empirical Investiga-
States may also have had the unfortunate side tion." American Economic Review 91 (5): 1369-401.
effect
Alberuni, A.-R. M. I. A. 1030. Indica. Munshiram Manoharlal, Delhi,
of raising the likelihood of violence against educated
2005 ed., edited by Sachau Edward C.
minority incumbents in skilled jobs by rendering them
Alesina, A., and E. La Ferrara. 2005. "Ethnic Diversity and Eco-
more replaceable by locals.41 nomic Performance." Journal of Economic Literature 43 (3): 762-
In contrast to physical and human capital, 800.
how-
ever, most ethnic trading networks are bothal Malibari,
diffi-Z. a.-D. 1528. Tohfut-ul-mujahideen. Oriental Transla-
tion Fund of Great Britain, London, translated from the Arabic
cult to steal- being intangible- and extremely costly
by M. J. Rowlandson in 1833.
to replicate. Because there are network externalities-
Arasaratnam, S. 1994. Maritime India in the Seventeenth Century.
the value of a trading network increases with theOxford
Delhi: size University Press.
of its membership- there will be high costs Baldwin,
for K.,
any and J. Huber. 2010. "Economic Versus Cultural Dif-
ferences: Forms of Ethnic Diversity and Public Goods Provision."
local to invest in a set of personal exchange relation-
American Political Science Review 104 (4): 644-62.
ships that would attain the scale necessary Baneijee,
to competeA., and L. Iyer. 2005. "History, Institutions and Economic
with an ethnic trading network. Thus, trading Performance:
networks The Legacy of Colonial Land Tenure Systems in
India." American
can provide the basis for sustained ethnic tolerance in Economic Review 95 (4): 1190-213.
Barbosa, D. 1519. A Description of the Coasts of East Africa and
mixed communities, as long as a nonviolent mechanism
Malabar in the Beginning of the Sixteenth Century. Hakluyt Society,
London, translated from the Portuguese by Henry E. J. Stanley in
41 Indeed anti-Semitism in Nazi Germany was particularly1866. pro-
Baron de Montesquieu, C. d. S. 1748. Esprit le lois.
nounced among the relatively literate Protestant population.
830

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
Bayly, C. A. 1985. "The Pre-History of

Habib, 1. 1982. An Atlas 'Communalism'?
of the Mughal Empire. Delhi: Oxford Uni- Religious
Conflict in India, 1700-1860." Modern
versity Press. Asian Studies 19 (2): 177-
203. Habyarimana, J., M. Humphreys, D. Posner, and J. Weinstein. 2009.
Coethnicity:
Bayly, S. 1989. Saints , Goddesses and Diversity
Kings:and the Dilemmas of Collective Action.
Muslims and New Christians
in South Indian Society 1700-1900.
York: Russell Sage.
Cambridge, UK: Cambridge
University Press. Hansen, T. B. 1999. The Saffron Wave: Democracy and Hindu Na-
Benbassa, É., and A. Rodrigue. 2000.
tionalism Sephardi
in Modern India. Princeton: PrincetonJewry:
University Press. A History
Haynes, D. E. 1991.
of the Judeo-Spanish Community, Rhetoric and Ritual inCenturies.
14th-20th Colonial India: The Shap- Berkeley:
University of California Press.ing of Public Culture in Surat City, 1852-1928. Berkeley: University
Bharadwaj, P., A. Khwaja, and ofA.California Press.
Mian. 2008. "The Big March: Mi-
Hearn, P. 2001.
gratory Flows After the Partition of "Global
India." GIS Database. Digital Atlas ofand
Economic South Asia."
Political
Weekly 43 (35). electronic resource, US. Geological Survey digital data series.
Hirschman,
Blank, J. 2001. Mullahs on the A. O. 1977. The Passions
Mainframe: Islamand the Interests:
and Political Ar-
Modernity
Among the Daudi Bohras. London:
guments forUniversity of Chicago
Capitalism before its Triumph. Press.
Princeton: Princeton
Bohlken, A. T., and E. Sergenti.
University2010.
Press. "Ethnic Violence and Eco-
nomic Growth: An Empirical Investigation
Horowitz, of Hindu-Muslim
D. L. 1985. Ethnic Groups in Conflict. Berkeley: University
Riots in India." Journal of Peace Research 47 (5): 589- of California Press.
600. Horowitz, D. 2003. The Deadly Ethnic Riot. Berkeley: University of
Bose, S. 2006. A Hundred Horizons : The Indian Ocean in the Age of California Press.
Global Empire. Cambridge, MA: Harvard University Press. Ibn Battuta, A. A. 1355. The Travels oflbn Battuta. New York: Lenox
Botticini, M., and Z. Eckstein. 2012. The Chosen Few: How Education Hill, 1971 ed., translated from the Arabic by Samuel Lee, 1829.
Shaped Jewish History, 70-1492. Princeton: Princeton University Jha, S. 2007. "Maintaining Peace Across Ethnic Lines: New Lessons
Press. from the Past." Economics of Peace and Security Journal 2 (2),
Bouchon, G. 2000. "A Microcosm: Calicut in the Sixteenth Century." 89-93.
in Lombard and Aubin (2000), chap. 4, pp. 78-87. Jha, S. 2009. "A theory of ethnic tolerance," mimeo, Stanford Uni-
Brass, P. R. 2003. The Production of Hindu- Muslim Violence in Con- versity.
temporary India. Seattle: University of Washington Press. Jha, S. 2013. ""Unfinished Business": Ethnic Complementarities and
Campbell, J. M., ed. 1899. Gujarat Population- Musalmans and Par- the Political Contagion of Conflict and Peace in Gujarat," working
sis , vol. 9 of Gazetteer of the Bombay Presidency. Bombay: Govt. paper 19203, National Bureau of Economic Research, Cambridge,
Central Press, reprinted, 1990 edn. MA.
Casson, L., ed. 1989. The Periplus Maris Erythraei. Princeton, Prince- Jha, S., and A. Mahajan. in progress. "Complementarities, trust and
ton. the success of public health programs: Evidence from polio in
Chakravarti, R. 2000. "Nakhudas and Nauvittakas: Ship-Owning India," mimeo, Stanford University.
Merchants in the West Coast of India (C. AD 1000-1500)." Jour- Jha, S., and S. I. Wilkinson 2012. "Does Combat Experience Fos-
nal of the Economic and Social History of the Orient 43 (1): 34- ter Organizational Skill? Evidence from Ethnic Cleansing during
64. the Partition of South Asia." American Political Science Review
Chaudhuri, K. 1995. Trade and Civilisation in the Indian Ocean: An 106 (4): 883-907.
Economic History from the Rise of Islam to 1750. Cambridge, UK: Jukua, C. 1225. His Work on the Chinese and Arab Trade in the
Cambridge University Press. Twelfth and Thirteenth Centuries, Entitled Chu-fan-chi. New York:
Chua, A. 2003. World on Fire: How Exporting Free Market Democ- Paragon Books, translated from the Chinese by Friedrich Hirth
racy Breeds Ethnic Hatred and Global Instability. New York: Dou- and W. W. Rockhill in 1966.
bleday. Kohli, A. 2012. Poverty amid Plenty in the New India. Cambridge,
Dasgupta, A. 1994. Merchants of Maritime India, 1500-1800. Alder- UK: Cambridge University Press.
shot: Variorum. Landa, J. T. 1994. Trust, Ethnicity and Identity: Beyond the Institu-
Dasgupta, A. 2004. India and the Indian Ocean World: Trade and tional Economics of Ethnic Trading Networks, Contract Law and
Politics. Oxford, UK: Oxford University Press. Gift- Exchange. Ann Arbor: University of Michigan Press.
Lombard, D. 2000. "Introduction." in Lombard and Aubin (2000),
Desai, S., A. Dubey, B. Joshi, M. Sen, A. Shariff, and R. Vanneman.
2009. "Indian Human Development Survey." ICPSR 22626v2, chap. 1.
University of Maryland and NCAER, Ann Arbor, MI. Lombard, D., and J. Aubin, eds. 2000. Asian Merchants and Business-
Dippel, C. 2011. "Forced Coexistence and Economic Development: men in the Indian Ocean and the China Sea. New Delhi: Oxford
Evidence from Native American Reservations." mimeo, UCLA. University Press.
di Verthema, L. 1503. "The Nauigation and Voyages of Lewes Ver- Mazower, M. 2005. Salónica, City of Ghosts. New York: Alfred A.
tomannus, Gentleman to the Citie of Rome. Hakluyt, translated Knopf.
from the Latin by Richard Eden, 1576. Mehta, M. 1991. Indian Merchants and Entrepreneurs in Historical
Esteban, J., and D. Ray. 2007. "On the Salience of Ethnic Perspective: With Special Reference to the Shroffs of Gujarat, 1 7th
Conflict." working paper, NYU, available at www.econ.nyu. to 19th Centuries. Delhi: Academic Foundation.
edu/user/debraj/Papers/EstebanRaySalience.pdf. Miguel, E., S. Satyanath, and E. Sergenti. 2004. "Economic Shocks
Fearon, J. D., and D. D. Laitin. 1996. "Explaining Inter-Ethnic Co- and Civil Conflict: An Instrumental Variables Approach." Journal
operation." American Political Science Review 90 (4): 715-35. of Political Economy 112 (4): 725-54.
Fearon, J. D., and D. D. Laitin. 2003. "Ethnicity, Insurgency and CivilMilgrom, P., and J. Roberts. 1990. "The Economics of Modern Manu-
War." American Political Science Review 97 (1): 75-90. facturing: Technology, Strategy and Organization." American Eco-
Fearon, J. D., and D. D. Laitin. 2011. "Sons of the Soil, Migrants and nomic Review 80 (3): 511-28.
Civil War." World Development 39 (2): 199-211. Mitra, A., and D. Ray. 2010. "Implications of an Economic Theory
Gillion, K. L. 1968. Ahmedabad: A Study in Indian Urban History. of Conflict: Hindu-Muslim Violence in India."
Berkeley: University of California Press. Nunn, N, and L. Wantchekon. 2011. "The Slave Trade and the Ori-
Glaeser, E. L., R. La Porta, F. Lopez-de Silanes, and A. Shleifer. 2004. gins of Mistrust in Africa." American Economic Review 101 (7):
"Do Institutions Cause Growth?." Journal of Economic Growth 3221-52.
9 (3): 271-303. Olzak, S. 1992. The Dynamics of Ethnic Competition and Conflict.
Goitein, S. D. 1966. Studies in Islamic History and Institutions. Lei- Stanford, CA: Stanford University Press.
den: E.J. Brill. Penrad, J.-C. 2000. "The Ismaili Presence in East Africa: A Note on its
Greif, A. 2005. Institutions and the Path to the Modern Economy: Commercial History and Community Organization." in Lombard
Lessons from Medieval Trade. Cambridge, UK: Cambridge Uni- and Aubin (2000), chap. 17, pp. 223-38.
versity Press. Polachek, S., and C. Seiglie. 2007. "Trade, Peace and Democracy: An
Greif, A., and D. D. Laitin. 2004. "A Theory of Endogenous Insti- Analysis of Dyadic Dispute." in Handbook of Defense Economics ,
tutional Change." American Political Science Review 98 (4): 633- ed. T. Sandler and K. Harley. New York, Amsterdam: Elsevier,
52. 1017-73.
831

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00
Subrahmanyam,
Prior, K. 1993. "Making History: The State's Intervention in UrbanS. 1990. The Political Economy of Commerce:
Religious Disputes in the North-Western ProvincesSouthern
in the India, 1500-1650. Cambridge, UK: Cambridge Univer-
Early
Nineteenth Century." Modern Asian Studies 27 (1):sity179-203.
Press.
Subrahmanyam,
Qutbuddin, S. 201 1 . "History of the Da'udi Bohra Tayyibis in ModernS. 2000. "Introduction." in Lombard and Aubin
Times: The Da'is, the Da'wat and the Community."(2000), in A chap. 1.
Modern
History of the Ismailis: Continuity and Change in Thapar,
a Muslim R. 2004.
Com-Somanatha: The Many Voices of a History. New
munity, ed. F. Daftary, chap. 12. London: I.B. Tauris Delhi: Penguin.
Publishers.
Rauch, J., and A. Casella. 2002. "Anonymous Market Varshney,and Group
A. 2002. Ethnic Conflict and Civic Life: Hindus and Mus-
Ties in International Trade." Journal of International lims inEconomics
India. New Haven: Yale University Press.
(58): 19-47. Varshney, A., and S. I. Wilkinson 2004. "Varshney Wilkin-
Richards, J. E 1996. The Mughal Empire , New Cambridge son Dataset on Hindu-Muslim Violence in India, Version
History
of India. Cambridge, UK: Cambridge University Press. 2."
Risley, H., W. Meyer, R. Burn, and J. Cotton, eds. 1909. Voigtländer,
The ImperialN, and J. Voth 2011. "Persecution Perpetuated: The
Gazetteer of India, vol. various volumes. Oxford: Clarendon MedievalPress.
Origins of Anti-Semitic Violence in Nazi Germany."
Rodrik, D., A. Subramanian, and E Trebbi 2004. "Institutions working paper Rule:17113, NBER.
The Primacy of Institutions Over Geography andWade, Integration in
R. 1988. Village Republics: Economic Conditions for Collective
Economic Development. "Journal of Economic Growth Action9in(2):
South India. Oakland: ICS Press.
131-
65. Watson, J. W. 1886. History of Gujarat (Musalman period, A.D. 1297-
Sircar, D. 1962. "Veraval Inscription of Chaulukya-Vaghela Arjuna." 1760). Bombay: Government Central Press.
Epigraphia Indica , XXXIV, 141. Wilkinson, S. I. 2004. Votes and Violence: Electoral Com-
Smith, R. J., and R. W. Blundell. 1986. "An Exogeneity Test for a petition and Ethnic Riots in India , Cambridge studies in
Simultaneous Equation Tobit Model with an Application to Labor comparative politics. Cambridge, UK: Cambridge University
Supply." Econometrica 54 (4): 679-86. Press.
832

103.150.40.18 on Tue, 19 Mar 2024 17:59:44 +00:00

JHA TradeInstitutionsEthnic 2013

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

JHA TradeInstitutionsEthnic 2013

Uploaded by

Copyright:

Available Formats

Trade, Institutions, and Ethnic Tolerance: Evidence from South Asia

Author(s): SAUMITRA JHA

Stable URL: https://www.jstor.org/stable/43654035

This content downloaded from

doi:10.1017/S0003055413000464 © American Political Science Association 2013

Trade, Institutions, and Ethnic Tolerance: Evidence from

INTRODUCTION the mechanisms through which trade can foster lasting

This content downloaded from

Hajj , Muslims enjoyed exogenous Porbandar in 2006-07,complementary

This content downloaded from

This content downloaded from

institutions, along with case

This content downloaded from

This content downloaded from

FIGURE 1. Major Medieval Ports and Political Patron

This content downloaded from

This content downloaded from

Every major Bohra community

This content downloaded from

I M iriifft «> iii: . U «i

This content downloaded from

This content downloaded from

TABLE 2. Summary Statistics

Riots, 1850-1950 Obs Mean SD Obs Mean SD Obs Mean SD

# of Hindu-Muslim Riots 476 1.116 3.416 53 0.925 5.487 59 0.136 0.472

Harborages were typically located that

This content downloaded from

economic historians of the region

This content downloaded from

FIGURE 2. Medieval Legacies: Religious Composition and Hindu-Muslim Riots

This content downloaded from

O) ^ CO Õ W CVJ ^ CO CO CV1 ÏO O R O CO CO Õ Õ Ti- ÏO Tí- ^ ^ oT C3) CO = CO C30 CO

"£ _ CO^CO^CMO)^C01-NT-I0^i-C0ir)00c0i-W00c00)(x)0) = c/) co oo g

This content downloaded from

5J§ §l§§i.ì|i«J«P = cog x » i.

This content downloaded from

TABLE 5. Regression: Interactions with Human C

Probability of A Riot Number of Riots

OLS: Years 1850-1950 (1) (2) (3) (4) (5) (6)

Medieval Overseas Port -0.206** -0.250*** -0.237** -0.532 -0.902* -0.899**

Notes : All regressions include quadratic polynom

This content downloaded from

TABLE 6. Regression: Placebos and Robustness

Notes : Each cell represents a regression. A

human capital accumulation, we should expect Persistence and Change, 1850-1995

This content downloaded from

FIGURE 3. Timing of the first failure of religious

Note : This Kaplan-Meier curve compares the relative survival rate of to

also at risk for violence earlier

This content downloaded from

ín J cr> lo r» ifi^CNjcDw^oœifiœ^wor-r-cococû co - p >- __ Z o ^ c

iü-Q en CO TtCMCM^OOCMOT-O^OOOOOCOr^CM COLL E .9

sI s S|csi il- §S oId. U^£5«= !jy1

This content downloaded from

TABLE 8. Regression: Interactions with Mediev

Probability of A Riot Number of Riots

OLS: Years 1850-1950 (1) (2) (3) (4)

Medieval Overseas Port -0.235*** -0.053 -1.330** -1.478

Notes : All regressions include quadratic

wages in a given non-Muslim

This content downloaded from

TABLE 9. Regression: Interactions with Democratic Political Incentives, 1950-95

Any H-M Riot in Year? (OLS) (1) (2) (3) (4)

Medieval Overseas Port -0.206 -0.250* -0.237** -0.532 -0.902* -0.899**

Medieval Overseas Port -0.235* -0.053 -1.330 -1.478

Medieval Overseas Port -0.034 -0.034 -0.034 -0.035