Trade Shocks, Mass Mobilization and Decolonization:

Evidence from India’s Independence Struggle

Rikhil R. BhavnaniDepartment of Political ScienceUniversity of Wisconsin-Madison

Saumitra Jha⇤

Graduate School of BusinessStanford University

October 31, 2011

Abstract

A key challenge for political and economic development lies in generating broad coali-tions that span economic and ethnic divisions. We measure the e↵ects of a particularmechanism—shocks to trade—in mobilizing the Indian subcontinent’s remarkably di-verse population into one of the world’s first mass political movements in favour ofIndependence. Using novel data, we find evidence that residents of exports-producingdistricts that were negatively impacted by the Great Depression and Britain’s policyshift from free trade to an imperial preference regime favoring British manufactureswere more likely to support the Congress, the party of independence, in 1937 and 1946and more likely to engage in violent insurrection in the Quit India rebellion of 1942.However, districts experiencing both positive and extreme negative shocks were asso-ciated with lowered support. We interpret our results as inconsistent with a “peasantrebellion” interpretation of India’s independence and instead as reflecting the role ofthe Great Depression in aligning the incentives of South Asia’s producers of exportablegoods (broadly, providers of the labour) with import substituters (providers of thecapital) in favour of political independence, even while Imperial protectionism forgednew pro-Empire constituencies.

⇤Emails:bhavnani@wisc.edu; saumitra@gsb.stanford.edu. This draft is preliminary and incomplete. Weare particularly grateful to Dennis Appleyard for generously sharing his data and to Abhijit Banerjee, LatikaChaudhary, Helen Milner, Huggy Rao, Je↵rey Williamson and to participants at seminars at Stanford, theISNIE conference and the All-UC conference on the Great Specialization for helpful comments. Abhay Anejaprovided excellent research assistance. Thanks also to Ishwari Bhattarai for help with the data.

1

Introduction

India is a geographical term. It is no more a united nation than the Equator.

— Sir Winston Churchill, March 18, 1931

The fundamental role that the mass mobilization of disenfranchised groups can play in

large-scale institutional change, including revolutions and democratization, has long been

emphasized in many prominent theories of political development (Engels and Marx, 1848,

Boix, 2003, Acemoglu and Robinson, 2005, North, Wallis and Weingast, 2009).1 At the same

time, much blame for the lack of change and for persistent underdevelopment around the

world has been attributed to a failure to create broad coalitions in favor of beneficial reforms

in societies riven by di↵erences in ethnicity, wealth and other dimensions (e.g. Engerman

and Sokolo↵, 2000, Acemoglu, Johnson and Robinson, 2005a, Rajan, 2006, Jha, 2011). Yet,

much less is known about factors that have been successful as driving such mass mobilization

or forming such coalitions. In this paper, we assess the role of a particular mechanism—the

role of trade shocks—in reducing the shared interests between ruler and ruled, and forging

new coalitions in favour of democratization and political autonomy. We do this using novel

data on an important yet puzzling success: that of the mass mobilization of the inhabitants

of the countries that would become India, Pakistan and Bangladesh in favor of democratic

independence.2

India’s successful struggle for independence from Britain marked the first major rever-

sal of a process of global colonization and market integration by Europeans that had been

continuing since the early nineteenth century, making it both a central and a pioneering ex-

ample for future civil rights and independence movements around the world (Figure 1). Yet,

beyond its historical importance, India’s independence struggle poses a number of intrigu-

ing puzzles for social science. Surprisingly, both for contemporary observers like Winston

1See also Bhavnani (2010) and Jha and Wilkinson (2011).2In what follows we will follow contemporary usage and refer to that portion of the Indian subcontinent

under direct or indirect British rule as ‘India’, encompassing not only the nation that adopted that nameafter 1947, but also the territories that would form the future states of Pakistan and Bangladesh.

2

Churchill and students interested in collective action, India’s independence struggle emerged

as one of the world’s first mass political movements, spanning both rich and poor as well

mobilizing supporters across much of India’s remarkable ethnolinguistic diversity. Surprising

from the perspective of trade theory, the platform of the main party of Independence, the

Congress, was avowedly autarkic, and yet still proved broadly popular even though India was

(and remains) labour-abundant and capital-scarce, conditions that classic trade theory and

cross-country evidence suggests should favour political support for free trade (Stolper and

Samuelson, 1941, O’Rourke and Taylor, 2006, Lopez-Cordova and Meissner, 2008, Milner and

Mukherjee, 2009).3 This is still more remarkable because in its last two decades, India’s in-

dependence struggle had become, to an important extent, a struggle over control over India’s

trade and foreign policy. By 1937, ten years before Independence, India’s first broadly rep-

resentative provincial legislatures had already acquired substantial local autonomy, with the

British retaining control over foreign policy and overseas trade. The foundations of India’s

post-Independence redistribution systems and its developmental state were in fact largely

the product of British India’s war e↵ort (Kamtekar, 1988). Yet the intervening years were to

see continued mass mobilization, often at high risk, by both rich and poor in favour of seiz-

ing Britain’s remaining imperial rights, with the avowed aim of Purna Swaraj—“complete

independence.” A large coalition of Indians chose not to take the path of self-governing

dominion within the empire o↵ered by the British, a path trod by Australia and Canada,

with its accompanying ease of access to within-empire trade and immigration.4

How and why then did a broad coalition of South Asians form across ethnolinguistic and

economic lines to push for democratic self-determination? In this paper, we provide the

first systematic empirical evidence on the determinants of support for the Indian indepen-

dence movement. In particular, we test whether trade shocks caused mainly by the Great

3The essential intuition is that because free trade allows the flow of capital goods into the country, thisshould raise the value of labour. Thus, workers should prefer free trade, and in labour-abundant societies,workers will have the median vote.

4The importance of trade and foreign policy to India’s independence struggle may seem remarkable tomodern observers: prior to the reforms in the 1990s, India was a classic closed economy. Yet, these policiesthemselves were the post-Independence implementation of national policies by the Congress party.

3

Depression decreased the complementarity between sub-continental producers of exported

goods and metropolitan industries, instead aligning the incentives of such producers with

sub-continental industrial interests in favour of democratic self-determination, and with it

an autonomous trade and macroeconomic policy.

We perform this exercise using a range of hitherto untapped district-level data sources,

assembling novel data on mobilization in favour of democratic self-determination, including

votes and turnout in the first provincial elections in 1937, secret intelligence reports on violent

insurrection during the “Great Rebellion” of 1942 against British rule, and Congress member-

ship on the eve of Independence in 1946. These data are supplemented with Depression-era

data on crop-growing patterns, agricultural yields and employment in import and export

crops and manufactures. We find remarkably consistent evidence that the residents of more

industrialized districts and districts producing exportable goods that faced lowered British

demand were more likely to support the Congress, the party of Independence, in elections in

1937 and in its membership on the eve of Independence in 1946. These residents were also

more likely to engage in violent insurrection in the Quit India rebellion of 1942. However,

districts that experienced positive and extreme negative shocks in this period were associated

with less, rather than more, support for Independence.

We interpret these results in light of the benchmark theory of trade. Under the broadly

free trade regime that characterized South Asia under British rule in the early 1920s, the sub-

continent’s industrialists were largely uncompetitive overseas and faced world and British

competition in their domestic markets. Not surprisingly, these industrialists and mill-owners

often voiced a strong demand for import trade barriers (Rothermund, 1992). Yet, India

under British rule had long been remarkably open to world markets. Capital inflows under

free trade naturally favour labour (Stolper and Samuelson, 1941), even while the residents

of many districts were enjoying relatively cheap foreign manufactures and added demand for

exportable primary goods, including both staples and cash crops (Figure 3). Though much of

the surplus from India’s trade likely accrued to intermediaries, including landlords (eg Kran-

4

ton and Swamy, 2008), there were still strong economic benefits to producers of exportable

goods from the Raj.

Yet, metropolitan and world demand for sub-continental goods fell with Britain’s 1925

decision to return to the Gold Standard at its overvalued pre-war parity, followed by the

Great Depression, whose major impact was felt in India in 1930 (Figure 2). Britain’s subse-

quent abandonment of free trade in favor of an “imperial preference” regime favoring British

manufacturers also protected a small number of Indian exports that did not compete with

British goods. Thus, even while the imperial preference regime created new pro-Empire

constituencies among protected exporters, exogenous trade shocks reduced the benefits of

Empire for India’s unprotected exportable goods producers.

At the same time, districts di↵ered in their ability to mitigate the shock. Districts where

producers were better able to switch from exports to food crops were likely also able to reduce

the need that their producers faced for the risk-sharing and trade intermediation services

provided by landlords. 5 With the capital for India’s Independence movement available

from industrial rather than landed interests, the promise of redistribution of land from the

group frozen out of the deal may have helped forge the coalition. Thus trade shocks may

have facilitated the formation of a broad coalition of workers, from former indigo growers

in Bihar to factory workers in Gujarat providing the labour that complemented the capital

of textile manufacturers and other industrial interests necessary for India’s successful mass

mobilization.6 In contrast, districts experiencing extreme shocks were likely those most

unable to switch away from exportables, and thus those with a continued interest in an

imperial link and a (relatively) open trade policy.

Our paper provides not only evidence for a novel interpretation for the movement that

led to the democratic self-determination of one-fifth of the world’s population but also con-

5These landlords include the explicit zamindars in the areas that fell under the Permanent Settlementbut also included landlords in ryotwari areas–the distinctions are quite blurred in this period, since landrights had often been sold on. See also Banerjee and Iyer (2005).

6NB this portion of the interpretation is our current working hypothesis but this version of the papercontains only suggestive evidence for this piece of the puzzle.

5

tributes to the social science literatures on the role of coalition formation in institutional

change, on democratization and trade as well as on decolonization and nationalism. We

also speculate on how our work might explain the roots of economic policy making in the

post-Independence period.

As discussed above, the role of shocks that encourage the mass mobilization of disenfran-

chised groups plays a fundamental role in many of the most prominent theories of institu-

tional change (Lipset, 1960, Moore, 1966, Boix, 2003, Acemoglu and Robinson, 2005). While

the particular importance of trade shocks has been emphasized in encouraging the relative

empowerment of trading groups in engendering change (Acemoglu, Johnson and Robinson,

2005b, Jha, 2008), less work has focused on the role of trade shocks in aligning the interests

of sub-groups possessing the capital and the labour necessary for successful mobilization in

favour of democratic self-determination.7

We also build upon and contribute to an important literature in the political economy

of trade that finds, consistent with the Stolper-Samuelson intuition, that labour-intensive

democracies tend to have lower trade barriers, and in turn that variation in world trade

volumes (Ahlquist and Wibbels, 2010), or natural openness to trade (Eichengreen and

Leblang, 2008, Lopez-Cordova and Meissner, 2008) explain democratization.8 We break

new ground and look at within- country, rather than cross- country variation, which enables

us to reconcile these works with the puzzling coincidence between the movement of South

Asian and many other post-Independence countries towards both increased democratic self-

determination and higher trade barriers. We argue that part of the answer may be found in

the interaction between negative trade shocks– that reduce the economic benefits from trade

intermediaries and risk-sharing through concentrated land ownership– and democratisation,

that makes redistribution of these newly available rents credible.

Our paper also contributes to works on decolonization and nationalism, being the first

7Indeed, there are reasons to expect that, in the absence of such trade shocks and the possibility of futureredistribution, the complementarity between capital and labour in mobilization may have made ethnic-basedmobilization more likely (Esteban and Ray, 2008).

8Milner and Mukherjee (2009) provides a very useful overview.

6

paper, to our knowledge, to use within-country empirical variation to examine these phe-

nomena. While the questions we can address using sub-national data are naturally narrower

than possible with the immense institutional variation of a cross-country study, our focus on

South Asia, which includes data on districts that would form the three future countries of

India, Pakistan and Bangladesh, allows us to shed light on an environment that housed one-

fifth of the world’s population that is less subject to reverse causality and confounds, while

still encapsulating remarkable political and socioeconomic diversity. We briefly summarize

our contributions to these literatures below.

India has long been a puzzle for the literature on nationalism. While prominent strands

of this literature has emphasized the rise of print journalism in creating a common “high

culture” (Anderson, 1983) and uneven modernization (Gellner, 1983), a common theme is

that nationalism, when it occurs, tends to be concentrated within ethnically homogeneous

units with high or growing literacy rates. Such explanations fail to explain India’s remark-

able mass mobilization across ethnic and social lines, particularly in an environment of low

literacy. We resolve this puzzle by examining the political economy of India’s trade. Our

account has commonalities with Gourevitch’s 1979 argument that nationalism arises when

there when economic and political centers are distinct. However by tracing the emergence

of nationalism to district-specific shocks due to the Depression, however, our account under-

lines how nationalism can even vary across space when there is a general disjuncture between

political and economic centers.

Further, by looking within a single independence struggle, we are also able to hold

constant the competing factors emphasized by much of the existing literature on decolo-

nization and the growth of self-determination movements. These important works have

emphasized the metropole’s interests (Lustick, 1993), the inevitable growth of national-

ism (Brubaker, 1996), the obstruction of demands for representation (Lawrence, 2007), state

weakness (Lawrence, 2007), changes in international norms (Hailey, 1943), or the destruction

wrought by World War II (Clayton, 1994). Instead, we can isolate the substantial extent

7

to which India’s independence was driven by economic incentives. A broad coalition of In-

dia’s remarkably diverse population, rather than behaving as emotional nationalists, had

economic reasons to be rid of the Raj. India is a particularly good case with which to study

the drivers of decolonization since, being the first major decolonization since the aftermath

of the Napoleonic Wars in Latin America, its decolonization could not have been subject to

spillover e↵ects from elsewhere (Figure 1).9 Instead, our analysis has intriguing parallels with

recent theoretical work by Bonfatti (2010) who emphasizes the disincentive to independence

due to the potential loss of a metropole’s trade with the colony.

Our paper also connects with methodological literatures. In particular, by examining the

e↵ects of the Great Depression on three very di↵erent types of political mobilization—party

membership, voting and protests—our paper is one of the few to test the idea that di↵erent

types of political mobilization can have the same root causes (McAdam, Tarrow and Tilly,

2001).

Finally, by assembling novel data, which includes, to the best of our knowledge, the first

comprehensive assembly of archival intelligence data on the extent of non-violent and violent

insurrection in the war-time Quit India rebellion, we contribute to Indian history. The two

major strands of existing Indian historiography emphasize either the metropole’s reasons

for granting India independence (see, e.g., the Transfer of Power series published by the

U.K. government—Mansergh (1976)), or provide thick description of the micro-politics of

the movement in India (see the Towards Freedom series published by the Indian Council

for Historical Research—Gupta, ed (2010), Prasad, ed (2008), Panikkar, ed (2009), Gupta

and Dev, eds (2010)). These literatures, respectively, mention the Great Depression as a

factor weakening Britain’s will to rule India, and as a cause of a “peasant movement” in

the inter-war years, which provided the elite-led independence movement with the masses

it needed (Rothermund, 1992, 2006). We are able to test the latter claim empirically, and

find it incomplete as an explanation. Instead, we are able to propose and begin to test a

9India’s independence, on the hand, is often said to have inspired other anti-colonial movements (Rother-mund, 2006).

8

novel interpretation, based upon on the political economy of India’s trade, to explain not

only one of the pivotal historical episodes in the political and economic destinies of one-fifth

of the world’s population, but also why and how there was a mass mobilization in favour of

democratic self-determination that has since served as a central example to freedom struggles

around the world.

We start by outlining our alternative account of the Indian Independence movement.

The next section details the unique data and empirical strategy that we rely on. We then

present our results, and conclude.

An account of the Indian independence movement

The leading organization of the Indian independence movement—the Indian National Congress—

was founded in 1885, soon after the British abolished most import duties in India. For

much of its pre-Independence history, the movement was dominated and financed by rich

professionals—particularly lawyers and businessmen—who made their living from India’s

triangular trade with Britain and China. These elites pushed for greater self-government

within the British Empire. In a separate paper, we intend to trace the e↵ect of trade on the

birth and initial growth of this movement. In this paper, we estimate the impact of trade

shocks on the dramatic transformation of the movement in securing India her independence.

We start our account in the early-1920s, when the Independence movement still was—

despite its recent expansion under Gandhi, who had returned a hero from South Africa in

1915—largely a narrow, elite-led one, occasionally derided as a “talking shop.” By 1935, the

movement had transformed itself into a mass movement aiming for complete independence.

The transformation was so substantial that independence appeared eminently achievable by

the end of the decade, with World War II possibly acting in a delaying rather than expediting

role. We concern ourselves with describing this transformation.

The reasons for the broad-basing and change in aim of the Indian independence movement

9

were many, but most historical accounts highlight two factors—the impact of the Indian Na-

tional Congress’s strategic campaigns, particularly under Gandhi’s leadership, and the great

economic tumult of the inter-war period, which reached its nadir in the Great Depression.10

We focus on the latter, partly because Gandhi’s e↵orts were explicitly conditioned on eco-

nomic factors, and were therefore endogenous to the economic situation.

The 1920s were tumultuous for the world economy, and its boom and busts severely tested

the world. Until that time, Britain’s stewardship of Indian trade policy had brought with it

an openness to trade that India would not see again at least until the 1990s. Yet, 1923 is

considered the last “business as usual” year under the broadly free trade regime that India

had become accustomed to as a colony of the United Kingdom (Appleyard, 1968, 2006).

A series of questionable policies followed, beginning with the United Kingdom’s return to

the gold standard at pre-war (and now, overvalued) levels in 1925. This was followed by a

remarkable contraction of world trade during the Great Depression, which started in 1929.

Both a↵ected practically every sector of the Indian economy, the country’s relationship with

Britain and the rest of the world, and, as we will show, the dynamics of the independence

movement as well. An indication of the economic tumult of the time comes from the the

total value of imports into the United Kingdom from British India: these nearly halved from

£67 million in 1923 to £37 million in 1931 (see also Figure 2).

The negative e↵ect of the Great Depression was exacerbated by the Raj’s external-sector

responses, which reflected Britain’s economic and security imperatives more than India’s

needs.11 The first of these responses had to do with exchange rate. Britain abandoned the

gold standard in September 1931, e↵ectively devaluing the pound, while at the same insisting

that the rupee remain pegged to sterling at its existing high value.12 This allowed Britain

to reflate its economy—a policy that practically all the world followed—at the expense of

10Metcalf and Metcalf (2002), which barely discusses the economic dimension, is perhaps an exception.11Rothermund (1992) provides a compelling account of the over-ruling of the Finance Member of the

colonial government based in India, George Schuster, in seeking a devaluation by the Secretary of State forIndia in London.

12This stands in contrast to the devaluations that the dominions of Australia and New Zealand were ableto pursue.

10

India’s economy. British exports to India were favored over India’s exports to the world,

and a massive outflow of gold from the country and to Britain followed. Existing deflation-

ary pressures due to the collapse in demand due to the Great Depression were, in e↵ect,

exacerbated by the Empire’s exchange rate policy.

The second external-sector response to the Great Depression was an abandoning of free

trade. The 1931 “Ottawa Agreement” established “imperial preferences” between Britain

and her colonies. The Empire would operate as a preferential-trade zone, with the high tari↵s

to non-members, and preferential ones for members. The agreement o↵ered the British the

cover with which to extract low Indian import duties for 160 of its manufactures, while

agreeing to similar terms for a smaller number of Indian raw material exports (Rothermund,

1992)(p.147). While the former created opposition to Empire, the latter created—as we

detail below—new supporters of Empire.

British policy led to the segmentation of India’s populace into at least three distinct

groups, each of which reacted to the regime in di↵erent ways and for di↵erent reasons.

We consider each of these in turn, detailing how their interests were a↵ected by the Great

Depression, the overvaluation of the rupee, and the Ottawa agreement. The first group were

India’s “protected exporters,” who received preferential access to British markets under the

terms of the Ottawa agreement. This group mainly exported those Indian commodities that

the British turned to when in Depression: drugs, tea, co↵ee and tobacco. These were grown,

perhaps not coincidentally, chiefly on British-owned plantations.

The second group were India’s “unprotected exporters,” which included the bulk of the

population. This group included the producers of staples, such as wheat and rice, and of

export cash crops such as cotton, indigo and jute. This constituency su↵ered greatly under

the Great Depression, due to the fall in the demand for their products, which was exacerbated

by Britain’s decision to keep the rupee overvalued. However with a dramatic fall in trade

and a change in their allocation of factors to subsistence crops, this group may have also lost

its need for trade intermediaries and providers of risk-sharing services, including landlords.

11

The fall in trade may have thus created the potential for a promise of land reform, made

credible by the democratic franchise. This may have made possible a coalition between

erstwhile producers of export goods and the third group a↵ected by Great Depression— the

owners of India’s infant industries. These “import substituters” had strong incentives to

wrest Britain’s control of India’s external policy, both because of the overvalued exchange

rate that resulted from it, and because of the Ottawa agreement, which instituted preferential

tari↵s on manufactures from Britain. Both policies disadvantaged domestic manufacturers

in their domestic market. The only way to wrest control of such policies, was in fact, to

sue for complete independence. Indeed, it was as the Great Depression struck, on January

26th 1930—thenceforth celebrated as Independence Day—that Congress abruptly changed

its platform from self-government within the British empire to Purna Swaraj.13

A number of papers have pointed out that economic dislocation is oftentimes associated

with political participation, partially for expressive reasons, but also for instrumental rea-

sons, as people wish to do something to better their situation. Indeed negative economic

shocks have been seen as an instigator of peasant rebellion in India (Rothermund, 1992) and

increased social conflict more generally (e.g. Dal Bo and Dal Bo, 2004, Miguel, Satyanath

and Sergenti, 2004). Yet, we will provide evidence that the historical literature mistakenly

conflates the link between negative shocks, mobilization and support for democratic self-

determination. Districts that were worst hit by the Great Depression, while being more

politically active, were actually less likely to support the Independence movement. This is

entirely consistent with the intuition of classical trade theory: the autarkic platform of the

Independence movement did not make it the natural choice for labour.14

13Celebrations of India’s “Independence Day” would continue until 1947. Lord Mountbatten chose insteadAugust 15th as this was the anniversary of his greatest triumph—the surrender of Japan. Later January26th was rehabilitated as India’s Republic Day.

14Yet, other, possibly complementary, mechanisms that we are still in the process of testing may also beat play. For example, an increased need for relief from the incumbent government and landed intermediariesmay have led the worst hit to support local landlord parties rather than the Congress. Poverty may havealso enhanced risk aversion, thus favouring established interests. What we can distinguish is whether thepoor fail to coordinate due to a pure coordination dilemma (Kuran, 1991): while such an e↵ect might a↵ectviolent action, it would be less likely to influence voting under secret ballot, unless there was a possibility ofcollective punishments in the form of withholding of incumbent government relief.

12

The argument that much of the political mobilization of the 1930s was not for the

Congress is unusual in Indian historiography. This is partly because there has been lit-

tle systematic quantitative analysis of pre-independence era mobilizations. Evidence for our

claims can, however, be seen in some aspects of the historical record. Explaining how the

alliance crafted in the fire of the depression came to be born, Bose and Jalal (1998) argues

the Congress was practically “pushed, by the pressures which the colonial state’s economic

policies were generating from below, into taking positions they might otherwise have wanted

to resist” (140).15 The Congress could either ride the wave of economic disa↵ection that

confronted it, or be subsumed by it. Although the Congress chose to ride the wave of

disa↵ection, and this changed its subsequent demands, which now included both sops for

agriculturalists and industry,16 the alliance between the elite (mainly import-substituters)

and non-elites (mainly unprotected exporters) remained fragile.17

This was, as pointed out previously, because there was a substantial disjuncture between

the interests of import substituters and unprotected exporters. While the first of these

favored protection from imports, the latter will have preferred, per Stolper-Samuelson, a

free trade regime so as to benefit from capital inflows (O’Rourke and Taylor, 2006, Stolper

and Samuelson, 1941). This disjuncture might also help explain a recurring puzzle of India’s

pre-independence politics, where Gandhi—sometimes with, and at other times without the

Congress’s backing—would call o↵ their agitations against the wishes of the movement’s

15Rothermund (2006) discusses the forging of another coalition, between socialists and industrialists, thatalso helped fashion the independence movement and the country’s post-1947 economic policies. He notesthat “debates on British currency policy added to an increasing awareness among Indian industrialists thatnationalism was their best bet. Import substitution behind tari↵ walls guaranteed by a national governmentwas the ideal which they pursued. In this way socialists like Jawaharlal Nehru and Indian capitalists wereable to find a common denominator. Both preferred a national interventionist state to a pseudo-liberalcolonial state” (259).

16Bose and Jalal (1998) note that “Five of Gandhi’s eleven demands . . . related to economic issues. His callfor the abolition of the salt tax and a reduction of the land-revenue demand by half were designed for India’peasant masses. On behalf of India industrial bourgeoisie Gandhi demanded protection for the indigenoustextile industry, reservations of coastal shipping for Indians . . . , and a reduction of the rupee-pound exchangerate .. to stimulate Indian exports” (149).

17The Congress’s need for large amounts of funds to sustain the mass movement extended even to main-taining Mahatma Gandhi’s asceticism. Congress President Sarojini Naidu famously asked Gandhi “if youknew, Bapuji, how much it costs to keep you in poverty.”

13

rank and file. As Bose and Jalal (1998) note, the Congress was so uneasy with this alliance,

that the “the Gandhian Congress [was] ready to press the brakes, fearful of people running

ahead of the leadership and redefining the organization’s cherished goal of Swaraj” (140).

Though democratic self-determination might have made more credible the promise from

the industrial “capitalists” of the mass mobilization to the erstwhile agrarian exporters who

provided the “labour” to redistribute resources from the now-economically irrelevant landlord

intermediaries of India’s world trading past, the coalition remained an uneasy one.

Data and empirical strategy

We seek to measure the e↵ect of trade shocks due to the Great Depression and the institu-

tion of British protectionist “imperial preferences” on support or opposition to the Indian

National Congress, the main party of the Indian independence movement. The ideal compar-

ison would be to compare two districts with same levels of initial exposure to foreign trade

during the free trade regime of the 1920s, one of which received protection under “imperial

preferences” during the Great Depression, and one that did not. A third comparison cate-

gory are those districts which did not produce goods for export under free trade, and whose

producers were relatively insulated from the costs and benefits of imperial preferences.

Our benchmark specification will be cross-sectional regressions of the following form:

M1936,d = �1V d1920�23 + �2S

d1923�1933 +X 0⇣ + ✏d (1)

where M are measures of mobilization, V d is the average value of export goods per worker

in a district between 1920 and 1923, S is the percentage shock to the value of export goods

per person in a district due to the Great Depression and the imperial preference regime,

X are controls including provincial fixed e↵ects, ✏d are unobserved factors that may drive

mobilization that we assume to be independent between provinces but allow to be arbitrarily

correlated (clustered) within them, and d indexes administrative districts, which is the level

14

for our analysis.

We employ four new measures of colonial era mobilization in our analysis. One of these—

turnout during the 1937 elections—is a measure of overall mobilization. The other three—

Congress party support in the 1937 provincial elections, violent and non-violent political

activities during the Quit India “rebellion” of 1942, and Congress party membership in

1946—are measures of support for independence. The Congress Party membership data

were taken from the organization’s membership handbook; 1937 election data were taken

from the o�cial election returns, and the Quit India data were drawn from a series of secret

intelligence reports written by the British (please see the Data Appendix).

The initial value of export goods per worker in a district is calculated as follows:

V d1920�23 =

X

g

Vg,1920�23 ⇥ wdg

W Tg

(2)

where Vg,1920�23 provides the average c.i.f. value of British India exports to the UK in 1920-

23, g indexes all goods exported to the United Kingdom from British India appearing in

the Annual Statements of Foreign Trade of the United Kingdom for the relevant year, and d

indexes districts. wdg are those that work in the production of the good g in district d in 1931,

while W Tg is the total number of workers producing that good over all districts. Thus the

number of workers producing a good acts as a district-specific weight to changes in demand

for that good: those areas where relatively more workers are employed will be more a↵ected

by changes in value.

Note that as we are looking at the 1931 figures on employment, we are capturing those

individuals who chose not to or were unable to adjust to the 1923-33 trade shock by switching

out of export-oriented professions or crops. In a “peasant rebellion” interpretation, the

ability to adjust should mitigate the estimated e↵ect of the shock by lowering the demand

for mobilization among those groups who were able to adjust. Similarly, a demonstrated

unwillingness or inability to adjust should strengthen the e↵ect of an extreme negative shock.

15

In contrast, if it is the case, as we argue, that it was those erstwhile exporters who could

adjust to domestic production that had their interests most aligned with industrial interests

and the promise of future redistribution, we should expect intermediate negative shocks to

have the most impact.18

We then calculate the percentage shock to the value of export goods per person in a

district due to the Great Depression and the imperial preference regime:

Sd1923�1933 =

V d1930�34 � V d

1920�23

V d1920�23

⇥ 100 (3)

We use as our measure the change in the value of exports rather than just the world or

UK prices as this enables us to capture the changing export mix of goods in response to

world demand and the tari↵ regime, as well as giving us a measure that is intuitive: it is

the change in the average revenue product per worker in each district.19 V d can be broken

down into its component sectors (manufacturing, cash crops, staple crops, natural resources

etc) by doing the analogous calculation over the goods and producers in those sectors. The

Appendix provides details of which goods are assigned to which sector.

Our identification strategy rests on the assumption that the value (i.e. equilibrium price

and aggregate quantities) of UK imports from India are driven mainly by the fluctuations

in the pound, changes in world demand, and the broad tari↵ regime set in the Ottawa

agreement in 1931 favoring British manufactures, rather than by political mobilization by

individuals or groups within specific Indian districts.

The identification of the e↵ects of the great depression is particularly plausible given that

we do not use district-specific price measures to construct our shock measures. We instead

use the c.i.f. value of imported goods from India into Britain for various goods multiplied by

district-specific production of those goods in 1931 to construct our shock measure. Thus we

18The next iteration of this paper will examine the factor responses directly, by comparing the productionmix in 1923, prior to the Depression to the production mix thereafter.

19We also use price shocks as instruments for value shocks: though not precisely estimated in a number ofspecifications, we get results consistent in sign and magnitude. A key issue with these price shocks is thatthey do not account for changes in the basket of export goods.

16

are capturing those individuals who by 1931, had either chose not to or were unable to adjust

to the 1923-33 trade shock by switching out of export-oriented professions or crops. In a

“peasant rebellion” interpretation, the ability to adjust should mitigate the estimated e↵ect

of the shock by lowering the demand for mobilization among those groups who were able to

adjust. Similarly, a demonstrated unwillingness or inability to adjust should strengthen the

e↵ect of an extreme negative shock. In contrast, if it is the case, as we argue, that it was

those erstwhile exporters who could adjust to domestic production that had their interests

most aligned with industrial interests and the promise of future redistribution, we should

expect intermediate negative shocks to have the most impact.

The fact that we use three independent measures of mobilization to support our argument

should increase confidence in our results. Our regressions also employ provincial fixed e↵ects,

and therefore only leverage intra-provincial district variation in mobilization. We employ a

number of additional district-specific controls for our analysis. These vary depending on the

specific dependent variable considered, and are mentioned below, as we present the results

of our analysis.

Our key dependent, independent and control variables are summarized in Table 1. While

the average district in British India produced export goods worth around Rs. 1.1 per worker

in 1923, by 1933, the average Indian district su↵ered a 47.4% drop in the value of export goods

produced there, reflecting the general collapse of prices during the depression. Importantly

for our discussion, this mean value masks great variation: approximately 1/3 of the India’s

districts experienced net positive shocks during the depression, as the combination of imperial

preferences and the world demand rose for commodities such as cinchona and myrobalans

(for drugs), iron and steel, tin ore, oilseeds and oilnuts, spices and tobacco (Figures 3 and

5.)

17

Evidence

Figure 4 presents the raw relationship between export shocks until 1933 and the degree of

turnout in the 1937 elections. Separate local polynomial smooths are applied both above

and below a zero shock, i.e. for the winner and the loser districts from the Great Depression

and the imperial preference regime. Notice that the figure appears, at first, to confirm the

perspective of historians that the Great Depression led to mobilization by a ‘peasantry’

pushed to protest and rebel by the extreme negative shocks of the Depression and imperial

policy. The residents of districts that su↵ered greater negative shocks to the value of their

export goods appears to be somewhat more likely to turnout in the elections.

However, Figure 6 suggests that this account is incomplete. The Figure presents the

relationship between export shocks until 1933 and the vote share of the Congress party in

the 1937 elections. Notice that the shock data are bimodally-distributed above and below

zero. Further, there is a concave relationship between the export shock and the Congress

Party vote share, with support for Congress attaining a maximum (of around a 60% vote

share) with a negative shock to the value of export goods in the district of around 30%. In

contrast, districts that su↵ered greater negative shocks were actually less likely to support

the Congress. There is also a sharp drop o↵ in support for Congress among the “winners”

from the imperial preference regime, as the positive shock rises.

These patterns suggest that those worst hit by the Depression, particularly those who had

failed to change their factors away from exportables, was not coordinated into support for the

opposition. This is consistent with the lack of attraction that Congress’ autarkic platform

might yield to those who could not substitute easily away from export goods. Instead of being

a rebellion of those facing the hardest times, support for Congress came from intermediate

districts that were relatively insulated from the Depression shock or able to adjust relatively

easily to domestic production. Further, the introduction of imperial preferences appears to

have led to a new constituency of beneficiaries from imperial preferences who subsequently

also voted against the Congress.

18

Before we show that these patterns are robust to multivariate analysis, it is worth consid-

ering why we use Congress support as our measure of support for independence. We make

two points here. First, there is arguably some basis for the stance taken by the Congress

that since the nationalist movement needed to put up a united front against the British,

votes for non-Congress parties were essentially votes against independence. Second, other

than the Muslim League—which has limited electoral support for much of the period that we

are considering—most other parties were local parties that did not take a view on national

issues.20 This was the case since, all the way until 1947, Indian legislation only allowed for

electoral competition at the local level, which created parties focused on local issues. The

Congress’ focus on the national question was unusual in this regard, and stemmed from the

fact that it was a national movement that was beginning to compete in elections.

Given this discussion, we retain Congress party support as our dependent variable, and

proceed with the multivariate analysis. Consider first the analysis of the e↵ects of the

depression on the 1937 elections. Table 2 presents an analysis of the determinants of voter

turnout during these elections, and Table 3 presents the results of voter support for the

Congress party. The dependent variables are presented as a % of the total eligible votes,

and total votes polled, respectively. All regressions control for provincial fixed e↵ects, and

employ standard errors clustered at the provincial level.

Table 2 examines the determinants of percentage of eligible voters turning out to vote

during the 1937 elections. Notice first that, consistent with Figure 4 there is a weak, non-

robust negative relationship between the export shock and turnout (1-5), which once again

may appear at first to confirm the “Peasant Rebellion” view of the Great Depression and

the mass movement for Independence. However beneficiaries from the export shock are also

somewhat more likely to turnout (columns 4-5, 6-7). Other factors that appear to influence

turnout are the land tenure system, with voters in districts with more owner-cultivators and

20Many parties consisted of landlords and local elites, mobilized around local issues. Exceptions includevarious Communist groups, who had Soviet backing, and the Unionist Party of Punjab, who favoured con-tinued ties to Britain.

19

landless laborers much less likely to turn out to go to the polls (columns 3, 5, 6).

Table 3 suggests, however, that this weakly increased mobilization in adversely a↵ected

districts did not actually manifest itself in greater votes for the party of rebellion and inde-

pendence, the Congress.21 Notice first that, consistent with the raw data in Figure 6, there

is a robust inverted-U relationship between the export shock and the Congress vote share,

implying that support for Congress was maximised in districts which lost around 40% of

the value of their goods during the Depression (Cols 1-8). The partial residual plot for the

regression in column 7, displayed in Figure 7, is consistent with this analysis. This result is

robust to removing outlier exporter districts (column 2), controlling for the extent of employ-

ment in manufacturing, di↵erent types of land tenure, army recruitment and police presence

(columns 3, 4, 6) and for the extent of initial exports by sector (columns 4, 6, 8). The result

is also robust to controlling for the extent of turnout in the elections, which actually has a

negative e↵ect on the vote share of Congress (columns 5-10).

Thus the accounts of historians that conflate mobilization with support for independence

may be missing an important piece of the puzzle. Those districts adversely a↵ected by

the Depression did appear to mobilize more, however this mobilization did not appear to

favor Congress. Columns 7-10 explore the e↵ect on Congress vote share of a positive trade

shock, parametrising this first as an interaction (columns 7, 8) and next by decomposing

the export shock in gains and losses (columns 9, 10).22 Notice that, again consistent with

Figure 6, those districts that experienced the most gains from the Great Depression and the

system of imperial preferences, and thus the inter-dependence with the United Kingdom,

were significantly less likely to vote for the party of decolonization and independence.

21The o�cial election report for the 1937 election, tabled in Britain’s House of Commons, only notesthe votes received by winner and runner up candidates and their partisan a�liation. The Congress votereceived variable is calculated from this, and is therefore properly defined as the % of the votes received bythe Congress party in districts where there was at least one constituency where the party was the winner orrunner-up. This is an underestimate of the true Congress vote share, since it excludes the votes received byCongress candidates if they were not in the top two candidates. We drop the 18 districts where no Congresscandidate was the winner or runner-up.

22The gain (loss) is calculated as: 0 if the shock is negative (positive) and the value of the shock otherwise.Thus: shock = gains - loss

20

While various measures of land tenure do not appear to be major determinants of support

for Congress in the 1937 elections, perhaps because of the limited franchise, the proportion

of males employed in industry does appear to have had a robust positive e↵ect. This is

consistent with the Congress platform that would have favored protection for industry against

the UK manufactures that received preferential treatment under the imperial preference

system.23

Table 4 examines the extent to which the change in interests due to the Great Depression

and the institution of imperial preferences persisted until the eve of Independence, using data

on primary party membership by district published by the All-India Congress Committee

in 1946. Notice that there are similar patterns to the 1937 elections—the most adversely

a↵ected districts from the Great Depression, and those that gained from imperial preferences,

were both less likely to field paid-up party members (columns 6-10). By 1946, Congress

membership was greatest in districts that su↵ered around 20-30% losses to the value of their

exports. Congress membership was more prevalent in areas that had land tenure systems

that favored rentiers (non-cultivating landlords or tenants) and more landless labourers (see

also Figure 8).

A third measure of support for Congress can be found during the Quit India movement,

also known as the ‘Great Rebellion’ or the August Kranti, a violent uprising that took place

during 1942. Our Quit India dependent variable is a (log transformed) count of the number of

events—violent/non-violent, Gandhian/non-Gandhian etc.—listed in the British administra-

tion’s “Secret Reports” as having occurred in each district during the Quit India struggle.24

Quit India protests spread throughout the sub-continent, with Bengal, Bihar, Bombay, the

Central Provinces, Delhi, Madras, Sind and the North-West Frontier particularly a↵ected

(Figure 10).

Quit India activity is a particularly condign measure for our analysis for two reasons.

23Further, the interaction between measure of males in industry and the export shock is also negative,suggesting that industrialized districts that were adversely a↵ected by the shock were more likely to supportCongress (results not shown).

24Using negative binomial or Poisson specification yield very consistent results (not shown).

21

First, the national Congress party leadership was detained the very day the action started.

And second, the country’s district administration practically collapsed in some parts of the

country. Both factors meant that the political activity that did occur at the time was mostly

spontaneous, unmediated by national leadership or British e↵orts to restore order. Indeed, as

Figure 9 suggests, with the arrest of the Congress leadership, non-violent civil disobedience

quickly gave way to violent rebellion. Though areas that had civil disobedience were also

likely to have violent rebellions, violence was particularly concentrated in export-intensive

districts with zamindari (landlord) tenure systems (Figure 11 and Figure 3(a)).

As Table 5(1-5) reveals, Quit India protests once again show the inverted- U relationship

with our export shock, with falls both for gains and for large losses. Quit India protests

were also more likely both in districts more exposed to industry and in landlord districts.

Columns 6-16 decompose the Quit India relationship into violent and non-violent protests.

As Columns (6-10) reveal, export shocks show consistent, but not precisely estimated e↵ects

on the incidence of non-violent civil disobedience. Non-violent protests was also more likely in

industrial districts. In contrast, as Columns (11-15) suggest, a much stronger relationship is

visible between export shocks and violent protest. In contrast to the acts of civil disobedience

and consistent with Figure 11 , violence was also more likely to occur in landlord districts.25

Table 6 examines particular types of violence in the Quit India rebellion– targeted at

public infrastructure (the railways), at property, and at records (particularly of taxes and

debts). Once again there is a broadly consistent picture: violence was less likely in districts

that had a positive export shock in this period, and more likely in areas that su↵ered a

negative shock. Once again, industrial districts were more likely to see additional incidents.

Property and records were more likely to be destroyed in landlord districts.26

25This outbreak of violent conflict also seems consistent with the negative legacy of landlord areas notedby Banerjee and Iyer (2005).

26We, of course, have an ecological inference problem. While we have shown that districts that are subjectto greater shocks experienced less mobilization, we have not shown that the people who stayed home duringthe independence era mobilizations were truly the ones most a↵ected either positively or negatively by theDepression. There is an important need for additional qualitative evidence to clarify this mechanism.

22

Discussion and conclusions

We have argued that the Indian independence movement’s aims and dynamics were shaped by

the unfolding of a large economic shock—that of the Great Depression—as it coursed through

India. In the first large-n analysis of the subnational variation in overall and pro-Congress

mobilization, we have shown that the regions of India that experienced the largest negative

shocks were—in keeping with the received wisdom—somewhat more likely to turnout in

India’s first large-scale (provincial) elections of 1937. This result appears fragile when subject

to statistical tests, however. Furthermore, and completely contrary to the received wisdom,

being subject to the Great Depression is associated with decreased support for the Congress

in terms of votes received, membership and activity during the large-scale “Quit India” or

Congress mobilization of 1942. By conflating mobilization with support for the Congress

party, the historical literature has missed this story.

The one-third of the country that experienced positive price movements due to a com-

bination of the Great Depression and the colonial government’s imperial preference tari↵

system behaved in a decidedly di↵erent way. Those who benefited from the new Impe-

rial preference regime both turned out and supported the Congress less. Together, this

meant that Congress support was the greatest amongst those who were moderately—but

still substantially—negatively a↵ected by the depression. Gainers and extreme losers were

loath to support the Congress, even as extreme losers turned to vote in somewhat greater

numbers.

Although the Congress did well in the elections of 1937—the party formed majority

governments in five of the eleven provinces, and coalition governments in two additional

provinces—and went on to win India her independence, our results nuance current histori-

ography. Those areas that were most a↵ected by the depression and imperial preferences

supported the Congress the least. Rather, it was districts that experienced some losses, but

were more insulated or better positioned to adjust to domestic production in response to

changes in British demand and the imperial trading regime, where the alignment with the

23

interests of industrialists most coincided, and ultimately Congress support proved to be the

greatest.

Political and economic development often requires the forging of new broad-based con-

stituencies. Although this is the case, political economists all too rarely study the formation

of such coalitions. We have studied the formation of the coalition that led to the freedom

of 20 % of the world’s population, and to the founding of the modern states of South Asia.

Yet, the freedom to self-determination ironically meant restrictions on the free flow of goods.

The autarkic policies of the Congress would keep India’s trade closed to the world for more

than forty years (Milner and Mukherjee, 2011).

More generally, our study suggests a additional but neglected legacy of colonization. In

other states too, the forging of national parties of Independence that span broad groups

may displace traditional left-right or ethnic party competition in favour of strong, single-

party rule, with profound and lasting e↵ects on the future direction of policymaking and

reform.27 As in India, trade policy in particular may be a↵ected, as such parties may use

high tari↵ barriers to generate lobbying contributions that help buy and maintain single

party dominance (Milner and Mukherjee, 2011). South Asia’s struggle for independence has

long been an example for freedom struggles around the world. Yet there may yet be more

that it can teach us.

27A number of scholars have shown that founding elections set the patterns for subsequent political com-petition (Lipset and Rokkan, 1967, Wittenberg, 2006).

24

References

Acemoglu, Daron and James A. Robinson, Economic origins of dictatorship anddemocracy, Cambridge,UK: Cambridge University Press, 2005.

, Simon Johnson, and James Robinson, “Institutions as the fundamental cause oflong-run growth,” in Phillippe Aghion and Steven Durlauf, eds., Handbook of economicgrowth, North-Holland, 2005, chapter 6, pp. 385–472.

, , and , “The rise of Europe: Atlantic trade, institutional change and economicgrowth,” American Economic Review, June 2005.

Ahlquist, John S. and Erik Wibbels, “Riding the wave: trade, factor prices and politicalregimes,” mimeo, Duke, 2010.

Anderson, Benedict, Imagined Communities: Reflections on the Origin and Spread ofNationalism, London: Verso, 1983.

Appleyard, Dennis, “The Terms of Trade between the United Kingdom and British India,1858-1947,” Economic Development and Cultural Change, April 2006, 54 (3), 635–654.

Appleyard, Dennis R., “Terms of Trade and Economic Development: A Case Study ofIndia,” American Economic Review, May 1968, 58, 188–99.

Banerjee, Abhijit and Lakshmi Iyer, “History, institutions and economic performance:the legacy of colonial land tenure systems in India,” American Economic Review, Septem-ber 2005, 95 (4), 1190–1213.

Bhavnani, Rikhil R., “Political Inequality: E↵ects and Remedies.” PhD dissertation 2010.

Bo, Ernesto Dal and Pedro Dal Bo, “Workers, warriors and criminals: social conflict ingeneral equilibrium,” 2004. mimeo, UC-Berkeley.

Boix, Carles, Democracy and Redistribution, Cambridge: Cambridge University Press,2003.

Bonfatti, Roberto, “Decolonization: the role of changing world factor endowments,” 2010.mimeo, LSE: STICERD.

Bose, Sugata and Ayesha Jalal, Modern South Asia: history, culture and political econ-omy, 2nd ed., London: Routledge, 1998.

Brubaker, Rogers, Nationalism Reframed: Nationhood and the national question in theNew Europe, Cambridge University Press, 1996.

Clayton, Anthony, The Wars of French Decolonization, London: Longman, 1994.

Eichengreen, Barry and David Leblang, “Democracy and globalization,” Economicsand Politics, 2008, (3), 289–334.

Engels, Friedrich and Karl Marx, The Communist Manifesto 1848.

Engerman, Stanley L. and Kenneth L. Sokolo↵, “Institutions, factor endowments andpaths of development in the New World,” Journal of Economic Perspectives, 2000, 14 (3),217–32.

Esteban, Joan and Debraj Ray, “On the salience of ethnic conflict,” American EconomicReview, 2008, 98 (5), 2185–2202.

25

Gellner, Ernest, Nations and Nationalism, Ithaca: Cornell University Press, 1983.

Gourevitch, Peter Alexis, “The Remergence of ‘Peripheral Nationalisms’: Some Com-parative Speculations on the Spatial Distribution of Political Leadership and EconomicGrowth,” Comparative Studies in Society and History, 1979, 21 (3), 303–22.

Gupta, Amit Kumar and Arjun Dev, eds, Towards Freedom: Documents on the move-ment for Independence in India, 1941, New Delhi: ICHR and Oxford University Press,2010.

Gupta, Partha Sarathi, ed., Towards Freedom: Documents on the movement for Inde-pendence in India, 1943-1944, New Delhi: ICHR and Oxford University Press, 2010.

Hailey, Lord, The Future of Colonial Peoples, London: Oxford University Press, 1943.

Jha, Saumitra, “Shares, coalition formation and political development: evidence from17th century England,” GSB Research Paper 2005, Stanford Graduate School of Business,Stanford CA August 2008.

, “Sharing the Future: Financial Innovators and Innovation in Solving the Political Econ-omy Challenges of Development,” in Timur Kuran and Pranab Bardhan, eds., Institutionsand Development, Proceedings of the International Economic Association 2011.

and Steven Wilkinson, “Veterans, organizational skill and ethnic cleansing: evidencefrom the Partition of South Asia,” 2011. SSRN working paper 1739812.

Kamtekar, Indivar, “The End of the Colonial State in India 1942-47.” PhD dissertation,Cambridge University 1988.

Kranton, Rachel and Anand V. Swamy, “Contracts, Hold-Up and Exports: Textilesand Opium in Colonial India,” American Economic Review, June 2008, 98 (3), 967–989.

Kuran, Timur, “Now out of Never: The Element of Surprise in the East European Revo-lution of 1989,” World Politics, 1991, 44, 7–48.

Lawrence, Adria, Imperial Rule and the Politics of Nationalism, University of Chicago,2007.

Lipset, Seymour Martin, Political Man: the Social Bases of Politics, Johns HopkinsUniversity Press, 1960.

and Stein Rokkan, Cleavage Structures, Party Systems, and Voter Alignments: AnIntroduction, New York: Free Press, 1967.

Lopez-Cordova, J. Ernesto and Christopher M. Meissner, “The Impact of Inter-national Trade on Democracy: A Long-Run Perspective,” World Politics, 2008, 60 (4),539–75.

Lustick, Ian, Unsettled States: Disputed Lands, Ithaca, NY: Cornell University Press, 1993.

Mansergh, Nicholas, The Transfer of Power 1942-47, London: Her Majesty’s StationeryO�ce, 1976.

McAdam, Doug, Sidney Tarrow, and Charles Tilly, Dynamics of Contention, Cam-bridge University Press, 2001.

Metcalf, Barbara D. and Thomas R. Metcalf, A Concise History Of India, Cambridge:Cambridge University Press, 2002.

26

Miguel, Edward, Shanker Satyanath, and Ernest Sergenti, “Economic shocks andcivil conflict: an instrumental variables approach,” Journal of Political Economy, 2004,112 (4), 725–754.

Milner, Helen V. and Bumba Mukherjee, “Democratization and Economic Globaliza-tion,” Annual Review of Political Science, 2009, 12, 163–181.

and , “Democracy and Trade Policy in Developing Countries: Particularism and Do-mestic Politics with a Case Study of India,” in “Globalization, Democracy and TradePolicy in Developing Countries” 2011.

Moore, Barrington, Social origins of dictatorship and democracy: lord and peasant in themaking of the modern world, 1993 edition ed., Boston: Beacon, 1966.

North, Douglass, John Wallis, and Barry Weingast, Violence and social orders: a con-ceptual framework for interpreting recorded human history, Cambridge University Press,2009.

O’Rourke, Kevin H. and Alan M. Taylor, “Democracy and Protectionism,” workingpaper 12250, NBER 2006.

Panikkar, K.N., ed., Towards Freedom: Documents on the movement for Independence inIndia, 1940, ICHR and Oxford University Press, 2009.

Prasad, Bimal, ed., Towards Freedom: Documents on the movement for Independence inIndia, 1945, ICHR and Oxford University Press, 2008.

Rajan, Raghuram G., “The persistence of underdevelopment: constituencies and com-petitive rent preservation,” December 2006. manuscript.

Rothermund, Dietmar, India in the Great Depression, 1929-1939, New Delhi: ManoharPublications, 1992.

, The Routledge Companion to Decolonization, Routledge, 2006.

Stolper, Wolfgang F. and Paul A. Samuelson, “Protection and Real Wages,” TheReview of Economic Studies, 1941, 9 (1), 58–73.

Wittenberg, Jason, Crucibles of Political Loyalty: Church Institutions and Electoral Con-tinuity in Hungary, New York: Cambridge University Press, 2006.

27

Data Appendix

We detail the construction of our key dependent and independent variables here.

1937 election data

For each of British India’s 1,046 constituencies, we entered data on the following fields fromthe o�cial election returns (Secretary of State for India to Parliament 1937): total votespolled, votes polled and party a�liation for the winning candidate, the size of the electorate,the total number of candidates that ran for o�ce, the number of the seats (while 82% ofconstituencies were single-member seats, the rest had 2-4 members), and a variable indicatingthe type of constituency (general, general-urban, general-rural, reserved for scheduled castes,Muslims, Sikhs, Christian, Anglo-Indians, and some other small categories).

To collapse these data to the district level, we first mapped each constituency to anadministrative district or districts using the 1935 delimitation report (Secretary of State forIndia to Parliament 1936). In the 12% of instances where constituencies spanned districts,we divided the variables evenly between the districts. We then summed the variables acrossthe 199 districts of British India.

1942 Quit India data

These data are based on a series of secret reports written by the administration in responseto the Quit India agitations (Government of Bengal 1943, Government of Berar 1943, Gov-ernment of Bihar 1944, Government of India 1943a-h, Government of Madras 1943, Govern-ment of the United Provinces 1943). The reports provide detailed (often daily) accounts ofQuit India-related events. The Quit India dependent variable that we employ is the (log-transformed) count of the following events, between August and December 1942: violence,property damage, strikes, meetings, civil disobedience activities, and resignations.

1946 Congress primary membership data

These were obtained from a Congress Party handbook (All India Congress Committee 1946).Primary party members were required to pay an annual membership fee of four annas (equiv-alent to one-fourth of a rupee) a year. This entitled them to vote in party elections if theyhad maintained membership for a year. Data are divided by 100,000 and log-transformed.

Trade and shock data

Please see body of the paper.

Data sources

All India Congress Committee. 1946. Congress Handbook. Allahabad.Government of Bengal. 1943. “District O�cers’ Chronicles of Events of Disturbances

Consequent Upon the All-India Congress Committee’s Resolution of 8th August 1942 and

28

the Arrest of Congress Leaders Thereafter, August 1942 to the middle of March 1943.”Alipore: Bengal Government Press.

Government of Berar. 1943. “District Calendar of Events of Civil Disobedience Move-ment.” Nagpur: Government Printing, C.P. & Berar.

Government of Bihar. 1944. “Report on the Civil Disturbances in Bihar, 1942.” Patna:Government Printing, Bihar.

Government of India. 1943a. “Bombay: Six months of the Congress Movement.” NewDelhi: Government of India Press.

. 1943b. “A Brief Record of the Congress Movement, 1942-43, in the Province ofSind.” New Delhi: Government of India Press.

. 1943c. “Congress Disturbances in the N.-W.F.P, 1942-43.” New Delhi: Govern-ment of India Press.

. 1943d. “Congress Disturbances in the Punjab.” New Delhi: Government of IndiaPress.

. 1943e. “District Calendar of Events of the Congress Disturbances in Assam.” NewDelhi: Government of India Press.

. 1943f. “Disturbances at Delhi 1942: A Narrative Account.” New Delhi: Govern-ment of India Press.

. 1943g. “A Narrative Account of Congress Disturbances in Coorg.” New Delhi:Government of India Press.

. 1943h. “Narrative Account of Disturbances in Di↵erent Districts of the Provinceof Orissa, August-December 1942.” New Delhi: Government of India Press.

Government of Madras. 1943. “District Calendar of Events of the Civil DisobedienceMovement, August-December 1942.” Madras: Government Press.

Government of the United Provinces. 1943. “The Congress Rebellion in the UnitedProvinces, 1942.” Lucknow: Government Branch Press.

Secretary of State for India to Parliament. 1936. “Government of India Act 1935:Report of the Committee appointed in connection with the Delimitation of Constituenciesand connected matters.” Command paper 5,100. London: Her Majesty’s Stationery O�ce.

Secretary of State for India to Parliament. 1937. “Return Showing the Results of Elec-tions in India.” Command paper 5,589. London: His Majesty’s Stationery O�ce.

29

0.2

.4.6

Pr(

pers

on li

ves

in a

n in

depe

nden

t cou

ntry

)

1800 1850 1900 1950 2000Year

Source: Own calculations based on Cross-National Time-Series Data Archive.

Figure 1: World trends in decolonizationThe line marks the independence of India and Pakistan in 1947.

050

010

0015

0020

0025

00

1900 1910 1920 1930 1940 1950year

Imports-British Empire, Rs. Mils Exports-British Empire, Rs. Mils

Figure 2: India’s trade with the British EmpireSource: Mitchell: Historical Statistics

30

-100

-50

050

% c

hang

e in

val

. exp

ort g

oods

pro

duce

d in

dis

trict

, 193

3-19

23

0 10 20 30 40Rs.val. export goods produced per cap., 1920-23

(a) All districts

-100

-50

050

% c

hang

e in

val

. exp

ort g

oods

pro

duce

d in

dis

trict

, 193

3-19

23

0 2 4 6Rs.val. export goods produced per cap., 1920-23

(b) Excluding districts with initial export goods >Rs.10 in value per capita

Figure 3: Initial exports and Depression shocksSource: Own calculations, based upon Annual Statements of Foreign Trade of the United Kingdom and

various Censuses of India31

2040

6080

100

% T

urno

ut, 1

936

-100 -50 0 50% change in value of export goods in district, 1923-1933

Figure 4: Export shocks and % Turnout, 1937 electionsLocal polynomial smooths

32

Lege

ndIn

itial

Exp

ort V

alue

s, %

of T

otal

, (19

20-1

923

av)

initv

alto

t as

Per

cent

of T

otal

0% -

0.00

174%

0.00

175%

- 0.

0118

%

0.01

19%

- 0

.079

4%

0.07

95%

- 0.

18%

0.18

1% -

5.23

%

Nat

ive

Sta

tes

(a)Freetrad

e:Average

valueof

exportgo

odsper

producer,

%of

total,1920-23

Lege

ndEx

port

gai

ners

, 193

2-33

Exp

ort g

aine

rs, 1

932-

33

% c

hang

e in

val

exp

ort g

oods

, (19

23-1

933)

perc

vals

hkto

t-8

0.9

- -8

0.0

-79.

9 -

-60.

0

-59.

9 -

-40.

0

-39.

9 -

-20.

0

-19.

9 -

0.0

0.1

- 20.

0

20.1

- 40

.0

40.1

- 60

.0

(b)Im

perialpreferences:

%chan

gein

valueof

exportgo

odsper

producer,

1923-33

Figure

5:W

innersand

Losers

from

ImperialPreference

sand

theGreatDep

ression

Sou

rce:

Owncalculation

s,based

upon

the1931

census,1931

agriculturalcensusan

dvariou

seditionsof

theAnnualStatementofForeignTradeofthe

UnitedKingdom

33

020

4060

80%

Con

gres

s vo

te, 1

936

-100 -50 0 50% change in value of export goods in district, 1923-1933

Figure 6: Export shocks and % votes for Congress, 1937Local polynomial smooths

-30

-20

-10

010

20e(

% T

urno

ut, 1

936)

| X

-.4 -.2 0 .2 .4e(Pr. change exports/cap.) | X

-50

050

e(%

Con

gres

s vo

te, 1

936)

| X

-.4 -.2 0 .2 .4e(Pr. change exports/cap.) | X

-3-2

-10

12

e(Lo

g C

ongr

ess

Mem

bers

/100

,000

, 194

6) |

X

-.4 -.2 0 .2 .4e(Pr. change exports/cap.) | X

Figure 7: Partial residual plots

34

Lege

nd%

Con

gres

s vo

tes

1936

(qui

ntile

s)0.

00 -

12.0

6

12.0

7 - 2

2.83

22.8

4 - 4

5.67

45.6

8 - 6

3.33

63.3

4 - 8

8.07

Nat

ive

Stat

es

(a)1936

election

s:%

Con

gressvotes,

1936

election

s

Lege

ndC

ongr

ess

mem

bers

per

100

,000

(qui

ntile

s)

Nat

ive

Stat

es

(b)1946

Con

gressmem

bersper

100,000

Figure

8:Support

forCongress

priorto

Indep

enden

ceSou

rce:

Owncalculation

s,based

upon

o�cial

election

returnsan

dtheCon

gressParty

mem

bership

han

dboo

k1946

35

020

4060

80N

umbe

r of e

vent

s

July 1 October 1 January 1 April 1 July 1 October 1

Civil disobedienceAttempted violence

Figure 9: Civil disobedience preceded violent protest in the Quit India “Rebel-lion” of 1942Source: Own calculations, based upon secret intelligence reports for each province.

LegendNative States

Total incidents, Quit India 1942 (deciles)

01 -

23 -

67 -

19

20 - 3

5

36 - 4

7

48 - 6

3

64 - 9

3

94 - 1

22

123 -

695

Figure 10: Incidents of protest during the Quit India “Rebellion” of 1942Protests include: violence, property damage, strikes, meetings, other civil disobedience activities and resig-

nations. Source: Own calculations, based upon secret intelligence reports for each province.

36

Lege

ndN

ativ

e St

ates

Non

-vio

lent

civ

il di

sobe

dien

ce,

Qui

t Ind

ia 1

942

(dec

iles)

0

1

2

3

4 - 6

7 - 11

12 - 1

516

- 18

19 - 2

4 25 - 1

32

(a)Incidents:CivilDisob

edience,1942,Deciles

Lege

ndN

ativ

e St

ates

Viol

ent i

ncid

ents

, Qui

t Ind

ia 1

942

(dec

iles)

0

1

2

3 - 4

5 - 6

7 - 10

11 - 1

314

- 18

19 - 3

7 38 - 1

48

(b)Incidents:Violent

protest,1942,Deciles

Figure

11:In

ciden

tsofpea

cefuland

violentpro

test

duringth

eQuit

India

“Reb

ellion”of1942

Sou

rce:

Owncalculation

s,based

upon

secret

intelligence

reports

foreach

province.

37

Table 1: Summary Statistics

Variable Obs Mean Std. Dev. Losers GainersAll India

Congress membership per 100,000 459 9.7 16.4 12.6 2.9 ***Log. Congress membership per 100,000, 1946 459 1.3 1.5 1.7 0.4 ***Quit India event count 20 72.1 44.0 66.1 125.5Value export goods per worker, 1923 459 1.1 3.9 0.8 1.6Value export goods per worker- manufactures, 1923 459 0.2 0.3 0.2 0.2Value export goods per worker- natural resources, 1923 459 0.2 2.3 0.3 0.0 **Value export goods per worker- cash crops, 1923 459 0.6 3.3 0.2 1.5 **Value export goods per worker- staple crops, 1923 459 0.1 0.3 0.1 0.0 ***% change value export goods per capita, 1923-33 419 -32.0 32.6 -47.3 20.7 ***% gains: value export goods per capita, 1923-1933 419 4.6 10.0 0.0 20.7 ***% losses: value export goods per capita, 1923-1933 419 36.7 24.9 47.3 0.0 ***Gainer in value of export goods 472 0.3 0.5 0.0 1.0Log. Population, 1931 459 6.2 1.6 6.4 5.7 ***Population density, 100,000s/sqkm 446 0.3 2.1 0.2 0.7% Males in manufacturing industries, 1931 417 2.5 1.7 2.6 2.3% Males in agriculture, 1931 417 17.4 7.3 18.1 15.1 ***% Male non-cultivating landlords or tenants, 1931 417 0.5 0.7 0.5 0.4 **% Males owner-cultivators, 1931 417 5.3 5.5 5.6 4.6% Males unlanded agricultural labourers, 1931 417 3.7 3.7 4.0 2.5 ***Armymen per 100,000, 1931 417 1.5 6.2 1.8 0.7 **Police per 100,000, 1931 417 1.9 3.1 1.9 1.9Proportion Muslim 459 0.2 0.3 0.2 0.2 ***

British IndiaTurnout, % of eligible voters 204 56.7 12.5 57.1 52.2Congress vote, % of votes 188 43.7 23.8 44.2 39.3Number of candidates 204 16.4 9.7 16.3 18.0Number of seats 204 6.1 3.4 6.1 6.7Registered voters, 10,000s 204 13.5 10.1 13.7 11.5Congress membership per 100,000 203 19.1 18.6 19.5 15.2Log. Congress membership per 100,000, 1946 203 2.6 1.0 2.6 2.3Quit India event count 20 72.1 44.0 66.1 125.5Value export goods per worker, 1923 203 1.9 4.9 0.9 11.8 ***Value export goods per worker- manufactures, 1923 203 0.3 0.3 0.3 0.3Value export goods per worker- natural resources, 1923 203 0.2 1.5 0.3 0.0 *Value export goods per worker- cash crops, 1923 203 1.2 4.9 0.3 11.4 ***Value export goods per worker- staple crops, 1923 203 0.2 0.2 0.2 0.0 ***% change value export goods per capita, 1923-33 201 -47.4 22.5 -51.8 11.8 ***% gains: value export goods per capita, 1923-1933 201 0.8 3.3 0.0 11.8 ***% losses: value export goods per capita, 1923-1933 201 48.2 20.4 51.8 0.0 ***Gainer in value of export goods 204 0.1 0.3 0.0 1.0Log. Population, 1931 203 6.9 0.7 6.9 6.7Population density, 100,000s/sqkm 202 0.4 2.9 0.2 3.0% Males in manufacturing industries, 1931 200 2.7 1.7 2.8 1.6 ***% Males in agriculture, 1931 200 19.3 4.3 19.1 21.6 *% Male non-cultivating landlords or tenants, 1931 200 0.5 0.4 0.5 0.4% Males owner-cultivators, 1931 200 6.1 5.0 5.9 8.5% Males unlanded agricultural labourers, 1931 200 4.5 3.3 4.7 1.9 ***Armymen per 100,000, 1931 200 0.8 2.4 0.8 0.3Police per 100,000, 1931 200 1.5 1.2 1.5 1.1 **Proportion Muslim 203 0.2 0.3 0.2 0.2

Notes: *** p<0.01, ** p<0.05, * p<0.1 using two-sided difference in means Welch t-tests. Sources: Author's calculations. See text for details.

Mean

38

Table 2: Regression: % Turnout, 1937 elections

OLS with Native State / Province Fixed Effects (1) (2) (3) (4) (5) (6) (7)British

IndiaTrimming

ExportsBritish

IndiaBritish

IndiaBritish

IndiaBritish

IndiaBritish

IndiaValue export goods per worker, 1923 -0.516*** -1.656* -0.608*** -0.604** -0.432*** -0.679*** -0.658***

[0.119] [0.836] [0.134] [0.215] [0.139] [0.152] [0.113]Prop. change value export goods per capita -4.871 -9.391 -3.900 -34.493* -28.406*

[7.343] [7.173] [6.697] [15.870] [14.050]Prop. change value export goods per capita^2 -2.010 -7.216 -2.277 -30.954* -25.741*

[8.532] [8.595] [7.400] [16.762] [12.047]Gainer in value of export goods 10.171 15.392**

[6.058] [6.232]Gainer x % change in value 0.295 -0.584

[0.618] [0.567]% Gains: value export goods per cap., 1923-1933 0.274 0.041

[0.252] [0.227]% Losses: value export goods per cap., 1923-1933 0.047 0.024

[0.051] [0.055]% Males in manufacturing industries, 1931 -0.584 -0.640 -0.569

[0.721] [0.728] [0.719]% Males in agriculture, 1931 0.380 0.351 0.383

[0.337] [0.329] [0.337]% Male non-cultivating landlords or tenants, 1931 1.620 1.585 1.689

[2.157] [2.028] [2.105]% Males owner-cultivators, 1931 -0.828* -0.883** -0.828*

[0.398] [0.389] [0.397]% Males unlanded agricultural labourers, 1931 -0.544* -0.502* -0.543*

[0.259] [0.257] [0.266]Armymen per 100,000, 1931 -0.383 -0.364 -0.383

[0.442] [0.479] [0.438]Police per 100,000, 1931 -0.104 -0.009 -0.078

[0.835] [0.844] [0.847]Proportion Muslim 1.161 1.960 0.935

[3.849] [3.435] [3.632]Electoral controls Y Y Y Y Y Y YObservations 199 191 199 199 199 199 199R-squared 0.45 0.44 0.50 0.46 0.51 0.45 0.50

Robust standard errors in brackets, clustered at the Native State/ Province level. * significant at 10%; ** 5%; *** 1%; All regressions include controls for log. population 1931, population density.++: Electorate controls include: No of Candidates, No of Seats, No of Registered voters. Districts with 1923 export values per capita of Rs 10 are dropped in Col 2.

39

Tab

le3:

Reg

ression:%

Congress

Vote

Share,1937

OLS

with

Nat

ive

Stat

e/ P

rovi

nce

Fixe

d Ef

fect

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)

Brit

ish

Indi

aTr

imm

ing

Expo

rtsB

ritis

h In

dia

Brit

ish

Indi

aB

ritis

h In

dia

Brit

ish

Indi

aB

ritis

h In

dia

Brit

ish

Indi

aB

ritis

h In

dia

Brit

ish

Indi

aV

alue

exp

ort g

oods

per

wor

ker,

1923

0.61

7*-0

.634

0.09

6 -16

.228

***

0.53

7-15

.704

***

1.17

0***

-15.

361*

**1.

087*

**-1

4.87

3***

[0.3

07]

[2.3

83]

[0.4

64]

[3.4

72]

[0.3

15]

[3.2

78]

[0.2

21]

[3.4

23]

[0.3

09]

[2.8

80]

Prop

. cha

nge

valu

e ex

port

good

s per

cap

ita-5

4.48

7***

-36.

107*

-41.

884*

-41.

936*

*-5

4.49

2***

-42.

576*

*-5

3.92

3*-3

9.80

8[1

3.34

0][1

9.43

0][2

1.28

8][1

7.37

2][1

4.29

6][1

7.10

4][2

7.20

0][3

4.88

1]Pr

op. c

hang

e va

lue

expo

rt go

ods p

er c

apita

^2-7

0.70

5***

-54.

737*

*-3

5.86

2-4

9.96

8*-7

0.76

6***

-51.

571*

-68.

661*

-47.

901

[14.

740]

[21.

569]

[25.

219]

[25.

271]

[16.

929]

[25.

144]

[33.

006]

[42.

755]

Gai

ner i

n va

lue

of e

xpor

t goo

ds18

.501

18.4

07[1

2.49

1][1

6.26

4]G

aine

r x %

cha

nge

in v

alue

-2.1

89*

-2.3

37[1

.111

][1

.648

]%

Gai

ns: v

alue

exp

ort g

oods

per

cap

., 19

23-1

933

-2.3

66**

*-1

.892

***

[0.3

27]

[0.5

12]

% L

osse

s: v

alue

exp

ort g

oods

per

cap

., 19

23-1

933

-0.1

13-0

.06

[0.1

11]

[0.0

90]

% T

urno

ut-0

.367

**-0

.312

-0.3

83**

-0.3

41-0

.351

**-0

.307

[0.1

37]

[0.2

04]

[0.1

38]

[0.2

13]

[0.1

40]

[0.2

04]

% M

ales

in m

anuf

actu

ring

indu

strie

s, 19

314.

433*

2.76

7*2.

690*

2.71

5*2.

777*

[2.4

49]

[1.3

91]

[1.3

85]

[1.4

09]

[1.3

65]

% M

ales

in a

gric

ultu

re, 1

931

0.76

10.

818

0.89

10.

992

1.03

2[0

.738

][0

.650

][0

.647

][0

.671

][0

.634

]%

Mal

e no

n-cu

ltiva

ting

land

lord

s or t

enan

ts, 1

931

2.11

23.

444

3.69

3.96

24.

136

[7.5

92]

[3.4

52]

[3.5

82]

[3.5

13]

[3.5

18]

% M

ales

ow

ner-

culti

vato

rs, 1

931

-0.4

6-0

.428

-0.5

92-0

.795

-0.7

75[0

.673

][0

.718

][0

.793

][0

.828

][0

.784

]%

Mal

es u

nlan

ded

agric

ultu

ral l

abou

rers

, 193

1-0

.344

-0.3

61-0

.479

-0.5

35-0

.587

[0.9

23]

[0.9

94]

[1.0

25]

[1.0

25]

[1.1

16]

Arm

ymen

per

100

,000

, 193

1-0

.838

-0.3

3-0

.679

-0.7

55-0

.646

[0.5

24]

[0.7

22]

[0.7

28]

[0.7

68]

[0.7

54]

Polic

e pe

r 100

,000

, 193

13.

312*

1.14

1.11

61.

389

1.15

6[1

.601

][1

.557

][1

.709

][1

.673

][1

.633

]Pr

opor

tion

Mus

lim-2

7.63

-30.

15-2

6.90

5-2

5.43

3-2

7.67

4[2

0.45

4][1

7.87

2][2

0.25

2][2

0.75

8][1

8.89

1]%

Exp

ort s

hock

impl

ying

max

imum

Con

gres

s sup

port

-0.3

85-0

.330

-0.5

78-0

.420

-0.3

85-0

.413

-0.3

93-0

.416

Con

trols

for i

nitia

l val

ue b

y se

ctor

+N

NN

YN

YN

YN

YEl

ecto

rate

con

trols

++Y

YY

YY

YY

YY

YO

bser

vatio

ns18

317

518

318

318

318

318

318

318

318

3R

-squ

ared

0.52

0.53

0.58

0.60

0.54

0.61

0.54

0.62

0.54

0.61

Rob

ust s

tand

ard

erro

rs in

bra

cket

s, cl

uste

red

at th

e N

ativ

e St

ate/

Pro

vinc

e le

vel.

* si

gnifi

cant

at 1

0%; *

* 5%

; ***

1%

; All

regr

essi

ons i

nclu

de c

ontro

ls fo

r log

. pop

ulat

ion

1931

, po

pula

tion

dens

ity.+

: sec

tors

incl

ude:

Man

ufac

ture

s, N

atur

al re

sour

ces,

Cas

h cr

ops a

nd S

tapl

e cr

ops,

1920

-23.

++

: Ele

ctor

ate

cont

rols

incl

ude:

No

of C

andi

date

s, N

o of

Sea

ts, N

o of

Reg

iste

red

vote

rs.

Dis

trict

s with

192

3 ex

port

valu

es p

er c

apita

of R

s 10

are

drop

ped

in C

ol 2

.

40

Tab

le4:

Reg

ression:Log.Primary

Congress

Mem

bers,

1946

OLS

with

Nat

ive

Stat

e/ P

rovi

nce

Fixe

d Ef

fect

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)

Full

sam

ple

Dro

ppin

g A

hmad

.Tr

imm

ing

expo

rtsD

ropp

ing

Ahm

ad.

Dro

ppin

g A

hmad

.Fu

ll sa

mpl

eFu

ll sa

mpl

eD

ropp

ing

Ahm

ad.

Full

sam

ple

Dro

ppin

g A

hmad

.V

alue

exp

ort g

oods

per

wor

ker,

1923

0.02

30.

023

0.05

80.

025*

-0.0

250.

019

0.02

1-0

.022

0.02

0-0

.021

[0.0

16]

[0.0

16]

[0.0

45]

[0.0

14]

[0.0

29]

[0.0

16]

[0.0

15]

[0.0

33]

[0.0

17]

[0.0

28]

Ph

lt

dit

043

9**

042

6**

045

8**

039

9**

041

6**

228

2**

224

9**

195

7***

Prop

. cha

nge

valu

e ex

port

good

s per

cap

ita-0

.439

**-0

.426

**-0

.458

**-0

.399

**-0

.416

**-2

.282

**-2

.249

**-1

.957

***

[0.1

87]

[0.1

90]

[0.1

86]

[0.1

89]

[0.1

77]

[0.8

75]

[0.8

68]

[0.7

20]

Prop

. cha

nge

valu

e ex

port

good

s per

cap

ita^2

-1.1

69**

*-1

.091

***

-1.1

70**

*-0

.990

***

-1.0

43**

*-3

.255

***

-3.1

59**

*-2

.761

***

[0.3

83]

[0.3

92]

[0.3

94]

[0.3

20]

[0.3

23]

[1.0

51]

[1.0

18]

[0.8

66]

Gai

ner i

n va

lue

of e

xpor

t goo

ds0.

172

0.17

30.

127

[0.1

71]

[0.1

75]

[0.1

95]

Gai

ner x

% c

hang

e in

val

ue0.

035*

*0.

033*

*0.

029*

*[0

.014

][0

.013

][0

.012

]%

Gai

ns: v

alue

exp

ort g

oods

per

cap

., 19

23-1

933

-0.0

08**

-0.0

07**

[0.0

04]

[0.0

03]

% L

osse

s: v

alue

exp

ort g

oods

per

cap

., 19

23-1

933

-0.0

04-0

.002

[0.0

03]

[0.0

03]

[0.0

03]

[0.0

03]

% M

ales

in m

anuf

actu

ring

indu

strie

s, 19

310.

075

0.07

80.

076

0.07

9[0

.048

][0

.059

][0

.060

][0

.064

]%

Mal

es in

agr

icul

ture

, 193

1-0

.007

-0.0

07-0

.009

-0.0

08[0

.011

][0

.010

][0

.010

][0

.010

]%

Mal

e no

n-cu

ltiva

ting

land

lord

s or t

enan

ts, 1

931

0.09

4**

0.08

7*0.

079*

0.09

3**

[004

7][0

043]

[004

3][0

046]

[0.0

47]

[0.0

43]

[0.0

43]

[0.0

46]

% M

ales

ow

ner-

culti

vato

rs, 1

931

0.01

10.

012

0.01

30.

009

[0.0

14]

[0.0

13]

[0.0

14]

[0.0

13]

% M

ales

unl

ande

d ag

ricul

tura

l lab

oure

rs, 1

931

0.02

3**

0.02

4**

0.02

7**

0.02

5**

[0.0

11]

[0.0

12]

[0.0

12]

[0.0

12]

Arm

ymen

per

100

,000

, 193

1-0

.001

-0.0

01-0

.001

-0.0

02[0

005]

[000

5][0

005]

[000

5][0

.005

][0

.005

][0

.005

][0

.005

]Po

lice

per 1

00,0

00, 1

931

-0.0

41**

-0.0

41**

-0.0

38*

-0.0

42**

[0.0

18]

[0.0

19]

[0.0

20]

[0.0

18]

Prop

ortio

n M

uslim

-0.2

35-0

.214

-0.1

62-0

.298

[0.5

45]

[0.5

46]

[0.5

03]

[0.6

05]

% E

xpor

t sho

ck im

plyi

ng m

axim

um C

ongr

ess s

uppo

rt -0

.188

-0.1

95-0

.196

-0.2

02-0

.199

-0.3

51-0

.394

-0.3

54l

fi

iil

lb

Con

trols

for i

nitia

l val

ue b

y se

ctor

+N

NN

NY

NN

YN

YO

bser

vatio

ns40

540

439

640

440

441

740

540

441

740

4R

-squ

ared

0.82

0.83

0.83

0.83

0.83

0.83

0.83

0.84

0.82

0.83

Rob

ust s

tand

ard

erro

rs in

bra

cket

s, cl

uste

red

at th

e N

ativ

e St

ate/

Pro

vinc

e le

vel.

* si

gnifi

cant

at 1

0%; *

* 5%

; ***

1%

; All

regr

essi

ons i

nclu

de c

ontro

ls fo

r log

. pop

ulat

ion

1931

an

d po

pula

tion

dens

ity. +

: sec

tors

incl

ude:

Man

ufac

ture

s, N

atur

al re

sour

ces,

Cas

h cr

ops a

nd S

tapl

e cr

ops,

1920

-23.

Ahm

adab

ad- a

s loc

atio

n of

Gan

dhi's

ash

ram

at S

abar

mat

i w

as a

Con

gres

s hea

dqua

rters

, so

was

an

outli

er in

mem

bers

hip.

Dis

trict

s with

192

3 ex

port

valu

es p

er c

apita

of R

s 10

are

drop

ped

in C

ol 4

.g

q,

pp

pp

pp

41

Tab

le5:

Reg

ression:Non-violentand

ViolentPro

testsin

theQuit

India

‘Reb

ellion’,

1942

OLS

: Log

. Num

ber o

f inc

iden

ts

Full

sam

ple

Init.

Exp

ort

Sect

or

Con

trols

Trim

min

g Ex

ports

Full

sam

ple

Full

sam

ple

Full

sam

ple

Init.

Exp

ort

Sect

or

Con

trols

Trim

min

g Ex

ports

Full

sam

ple

Full

sam

ple

Full

sam

ple

Init.

Exp

ort

Sect

or

Con

trols

Trim

min

g Ex

ports

Full

sam

ple

Full

sam

ple

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

(13)

(14)

(15)

Val

ue e

xpor

t goo

ds p

er w

orke

r, 19

23-0

.011

***

-0.0

40-0

.036

-0.0

08-0

.008

-0.0

030.

010

0.03

6-0

.001

-0.0

01-0

.017

-0.0

070.

046

-0.0

16-0

.016

[0.0

04]

[0.0

30]

[0.0

64]

[0.0

06]

[0.0

06]

[0.0

02]

[0.0

31]

[0.0

32]

[0.0

03]

[0.0

03]

[0.0

12]

[0.0

25]

[0.0

95]

[0.0

15]

[0.0

15]

Prop

. cha

nge

valu

e ex

port

good

s per

cap

ita-0

.242

*-0

.386

***

-0.2

06-0

.259

*-0

.083

-0.1

11**

-0.0

76-0

.099

*-0

.294

**-0

.346

***

-0.2

41**

-0.3

01**

[0.1

26]

[0.1

10]

[0.1

24]

[0.1

30]

[0.0

53]

[0.0

51]

[0.0

53]

[0.0

53]

[0.1

31]

[0.0

98]

[0.1

03]

[0.1

38]

Prop

. cha

nge

valu

e ex

port

good

s per

cap

ita^2

-1.1

30**

-1.1

23**

-1.1

34**

-0.9

91**

-0.4

09*

-0.4

41*

-0.3

71-0

.376

-1.2

93**

*-1

.287

***

-1.2

05**

*-1

.160

**[0

.459

][0

.428

][0

.455

][0

.474

][0

.236

][0

.251

][0

.221

][0

.229

][0

.432

][0

.435

][0

.440

][0

.477

]%

Gai

ns: v

alue

exp

ort g

oods

per

cap

ita, 1

923-

1933

-0.0

09**

-0.0

03-0

.009

**[0

.004

][0

.002

][0

.004

]%

Los

ses:

val

ue e

xpor

t goo

ds p

er c

apita

, 192

3-19

33-0

.005

*-0

.002

-0.0

05*

[0.0

03]

[0.0

01]

[0.0

03]

Log.

Pop

ulat

ion,

193

10.

066

0.07

1*0.

061

0.09

60.

10.

017

0.01

70.

009

0.01

90.

021

0.04

60.

050*

0.03

80.

081

0.08

5[0

.040

][0

.039

][0

.040

][0

.061

][0

.061

][0

.021

][0

.022

][0

.021

][0

.022

][0

.022

][0

.030

][0

.028

][0

.029

][0

.054

][0

.055

]Po

pula

tion

dens

ity, 1

00,0

00s/

sqkm

0.41

40.

402

0.40

70.

421

0.43

70.

210

0.21

00.

216

0.18

60.

191

0.35

90.

345

0.33

90.

390

0.40

7[0

.337

][0

.300

][0

.337

][0

.375

][0

.380

][0

.204

][0

.194

][0

.203

][0

.208

][0

.210

][0

.286

][0

.260

][0

.269

][0

.344

][0

.350

]%

Mal

es in

man

ufac

turin

g in

dust

ries,

1931

0.08

1***

0.08

3***

0.03

1*0.

031*

0.06

7*0.

069*

[0.0

28]

[0.0

27]

[0.0

17]

[0.0

17]

[0.0

36]

[0.0

35]

% M

ales

in a

gric

ultu

re, 1

931

-0.0

05-0

.005

-0.0

06-0

.006

0.00

00.

000

[0.0

13]

[0.0

13]

[0.0

08]

[0.0

08]

[0.0

11]

[0.0

11]

% M

ale

non-

culti

vatin

g la

ndlo

rds,

1931

0.08

6*0.

092*

*0.

039

0.04

10.

065*

0.07

3*[0

.043

][0

.043

][0

.030

][0

.031

][0

.035

][0

.037

]A

rmym

en p

er 1

00,0

00, 1

931

-0.0

03-0

.003

0.00

10.

001

-0.0

01-0

.001

[0.0

04]

[0.0

04]

[0.0

02]

[0.0

02]

[0.0

04]

[0.0

04]

Polic

e pe

r 100

,000

, 193

1-0

.010

-0.0

11-0

.005

-0.0

05-0

.005

-0.0

06[0

.008

][0

.008

][0

.005

][0

.005

][0

.008

][0

.008

]Pr

opor

tion

Mus

lim-0

.385

-0.4

240.

083

0.06

7-0

.501

-0.5

50[0

.321

][0

.293

][0

.159

][0

.162

][0

.557

][0

.520

]%

Exp

ort s

hock

impl

ying

max

imum

inci

dent

s-0

.107

-0.1

72-0

.091

-0.1

31-0

.101

-0.1

26-0

.102

-0.1

32-0

.114

-0.1

34-0

.145

-0.1

75Pr

ovin

ce /

Nat

ive

Stat

e Fi

xed

Effe

cts

YY

YY

YY

YY

YY

YY

YY

YO

bser

vatio

ns40

940

940

040

940

940

940

940

040

940

940

940

940

040

940

9R

-squ

ared

0.85

00.

860

0.85

00.

860

0.86

0.73

00.

730

0.73

00.

730

0.73

0.69

00.

700

0.70

00.

700

0.7

Rob

ust s

tand

ard

erro

rs, c

lust

ered

at N

ativ

e St

ate/

Prov

ince

leve

l.* si

gnifi

cant

at 1

0%; *

* 5%

; ***

1%

. A

ll re

gres

sion

s inc

lude

Pro

vinc

e / N

ativ

e St

ate

Fixe

d Ef

fect

s as w

ell a

s con

trols

for L

og. P

opul

atio

n 19

31 a

nd

Popu

latio

n D

ensi

ty. (

2,7,

12) i

nclu

de se

para

te c

ontro

ls fo

r val

ue o

f exp

ort g

oods

per

wor

ker i

n na

tura

l res

ourc

es, c

ash

crop

s, m

anuf

actu

res &

agr

icul

ture

. (3,

8,13

) trim

dis

trict

s with

> R

s 10

expo

rts p

er c

apita

A

ll In

cide

nts

Non

-vio

lent

mas

s civ

il di

sobe

dien

ce

Vio

lent

mas

s pro

test

42

Tab

le6:

Reg

ression:Dec

omposingViolentPro

testsin

theQuit

India

Reb

ellion,1942

OLS

: Log

. Num

ber o

f inc

iden

ts, Q

uit I

ndia

194

2(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)(1

1)(1

2)V

alue

exp

ort g

oods

per

wor

ker,

1923

-0.0

14-0

.082

***

0.03

5-0

.011

-0.0

19-0

.047

0.01

8-0

.016

-0.0

13**

-0.0

25-0

.037

-0.0

11[0

.010

][0

.023

][0

.056

][0

.009

][0

.013

][0

.042

][0

.068

][0

.014

][0

.005

][0

.035

][0

.069

][0

.007

]%

gai

ns: v

alue

exp

ort g

oods

per

cap

ita, 1

923-

1933

-0.0

07**

-0.0

08**

-0.0

06**

-0.0

07**

-0.0

08**

-0.0

09**

*-0

.007

**-0

.007

**-0

.008

**-0

.010

***

-0.0

08**

-0.0

08**

[0.0

03]

[0.0

03]

[0.0

03]

[0.0

03]

[0.0

03]

[0.0

03]

[0.0

03]

[0.0

03]

[0.0

04]

[0.0

03]

[0.0

04]

[0.0

04]

% lo

sses

: val

ue e

xpor

t goo

ds p

er c

apita

, 192

3-19

33-0

.004

*-0

.003

-0.0

04*

-0.0

03-0

.004

*-0

.003

-0.0

05*

-0.0

03-0

.005

*-0

.004

-0.0

06*

-0.0

04[0

.002

][0

.002

][0

.002

][0

.002

][0

.002

][0

.002

][0

.003

][0

.003

][0

.003

][0

.003

][0

.003

][0

.003

]Lo

g. P

opul

atio

n, 1

931

0.05

1*0.

056*

*0.

036

0.07

5*0.

057

0.06

3*0.

043

0.09

20.

065*

0.07

0*0.

057

0.09

5[0

.027

][0

.027

][0

.024

][0

.043

][0

.034

][0

.034

][0

.030

][0

.058

][0

.038

][0

.037

][0

.038

][0

.059

]Po

pula

tion

dens

ity, 1

00,0

00s/

sqkm

0.09

0.07

30.

076

0.11

40.

392

0.37

40.

367

0.41

70.

387

0.37

30.

378

0.40

6[0

.105

][0

.072

][0

.090

][0

.097

][0

.329

][0

.280

][0

.309

][0

.353

][0

.320

][0

.277

][0

.318

][0

.355

]%

Mal

es in

man

ufac

turin

g in

dust

ries,

1931

0.07

2**

0.09

1**

0.07

5**

[0.0

28]

[0.0

45]

[0.0

30]

% M

ales

in a

gric

ultu

re, 1

931

0.00

2-0

.002

-0.0

03[0

.006

][0

.010

][0

.012

]%

Mal

e no

n-cu

ltiva

ting

land

lord

s, 19

310.

040.

086*

*0.

086*

*[0

.027

][0

.038

][0

.040

]A

rmym

en p

er 1

00,0

00, 1

931

-0.0

02-0

.003

-0.0

03[0

.002

][0

.004

][0

.004

]Po

lice

per 1

00,0

00, 1

931

-0.0

02-0

.011

-0.0

1[0

.007

][0

.009

][0

.008

]Pr

opor

tion

Mus

lim0.

014

-0.4

5-0

.423

[0.0

84]

[0.4

10]

[0.3

05]

Prov

ince

/ N

ativ

e St

ate

Fixe

d Ef

fect

sY

YY

YY

YY

YY

YY

YO

bser

vatio

ns40

940

940

040

940

940

940

040

940

940

940

040

9R

-squ

ared

0.59

0.62

0.6

0.6

0.79

0.8

0.79

0.8

0.85

0.86

0.85

0.85

Rob

ust s

tand

ard

erro

rs, c

lust

ered

at N

ativ

e St

ate/

Prov

ince

leve

l.* si

gnifi

cant

at 1

0%; *

* 5%

; ***

1%

. A

ll re

gres

sion

s inc

lude

Pro

vinc

e / N

ativ

e St

ate

Fixe

d Ef

fect

s as w

ell a

s con

trols

for L

og.

Popu

latio

n 19

31 a

nd P

opul

atio

n D

ensi

ty. (

2,6,

10) i

nclu

de se

para

te c

ontro

ls fo

r val

ue o

f exp

ort g

oods

per

wor

ker i

n na

tura

l res

ourc

es, c

ash

crop

s, m

anuf

actu

res &

agr

icul

ture

. (3,

7,11

) trim

di

stric

ts w

ith >

Rs 1

0 ex

ports

per

cap

ita

Des

truct

ion

of P

rope

rty

D

estru

ctio

n of

Rai

lway

Infr

astru

ctur

e

Des

truct

ion

of R

ecor

ds

43