Professional Documents
Culture Documents
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
/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.
806
807
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
809
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
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
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
813
^ (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
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) lâ^I £ 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 ^OJ
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
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
816
817
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
C O TtOCOCOOCVJOOOOOO O) 1^ = u ^ o oo S- Lfì
S^OT-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-*
> 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^l^OlOOi-OlOCO 1^1- Wo5r-<0 O «?
T-O^tOOOOOOOOO^OOOOOOOOi-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 I^OzjOSkCOc = ^ CO
(0
c ^ T-O^OOOOOOOOO 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
* *
* * * *
^ 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)^i-COO^COi-Oi-0)in^OO)^^COCOCO^OMí)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 _ * «OÍb^O>OOCMO>TMn^lO^I^®COOO^ÍřCM^IOCOCMÍ-^ „ „ _* _ 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 ^-^oTcVJOOOO^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 cbīoiO^OOĪOOO^CŪCDCOOOO * CD CO ¡O C» * CM CD OO O^OOO) 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 ^wcvi^or-T-r-ooNWi-wMnoi-comT-o)oifìw^ co o ;r 0 00
yâdooooodododo^odoT^o^oyooo Q V ¿5
*
rn bo7oí"iococoF^,N.O)cocO'«-oícoioo>r^oo)coc'jcoTtcDi^ _ z
^ rn ^ CONOOOOU)WCDCO<DO)ini-CONi-m^Oi-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-NOC0L0CD^Ol00)0)C0 O CO ^ ^
LO - CO i-CvJOJCVJinr^-r-OlOCMi-COCOOOCDi-COCVJCDOCÚT-CDCOT-CO «o 0® ^ co
0)£ - CO OOCVI^O)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^co^or-'odoin ¿dodo ¿V J ^
3 .
Tñ ¿
^ .Eg * ^ „ „ „ „J „ r-, r-^ 0
A ^Cû ■£= ífOi-^OOOOCDCOCOCOCMCD
o -o co • cc r^ioi^oo^-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 ¿co^o^i^ ¿-1-' di-' ¿evi ooo^' ¿ ^
¿
U)
O * §
00 ^CÛ ^ T-mOCON0CVIN^COi-m(DONOCO^OCOi-(DONCDO = o <5
T- ^ CM .CO OCVINT-^TtNtOin^Oï^ONCÛT-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- o^odo 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
^OCOCMi-NOîi-WWOOOCOO
O CD 05 CO IO O 00 O) CD CO IO ^ F
U- ^ C'J E F
3 Z¿ ddcod^or^o^odo 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
820
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
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
822
823
* * * *
***** *
!_
!_ «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 >.
*
* 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
824
825
826
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
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
* * *
** * * ** , . u
_ . O) ^Fo f^oTcbo' .2 b^o^ ^ -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^-^-^uoLO loouo^tcviio „ $> oocm^-^odlot-t- „ „ o | ^
íC ^ _ i-Oi-O^Oi-
^ 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
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 ^O^.. =3 0 © © ^ S ^XmÛd) 0 "P
cr X il o S 2 2 -í cr il co īž
_é_
829
830
831
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