accountability without democracy: lessons from african famines of the 1980s

Accountability Without Democracy:

Lessons from African Famines in the 1980s

A Thesis

Presented to

The Division of History and Social Sciences

Reed College

In Partial Fulllment

of the Requirements for the Degree

Bachelor of Arts

Daniel J. Sellers

May 2013

Approved for the Division(Political Science)

Mariela Szwarcberg

Acknowledgments

First and foremost, I must thank my parents and family, who have provided supportin all my endeavors. Without them I may very well not have made it to this point.Thanks go as well to all my friends, whether from Reed, Georgetown Day, or anyother context, for helping make daily life that much more exciting. Of course, Ialso must thank my professors, especially Mariela Szwarcberg, my thesis advisor, andChris Koski, my qual advisor, for providing guidance of all varieties, Xin Zhang, forintroducing me to Amartya Sen and the realities of international development, andthe members of my orals board, for providing valuable feedback and suggestions forfuture research. I will miss all of you, and value greatly our interactions and yourteachings.

Table of Contents

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Chapter 1: Literature Review . . . . . . . . . . . . . . . . . . . . . . . . 71.1 Entitlements, Acquirement, and Freedom . . . . . . . . . . . . . . . . 71.2 Famine Prevention in Post-Colonial Africa . . . . . . . . . . . . . . . 111.3 Conceptualizing Famine . . . . . . . . . . . . . . . . . . . . . . . . . 15

1.3.1 Identication . . . . . . . . . . . . . . . . . . . . . . . . . . . 181.3.2 Quantication . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

Chapter 2: Theoretical Argument . . . . . . . . . . . . . . . . . . . . . . 292.1 Democracy and Elections . . . . . . . . . . . . . . . . . . . . . . . . . 29

2.1.1 Press . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342.2 Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

Chapter 3: Empirical Evidence . . . . . . . . . . . . . . . . . . . . . . . . 413.1 Case Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413.2 Analytical Narratives . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

3.2.1 Botswana . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413.2.2 Zimbabwe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453.2.3 Kenya . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503.2.4 Uganda . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 573.2.5 Mozambique . . . . . . . . . . . . . . . . . . . . . . . . . . . . 613.2.6 Ethiopia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 673.2.7 Sudan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

Conclusion: Lessons From the African Cases . . . . . . . . . . . . . . . 81

Appendix A: Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

List of Tables

1 Case Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.1 Estimated Mortality of Select Twentieth Century Famines, Devereux(2000) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

1.2 Indexes of per capita food production in famine-aected and not-aected African countries, 1961{95/96 . . . . . . . . . . . . . . . . . 17

1.3 Famine Intensity, Howe & Devereux (2004) . . . . . . . . . . . . . . . 261.4 Famine Magnitude, Howe & Devereux (2004) . . . . . . . . . . . . . . 27

2.1 Polity IV Scores, Cases . . . . . . . . . . . . . . . . . . . . . . . . . . 322.2 Freedom House Scores, Cases . . . . . . . . . . . . . . . . . . . . . . 36

3.1 Zimbabwe: Ocial Maize Sales, 1979{85 . . . . . . . . . . . . . . . . 483.2 Food Production in Ethiopia, 1977-84 . . . . . . . . . . . . . . . . . . 693.3 Average Grain Prices in Northern Ethiopia, 1981{5 . . . . . . . . . . 703.4 Relief and Political Representation by Region, Sudan . . . . . . . . . 78

A.1 Translation of Polity IV Executive Recruitment Concepts and Compo-nent Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

A.2 POLITY Scores for All African Countries . . . . . . . . . . . . . . . . . 84A.3 XRREG Scores for All African Countries . . . . . . . . . . . . . . . . . 85A.4 XRCOMP Scores for All African Countries . . . . . . . . . . . . . . . . . 86A.5 XROPEN Scores for All African Countries . . . . . . . . . . . . . . . . . 87A.6 Press Freedom Scores for All African Countries . . . . . . . . . . . . 88

List of Abbreviations

ARAP Accelerated Rainfed Arable Production Programme

BDP Botswana Democratic Party

DRP Drought Relief Programme

EPLF Eritrean People's Liberation Front

FRELIMO Front for the Liberation of Mozambique

KANU Kenya African National Union

NCPB National Cereals and Produce Board

RENAMO Mozambican National Resistance

RRC Relief and Rehabilitation Commission

SPLA Sudan People's Liberation Army

SSU Sudan Socialist Union

TPLF Tigrayan People's Liberation Front

UNEOE United Nations Emergency Oce for Ethiopia

ZANU Zimbabwe African National Union

ZAPU Zimbabwe African People's Union

Abstract

Development economist and Nobel laureate Amartya Sen argues that democracies,by virtue of mechanisms of accountability, are better able to avert famines thannon-democratic regimes. Using empirical evidence from colonial and independentIndia, Sen claims an anti-famine political contract, between the government and itssupporters, predicated on the prevention of famine. Building on this theory, Senlater tested his argument using cases in Africa. Whereas Sen's theory accuratelypredicts the outcomes, the causal mechanism he uses to explain variation is falisable.In studying the experiences of each African country that succumbed to famine inthe 1980s, I nd that a free press and competitive elections are neither necessarynor sucient conditions for a country to avert famine. The results presented inthis thesis question the presence and role of other causes contributory to famineprevention eorts, as well as the possibility of anti-famine commitments within less-than-democratic polities.

Introduction

Despite modern advances in the production and distribution of food, episodes ofhunger and food insecurity still occur today in many of the world's least developedcountries. Though \hunger is not a modern malady," as Jean Dreze and AmartyaSen (1989:3) remind us, it is recurrent in parts of the world, and no civilization istotally immune. During the twentieth century, however, endemic famine was all buteliminated from regions outside Africa,1 and even within the continent there has beenconsiderable variation. While certain countries have managed to avoid famine, otherssuccumb repeatedly. Theories abound as to why this is the case, though perhaps thebest-known explanation comes from Amartya Sen, who argues that \with a relativelyfree press, with periodic elections, and with active opposition parties, no governmentcan escape severe penalty if it delays preventive measures and permits a real famine tooccur. That threat keeps governments on their toes" (Sen 1990b). Democracies, Senargues, are better equipped to prevent famines than non-democratic governments,which may be less responsive to the demands of the people. However, this argumentis problematic for several reasons.

First, while Sen's argument does accurately predict the outcome of various Africancountries' famine prevention eorts, in each case the inuence of their free press andcompetitive elections was less than clear, and in each case was dierent than in India,on whose experience Sen's argument was originally based. While Sen came to explainhis correlation using cases from Africa, it was derived from comparing the experiencesof China and pre-independence India, both of which succumbed to famine, to post-independence India, which has not. Briey, while in colonial India, famine preventionbecame linked with independence as a cause for nationalist sentiment, signaling itsimportance to the post-colonial government, in Africa, famine prevention has (largely)remained an administrative obligation.

Second, despite the predictive accuracy of Sen's theory, in each (African) case,the reality is slightly, yet signicantly dierent than Sen claims. Thus, while Sen usesEthiopia and Sudan as examples of countries that had neither competitive electionsnor free press and experienced famine, this does not tell the whole story. In the earlystages of Ethiopia's 1983-5 famine, the dictatorial government did not suppress infor-

1\The last famine in Europe occurred in the Soviet Union immediately after World War II, thelast famine in China was a by-product of the Great Leap Forward of 1958-62, and the last faminein South Asia (to date) occurred in Bangladesh in 1974. Occasionally famines due to exceptionalcircumstances aict Southeast Asia (Cambodia in the 1970s, North Korea in the 1990s), but famineas an endemic problem in Asia and Europe seems to have been consigned to history" (Devereux2000:3).

2 Introduction

Table 1: Selected Cases

Table 1a: Sen's Cases

Country Famine Democracy2 Press Freedom3

Botswana No Yes FreeEthiopia Yes No Not FreeKenya No No Not FreeSudan (pre-1985) Yes No Not FreeSudan (post-1985) Yes Yes Free4

Zimbabwe Yes No Partly Free

Table 1b: Additional Cases

Country Famine Democracy Press Freedom

Mozambique Yes No Not FreeUganda (pre-1984) Yes Yes Partly Free

mation about the impending crisis, with a brief exception during a time of nationalcelebration. Rather, clear predictions were published about the food shortage that,nevertheless, were met with no response from Western governments or the UN, dueto the country's domestic politics. Sudan also descended into famine under an au-thoritarian government, and though between 1986 and 1989 the country experienceda free press and competitive elections, these were not enough to diminish or eliminatefamine conditions. In truth, the democratic government was to blame not only forfailing to respond to the famine, but also for encouraging its continuation.

Botswana and Zimbabwe, on the other hand, are heralded by Amartya Sen andJean Dreze as \ `positive' examples of famine prevention, claiming that democratic ac-countability [was] instrumental in averting famine despite severe food crises" (de Waal1996:195). Although Zimbabwe did manage to avoid famine following drought in 1982,around this time the province of Matabeleland was subject to government suppres-sion of dissidents, which included the strategic withholding of relief. Though in 1982,Zimbabwe was fairly democratic, the next year marked the start of a downward trend,and by the end of the decade, the country was strongly autocratic. Throughout thistime, the press was decidedly not free. Sen's fourth case, Botswana, has maintainedan extended period of electoral democracy and has managed to avoid famine, thoughthe causality is dierent than in the Indian case: Since independence, Botswana has

2 (Marshall et al. 2011), Boolean average of POLITY from 1980{1980. See Appendix A andSection 2.1 (page 31) for tables and further discussion. Though POLITY measures more than just`executive recruitment' (elections), it is more reective of each country's overall political climate.

3 (Freedom House 2012). See Section 2.1.1 (page 34).4Freedom House reports Sudan's press as `Not Free' during this time, though de Waal reports

that during this time, \[Sudan's] press was uncensored and vigorous, and often highly critical of thegovernment" (de Waal 1997:93).

Introduction 3

been run by a single party which has maintained its power through patronage net-works, including relief programs. Throughout the decade, the country's press waslargely free from government intervention.

Finally, Sen considers Kenya, which managed to avert famine following droughtin 1983. Though Dreze and Sen attribute the government's responsiveness to eortsof elected members of parliament in the face of a de jure one-party state, limitedbut signicant press freedoms, and the threat of political instability, de Waal arguesthat Kenya's \success in escaping famine was largely attributable to the political as-tuteness of President Daniel arap Moi, who recognized a rudimentary and implicitpolitical contract: feed the central highlands and (most importantly) the cities, andthe government will survive" (de Waal 1997:35). However, despite arap Moi's appar-ent recognition of this political pressure, his regime \avoided tackling the structuralcauses of poverty and vulnerability" (de Waal 1997:42), and committed innumer-able human rights violations during and after this time (Adar & Munyae 2001).There are also reports of interference in press aairs by arap Moi's regime (Ochieng2011,Wanyande 1995) which further complicate the picture.

In an attempt to further clarify Sen's thesis, I will also consider the experiencesof Mozambique and Uganda, both of which succumbed to famine. Mozambique, oncea Portuguese colony, was ravaged by civil war from 1977 to 1992, and experiencedfamine twice throughout the 1980s: in 1983-4, from natural causes, and again in 1987from the conict. Throughout the civil war, the country did not hold elections, andthroughout the decade, the press was not free. Uganda also experienced famine, in1980, resulting from environmental, economic, and post-conict issues following thedeposition of Idi Amin the previous year, though the country's political situationotherwise around this time likely contributed as well. In 1980, the country's rstelections in eighteen years were held, though they resulted in the reelection of MiltonObote, who was previously deposed by Amin in 1971. Under Obote II, as his secondterm was known, the country's elections were not totally free and fair, though theywere rated as among the most legitimate in this report, and on the continent. WhileUganda's press was deemed partly free in 1980, these freedoms were tenuous and

uctuated throughout the decade.

A third complication with Sen's argument stems from a much more fundamentalquestion: for what reasons|if any|should democracy be supposed to have an advan-tage in ghting famine? A basic assumption of Sen's claim is that \civil and politicalrights|to free speech, to free association, to elect representatives of one's choice|contribute to the protection of social and economic rights|[such as] the right to foodand livelihood" (de Waal 2000:12). Indeed, throughout history, one of the most im-portant aspects of civil and political rights has been their use in promoting social andeconomic rights. However, though it seems logical that citizens of a democracy wouldexploit their civil and political liberties to ensure their protection against famine,the reality is not so simple. Certainly, abuses of social and economic rights can anddo occur in democracies, and often stem from a failure of related civil and politicalrights. How, then|if at all|is famine dierent?

Despite variation in each country's response to the threat of famine, across allcases there emerges a pattern: In each case, the dominant motivation underlying

4 Introduction

governmental attempts to prevent famine (or underlying government inaction) is thatits primary interests|maintaining power|depend upon it. Thus, in some cases, thereexists a political incentive to prevent famine. However, this incentive is by no meansunique to democratic polities. Indeed, in all democratic states there exist interests andinterest groups whose demands are not met, and an authoritarian government mayjust as easily derive some legitimacy from meeting certain demands of the populace|whether on the people's conditions or the leader's.

According to de Waal (2000), though, while eorts to this eect taken by auto-cratic or otherwise less-than-democratic governments may constitute an anti-famineprogram or commitment, these are mere privileges, granted to the population by thegrace of the ruler. Only in the context of \ `real' democratic institutions" (de Waal2000:11), it is argued, can an anti-famine contract emerge|a lasting and, crucially,enforceable solution to famine. However, while the dierence between a repressiveand a responsive government may be relatively easy to identify, the dierence, boththeoretical and practical, between a commitment and a contract is less than clear(though a program seems more denable), and raises further questions regarding thenature of political incentives and government responses to them.

As a result, we may extrapolate Sen's thesis to a universal form: mechanisms ofaccountability, such as a free press and competitive elections, promote responsive gov-ernments. Yet, the mechanisms Sen describes may be distinct from the institutionalform of government, or even any pre-existing institutional measures of accountability,which may exist without necessarily promoting any specic interest. It is apparentthat neither the form of government nor the openness of the press signicantly accountfor a government's responsiveness to famine among its citizens. Instead, preventiveaction to famine seems more heavily inuenced by the desire of those in power topreserve and perpetuate the regime, a motive more widely shared across regimes ofall forms.

Having established that the presence and application of factors instrumental toanti-famine measures is independent of the form and structure of government, thisthesis will attempt to reconcile Sen's hypothesis with the experiences of African coun-tries during the 1980s. In countries that experienced famine, what were the signicantcausal mechanisms? And in countries that avoided famine, how (if at all) did gov-ernmental action contribute to the eort? More importantly, though, why did thegovernment act at all?

Beyond Sen's initial cases|Botswana, Ethiopia, Kenya, Sudan, and Zimbabwe|Ihave also included Mozambique and Uganda, both of which also experienced faminein the 1980s. By reevaluating each case, I argue that a free press and competitiveelections|supposed preconditions to averting famine|are, in reality, neither neces-sary nor sucient to achieving this objective. While these mechanisms of account-ability are undeniably important to improving the quality of governance, they mustbe considered distinct from other specic issues, and from the political will to act.

The structure of the thesis is as follows. After this introduction is a review of Sen'sthesis over time, followed by a review of literature on the topics of famine, elections,and press, and ultimately, an elaboration of my argument. In the second chapter, Iprovide further background on each case, focusing on the countries' political regime

Introduction 5

and elections, operational freedom of the press, and food security and famine preven-tion eorts (or the lack thereof). Finally comes a conclusion, including implicationsand recommendations for future action.

Chapter 1

Literature Review

1.1 Entitlements, Acquirement, and Freedom

Though Sen's theory appears most explicitly in his 1990 speech Individual Freedom asa Social Commitment, the notion that a free press and competitive elections could holda government accountable stems from an earlier development by Sen: entitlements.Described as \the set of dierent alternative commodity bundles that [a] person canacquire through the use of various legal channels of acquirement open to someone in[their] position" (Sen 1990a:36)1, a person's entitlements not only reect the goodsavailable to them, but also the conditions under which such exchanges are carried out.Thus, a person who owns and lives o their land, for example, will establish commandover food in a very dierent way than a wage laborer who is paid in cash. While theformer conducts, so to speak, an exchange with `nature' (own-labor entitlement),the latter transacts with other members of society (trade-based entitlement2). It isof great importance, Sen argues, that these dierences in exchange conditions areconsidered in matters of hunger and food policy|without them, it is impossible tounderstand how people can or cannot acquire enough food.

The problem of establishing command over commodities such as food Sen callsthe `acquirement problem.' This problem, he explains, has been neglected throughouthistory, with Thomas Malthus and his intellectual descendants assuming the distribu-tion of food within a society to be secondary to the per capita level of food production.Related to this so-called `Malthusian pessimism'3 is `Malthusian optimism,' \not be-ing worried about the food problem so long as food output grows as fast as|or fasterthan|population" (Sen 1990a:35), which has historically helped justify delays in pol-icy responses to hunger in light of stagnant or rising output per capita. Though this

1Though this volume, which contains Food, Economics, and Entitlements was published in 1990,the chapter is noted as being an abridged version of a speech originally given in 1985.

2Sen (1981) also identies production-based and inheritance and transfer-based entitlement re-lations. While the latter is generally of limited relevance to famine theory, the former is relevantinsofar as it interacts with others' trade-based entitlement prospects.

3While population growth is supposed to be exponential, agricultural productivity is roughlylinear, such that eventually demand will outpace supply capacity, thereby causing mass starvationand/or population control|an equilibration of supply and demand.

8 Chapter 1. Literature Review

is a serious problem under conditions of \regular but non-extreme hunger (withoutstarvation deaths but causing greater proneness to morbidity and mortality)" (Sen1990a:35), it can be disastrous if and when famine develops without any decline infood production per capita (such as India's Bengal famine of 1943).

As a result, Sen advocates considering entitlements|the means by which peopleacquire commodities, including food|in addition to the level of food produced oravailable. Though this approach does not seek to provide a complete explanation offamine, Sen notes that \famine reects widespread failure of entitlements on the partof substantial sections of the population" (Sen 1990a:36). Because a person's enti-tlement is greatly inuenced by their `initial endowment,' starvation may be thoughtof as any commodity bundle that does not include sucient food. Thus, a person'sentitlements may reduce them to starvation due to changes in their endowment (e.g.loss of land or labor power), or their `exchange entitlement mapping' (e.g. fall inwages, rise in commodity prices, or changes in employment).

A few years later, Sen dramatically expanded the concept of entitlements byremoving its tangible boundaries. In Individual Freedom as a Social Commitment(1990), Sen argues that famine constitutes not only a breakdown of vulnerable groups'entitlements, but also violations of individuals' positive freedom to survive. Buildingon Isaiah Berlin, Sen distinguishes between `positive' freedom, involving \what, ev-erything considered, a person can or cannot achieve," and `negative' freedom, which\concentrates precisely on the absence of a class of restraints that one person mayexercise over another" (Sen 1990b).

To illustrate the dierence, Sen imagines he desires to walk in a park. Were hisinability to walk in the park due to a physical disability, only his positive freedomwould be aected. If, on the other hand, walking in the park would certainly meangetting beat up by a gang of thugs, his negative freedom is aected as well. However,while Berlin interprets positive and negative freedom as disjoint, Sen argues thatany violation of negative freedom is also a violation of positive freedom, though theconverse does not hold. Thus, while the thugs act as a restraint on Sen's ability towalk in the park (a violation of his negative freedom), they also diminish his abilityto realize his desires of walking in the park (a restraint of positive freedom). Thoughthis seems redundant, the crucial distinction is between violations of freedom as beingof a `positive' versus a `negative' variety|incapacity versus prohibition.

Libertarianism, for example, focuses solely on freedom from restraint. However,it is seen easily enough that these (negative) freedoms may yield violations of others'(positive) freedom as well. Thus, Sen argues that poverty is not necessarily a violationof negative freedom: \A person in extreme poverty is not free to do many things (e.g.,feeding his family well, staying home when riots threaten his life), but the poverty andconsequent failure of positive freedom may not be due to interference by others" (Sen1990b). Rather, poverty and famine represent serious violations of a person's positivefreedom to survive. Despite the change in language, though, Sen's policy solutionremains the same: focus on vulnerable groups' entitlements, and changes thereof.Within this framework, explanations of famine would take the form of economic andpolitical changes that alter various groups' relative economic power.

By way of these considerations, Sen argues that the dierence between India prior

1.1. Entitlements, Acquirement, and Freedom 9

to independence|which experienced the Bengal famine of 1943|and India post-independence|the push for which was provoked in no small way by the Bengal famineof 1943|was not the formation of a famine response policy, but rather a willingnessto invoke the policy when necessary (Sen 1995).

Though India's Famine Codes of 1880 were designed under colonial administra-tion, they were rarely invoked in time, if at all. Since independence, though, famineprevention measures have been used judiciously. Thus, Sen concludes, the govern-ment's accountability to its people, as made paramount by the country's democraticrevolution (and its resulting political contract), is the causal factor in the coun-try's newfound ability to avert famine. This case, then, illustrates that \one setof freedoms|to criticize, to publish, to vote|are causally linked with other typesof freedoms, such as the freedom to escape starvation and famine mortality" (Sen1990b).

In colonial India, it was the reluctance of the British Government to take respon-sibility for famine relief that helped discredit it and foment nationalist movements,who used famine as an issue for mobilization. Eventually, \sustained political agita-tion forced the government to take serious anti-famine measures, implicitly acceptinga social contract" (de Waal 1997:12). The Bengal famine of 1943 represented a signif-icant breach of this contract, and independence came four years later. Post-colonialgovernments have continued to honor this anti-famine contract, and continue to beheld accountable by the press and competitive elections (Dreze & Sen 1989:212).

On the other hand, Sen considers China, which experienced no such populardemocratic uprising. As a result of the country's Great Leap Forward social andeconomic reforms|which attempted to rapidly modernize the country's economythrough processes of industrialization and collectivization|a tremendous famine be-fell them. From 1958-61, between sixteen and thirty million people are estimatedto have died (Dreze & Sen 1989:210), though other estimates place the toll evenhigher. At this time, \the government faced no pressure from newspapers, whichwere controlled, or from opposition parties, which were absent" (Sen 1995). Indeed,the Chinese government itself was misled by the lack of free reporting, with its ownpropaganda and party ocials|all competing for credit in Beijing|distorting orfabricating statistics from around the country (Dreze & Sen 1989:213).

Following \the demise of [Soviet] communism in 1989 and the joint triumphof political and economic liberalism" (de Waal 1996:194), though, Sen's argumentgained a new lease on life. However, instead of India and China, the new focus wason ve African cases: democratic Botswana and Zimbabwe and the questionably-democratic Kenya, which averted famine in the early 1980s, and authoritarian Sudanand Ethiopia, which succumbed. Though Sen undoubtedly considers his theory robustto the circumstances of individual countries, there are important dierences amongthe countries to consider that present a challenge to Sen's hypothesis.

As in India, under colonialism, African governments faced little pressure to intro-duce anti-famine measures. Rather, it was only towards the end of the era that therearose any \sense of administrative responsibility towards colonial subjects" (de Waal1997:26). However, before then, colonialism in the area was primarily exploita-tive, and relied on military power. In the early twentieth century, for example,


famine was brought about in British Tanganyika (later Tanzania) (Little 1991) andGerman Ruanda-Urundi by colonial troops, in an attempt to suppress the popula-tions (de Waal 1997).

Elsewhere in Africa, though, colonial administrations had begun to acknowledgefamine as a detriment to the local population. Following famine and food insecurityin Sudan during the 1910s, the colonial government attempted to transplant theIndian Famine Code to the country, resulting in the 1920 Sudan Famine Regulations.However, Sudan was very dierent than Madras, the Indian state on whose policiesSudan's were based. First, there was not enough administrative capacity in Sudanto organize a system necessary to predict and prevent famine, let alone to undertakethe prescribed relief eorts. Second|reective also of \the generic dierence between(most) South Asian famines and (most) African ones" (de Waal 1997:28)|was thatwhile in India, those most at risk of famine were agricultural laborers and ruralartisans, in Sudan, those most at risk were largely pastoralist farmers and herders.Crucially, while the fortunes of the former depended on market conditions (i.e. foodprices and employment), the wealth of the latter was held primarily in assets (suchas livestock), meaning that famine in Sudan would be harder to predict but slower toarrive, reducing the advantage of responding quickly4. The area's lack of transportinfrastructure and the absence of any food markets also meant that \the deferenceof the regulations to classical economics was inappropriate" (de Waal 1997:28). Inreality, de Waal argues, the Sudanese Famine Regulations were constructed primarilyto deter political threats by prioritizing for subsidized food government employeesand townspeople, followed by those in ood-prone riverine areas, while dwellers ofmore remote regions would be lucky to receive any assistance whatsoever.

Though Sudan's famine codes of the early twentieth century are notable for beingsomewhat inappropriate given their context, other African colonies also experienced`relief' policies which provided a bare minimum of assistance. While in certain in-stances, the legacies of colonial administrations did help instill an expectation ofemergency relief provision (e.g. Tanzania), this was not the norm. There is thus aclear dierence between the approaches of famine-prevention regimes in India and inAfrica: in the former, famine relief had become a right, while in the latter it was stillseen as little more than an administrative duty. However, the situation was dierentin African cities, where, due to the growth of trade unions and nationalist sentiment,keeping the people fed became a political imperative, and consequently was perceivedas a right.

Politics of famine prevention in colonial Africa were further complicated by dis-putes regarding land use and soil conservation, with colonial governors seeing tra-ditional techniques as backwards and damaging to the environment. As one mightexpect, the policies which resulted from this view generated resentment and resistancethat signicantly helped strengthen nationalist sentiment. Policies dealing with epi-demic disease control were met with similar hostility, owing both to their rudimentary

4In terms of entitlements, while the wealth of agricultural laborers and rural artisans is primarilytrade-based (or possibly production-based), that of pastoralist farmers and herders is more along thelines of own-labor entitlements.

1.2. Famine Prevention in Post-Colonial Africa 11

nature (often involving population relocation) and the authoritarian manner in whichthey were implemented.

Even as African countries achieved independence, there were no nationalist move-ments that utilized the prevention of famine as a party platform. Instead, other issueslike racial identity and economic development tended to prove more salient, reectingboth a relative lack of rural civic mobilization, and the problematic legacy of colonialanti-famine land-use policies.

In promising macroeconomic and social welfare development, these regimes pur-sued technocratic policies which in some instances succeeded in averting famine.Ultimately, though, the authoritarian nature of the governments undermined theirrelief programs, leaving those most vulnerable unable to protect themselves. Thus,while systematic anti-famine policies were implemented in colonial Africa, they weremarkedly simpler than their Indian counterparts.

1.2 Famine Prevention in Post-Colonial Africa

Perhaps the clearest example of a minimal commitment to famine relief is Sudan.A British colony until 1956, the country descended into civil war almost immedi-ately after gaining its independence. Though at this time, scholars note, the countryexhibited structures which served to guard against localized food shortages degener-ating into famine, their eectiveness was undermined by political instability. In 1969,Jaafar Nimeiri seized power and embraced political Islam, exacerbating the conictbetween the country's Northern (Muslim) and Southern (Christian) halves. Nimeiri's(corrupt) regime destroyed its domestic accountability during this time, exploitingits strategic position in the West's anti-communist strategy, and the state began todecay domestically.

In 1983, the specter of famine loomed in a few provinces, and though rains failedagain in 1984, the government took no action. In an attempt to persuade interna-tional nanciers of his country's stability, Nimeiri stated publicly that at that time,the situation in the South|civil war, by all accounts|was \reassuring" (de Waal1997:91). Though Nimeiri did ultimately acknowledge the gravity of the conditions,the famine helped solidify a coalition against him, resulting in a popular democraticuprising in 1985. However, though the coalition used the famine to mobilize sup-port, its use was largely tactical, such that the issue failed to attain any real politicalsignicance.

Following the regime change, Sudan's famine response eort was delegated tointernational relief agencies and became a depoliticized, technical issue. This shift wasreected in the creation of a technical early warning system that monitored economic,climatic, and agricultural data, in the spirit of the \neo-liberal agenda for evisceratinggovernment responsibilities" (de Waal 1997:93). Though under the democratically-elected Umma Party (1986-9), the country did possess liberal democratic institutions,in practice they did not extend beyond Khartoum and other (Northern) urban centers.War in the South continued to rage, as did the famine, and though at this time thepress was unrestrained, there was little interest in covering either situation.


Around this time, Ethiopia also failed to prevent famine, though the country'sprospects were not nearly so bright as Sudan's after Nimeiri. Though Ethiopia wasnever colonized by a European power, Haile Selassie's Solomonic empire came toan end in 1974 when he was deposed by a Soviet-backed Marxist-Leninist militaryjunta|the Dergue (\council" in Ge'ez). Led by Mengistu Haile Mariam, the countryimmediately became a one-party communist state. While after the country's 1973-4food crisis, the imperial government established a Relief and Rehabilitation Commis-sion (RRC) to help prevent future crises, following the revolution the Commissionquickly became a tool of the Dergue. Land reforms enacted around this time weredesigned by a small number of intellectuals, without the involvement of the peasantbase, and imposed from above. The Dergue also established a central grain market-ing corporation which extracted food from rural populations to sustain a few citiesand the ever-expanding army|\precisely the same pattern of distribution which pre-vailed before the revolution" (de Waal 1997:110). During the 1980s, the RRC becamea political tool to implement radical social reforms, and later the military used it torelocate people. The RRC was also tasked with procuring foreign aid|most of whichwent to fuel militias|and spreading propaganda identifying the country's 1983-5famine as the result of overpopulation and drought, as opposed to war.

However, though drought and harvest failure did contribute to the famine, theycannot be considered direct causes. Even the government's economic and agriculturalpolicies were not as integral to the famine's sustainment as the counter-insurgencycampaigns in Tigray and Wollo: \The zone of severe famine coincided with the warzone, and the phases of the developing famine corresponded with major military ac-tions" (de Waal 1997:115). After the famine was uncovered by international media inthe fall of 1984, the Dergue began to use \aid as a strategic alibi" (de Waal 1997:123),with a newly-formed UN Emergency Oce for Ethiopia acting as an international of-cial mouthpiece. On the other side of the frontline, the Tigrayan People's LiberationFront (TPLF) was much more cognizant of their reliance on the rural population, andso aligned their interests. By \[linking] the political fortunes of the Tigrayan peas-antry in the face of famine to the political fortunes of the TPLF" (de Waal 1997:129),the TPLF was also able to frame the government as having genocidal intentions.

Ultimately, de Waal explains, Ethiopia's famines of the 1970s and 1980s are \re-plete with ironies, which are explicable only by attention to the existence or not ofan anti-famine political contract" (de Waal 1997:132). While the failed response tothe 1973 famine was elitist and institutionalized (and resulted in the fall of HaileSelassie), the response to the 1983-5 famine was, for a short time, thought of as amajor policy success for the government of Ethiopia. Due to the way discourse aboutghting famine is disconnected from discourse of internal conict, though, the TPLFhas largely been unable to take credit for its actions.

In Mozambique, too, did portions of the country experience famine conditions ow-ing to domestic conict. Even before Mozambique's 1975 independence from Portugal,the territory was subject to a decade of sporadic warfare. Additionally, following theformation of the new government|a one-party state based on Marxist principles|most of the country's roughly 250,000 Portuguese returned to Europe, leaving theeconomy in shambles. Civil war broke out in 1977, and lasted until 1992. Though

1.2. Famine Prevention in Post-Colonial Africa 13

during times of peace, Mozambique is a fertile country which has little trouble feed-ing its population and producing exports, between 1983 and 1984, the south-centralregion descended into famine, with most commonly cited statistics placing the deathtoll at around 100,000. A second major famine occurred in 1987, centered on theeastern Zambezia province, which was the result of conict. The civil war was char-acterized by extensive human rights violations by both sides, and only ended afterthe accidental death of the president and the end of the Cold War.

Though political instability in Uganda arguably did not directly contribute tothe country's famine in 1980, the situation was indeed exacerbated by confusion andfragility following the fall of Idi Amin. After being deposed in 1979, Amin and hisforces ed the the country, and around the same time, crisis struck Karamoja, a re-gion periodically aicted by drought. Amin's soldiers' abandonment of a barrackswithin Karamoja further contributed to the instability, as traditional power rela-tions were upset, and a power struggle following (possibly rigged) elections resultedin Karamoja|already a marginalized region|being caught in the crossre. In theend, an estimated 50,000 Karimojong perished. Though the human rights record ofMilton Obote's regime was below par, the country did manage to avoid famine inKaramoja following drought in 1984, just before a coup removed Obote from thepresidency. Dodge and Alnwick (1986), though, attribute this success entirely toeorts by international aid organizations.

At the other end of the spectrum lies Botswana, whose government is noted for ex-hibiting an inherent sense of obligation and a keen sense of its electoral interests. Hav-ing gained independence from Britain in 1964, Botswana has consistently maintainedwhat Freedom House judges as free and fair elections. In reality, though, the BotswanaDemocratic Party (BDP) has \beneted from an uneven playing eld" (Levitsky &Way 2010:255)5, winning every contest since independence by a considerable margin.When drought struck in 1982 (and continued for six years), the government enactedvarious programs in an attempt to prop up rural incomes. Though the country's useof plow and tractor subsidies was socially regressive, the country did manage to avoidfamine conditions. There are some, however, who argue that the country's successwas in no small part bolstered by its economic growth throughout the decade, ratherthan any political commitment to relief. The country's constitution does not containexplicit guarantees for freedom of the press, and though the government occasionallyinterferes in media aairs, abuse by the BDP is \neither as frequent nor as severe asother countries in the region" (Levitsky & Way 2010:256).

Though Dreze and Sen similarly laud Zimbabwe for enacting eective entitlementprotection programs, preventing the country's 1982-4 drought from precipitating amajor famine, this is far from a complete picture. After gaining independence in 1980,Robert Mugabe and his ZANU party were elected to power. Less publicized than thecountry's aversion of famine, however, was the ZANU military campaign conductedwithin the province of Matabeleland, home to supporters of the rival ZAPU party6.This campaign|Gukurahundi7|resulted in widespread famine conditions and the

5XRCOMP in Polity IV. See Section 2.1 (page 31).6While ZANU was Western-aligned, ZAPU supported the USSR.7Gukurahundi, meaning \the early rain which washes away the cha before the spring rains," in


deaths of between 10,000 and 20,000 people. Though this was obscured partly byZimbabwe exploiting its strategic position between East and West, the Gukurahundiwas also kept secret by restrictions on the content of news reports. It was not until1987 that the campaign ended, following an agreement to unify ZANU and ZAPU(forming ZANU-PF). However, this relationship broke down some years later, andthough the country has experienced \limited multi-party democracy" (Dreze & Sen1989:146), in practice, Mugabe has held power throughout.

Last, but not least, Dreze and Sen consider the case of Kenya. After peacefullygaining independence from the British in 1963, Kenya became a republic, with JomoKenyatta as its president. Political pressure prompted Kenyatta to consolidate powerin 1966 after winning re-election, and established a de facto one-party state. Uponhis death in 1978, then-Vice President Daniel arap Moi took his place, and thoughhe enjoyed support from around the country, he was too weak to consolidate power.However, in 1982, some ocers of the air force attempted and failed to overthrow him,and arap Moi dismissed political opposition, establishing a de jure one-party state.In 1984, the country experienced a severe drought, in response to which an `inter-ministry drought response coordinating committee' was established and given toppriority. Ultimately, the country managed to avert famine, though there is evidenceof widespread hunger leading to malnutrition.

While Dreze and Sen attribute the government's responsiveness to political pres-sure from elected MPs, the (somewhat free) press, and the threat of political instabil-ity, de Waal argues that Kenya's \success in escaping famine was largely attributableto the political astuteness of President Daniel arap Moi, who recognized a rudimen-tary and implicit political contract: feed the central highlands and (most importantly)the cities, and the government will survive" (de Waal 1997:35). However, despite rec-ognizing this pressure, action taken by arap Moi's regime seemed oriented primarilytowards the placation of his opponents, failing to address the country's more funda-mental causes of poverty. Additionally, there are reports of numerous human rightsviolations during and after this time (Adar & Munyae 2001), as well as of govern-mental interference in press aairs (Ochieng 2011, Wanyande 1995) which furthercomplicate Dreze and Sen's argument.

To summarize, then: while Botswana and Kenya managed to avert famine condi-tions, neither were fully democratic. On the other hand, Sudan, Ethiopia, Mozam-bique, and Uganda were all autocratic, and all failed to avert famine conditions (inmost cases due to war-time politics). Finally, on rst glance, Zimbabwe also man-aged to avert famine, though in reality, the government indirectly created hunger tosuppress political opposition.

However, while anecdotal evidence is largely sucient to reconcile Sen's theorywith reality, it cannot provide general (or generalizable) denitions of famine, com-petitive elections, or a free press|all of which may be used to further examine thehypothesis' validity. While for each variable, a simple binary measure (e.g., free pressvs. restricted press) would seem to suce, in reality it is the gradations betweenwhich provide information pertinent to this endeavor, as they often reect decisions

the Chi-Shona language, was also known as the Matabeleland Massacres.

1.3. Conceptualizing Famine 15

and actions of a political variety.

1.3 Conceptualizing Famine

While most popular denitions of famine include scenes of malnutrition and starva-tion resulting from a general lack of food, this is somewhat of a simplication|thoughfamine conditions may obviously result in starvation and destitution, these symptomsare generally indicative of breakdowns of other social systems as well. Exactly whatsystems are aected, though, and how, is a matter of debate. While some scholarsargue that famine is the result of a decrease in relative exchange power of (vulnerable)populations (Ilie 1989,Sen 1981,Walker 1989), others emphasize nutrition (Dreze &Sen 1989, O Grada 2009), disease and social disruption (de Waal 1990), excess mor-tality (Howe & Devereux 2004), or other indicators as both reecting and stemmingfrom `famine conditions.' Clearly, then, we are faced with a complex phenomenon.Seeing as denitions of nearly every aspect of famine are debated, it will serve us wellto start with classical denitions of this unfortunate occurrence.

Table 1.1: Estimated Mortality in Select Twentieth Century Famines

Years Location (epicenter) Excess Mortality Causal Triggers

1943 India (Bengal) 2.1m{3m Conict1958-62 China 30m{33m Government1969-74 West Africa (Sahel) 101k Drought1972-75 Ethiopia (Wollo & Tigray) 200k{500k Drought1980-81 Uganda (Karamoja) 30k Conict & Drought1982-85 Mozambique 100k Conict & Drought1982-85 Zimbabwe (Matabeleland) 10k{20k Conict1983-85 Ethiopia 590k{1m Conict & Drought1984-85 Sudan (Darfur, Kordofan) 250k Drought1988 Sudan (south) 250k Drought

(Devereux 2000:6)

Early 'modern' scholars of famine were inuenced by Thomas Malthus, who fo-cused on the gap between food supply and demand, given a level of technologicaladvancement and a xed amount of land. In his 1798 work An Essay on the Princi-ple of Population as it Aects the Further Improvement of Society, Malthus identiedfamine as \a shortfall in the supply of food in a given area and, simultaneously, thedeath by starvation of a substantial proportion of the inhabitants" (de Waal 1997:20).However, de Waal notes, the picture painted by this description|one of great faminelooming constantly|is entirely without empirical foundation.

By the 1800s, though, Malthusian doctrine had reached a level of general popular-ity such that by the 1840s, Malthusianism was a guiding principle of British policy. Ofthe famine in Ireland, London believed that certain natural economic forces were to


blame, and that these forces should be left alone|eventually, the population wouldreduce on its own to match the food supply. As a result, relief was delayed andvery modest, and food was even permitted to leave the island. Malthus also inu-enced policies of the East India Company, which made him a professor of historyand political economy at its college in Haileybury. However, by the time of the 1880Famine Commission, doctrinaire Malthusianism had fallen somewhat out of fashion,likely sparked by the realization that in most circumstances, the theory failed to cor-respond to reality. The 1880 Report of the Famine Commission acknowledged thatfamine was caused not by a general lack of food, but by a general lack of purchasingpower, and in 1911, Lovat Fraser described famine as \now [meaning] a prolongedperiod of unemployment, accompanied by dear food..." (de Waal 2005:19). Despitecriticisms of Malthus' theory, though, it has remained popular since its inception, hisvision of a gigantic famine decimating the population used even today to instill fearsof unsustainable population growth (Datta-Ray 2013,Hassan 2013).

Though Malthus' theory of overpopulation leading to famine has fallen out offavor, the question of whether famines stem from `natural' or `articial' causes is stillpertinent. Sen, however, claims that the distinction can be misleading: famine, heargues, is fundamentally a social phenomenon, involving \the inability of large groupsof people to establish command over food in the society in which they live" (Dreze &Sen 1989:46) in the face of adverse meteorological conditions. Furthermore, though anatural catastrophe may cause (or exacerbate) food insecurity, a disaster's impact willnonetheless depend upon how a society is organized. For example, a country withan extensive irrigation network could theoretically weather a drought much betterthan one without such redistributive infrastructure. That said, even the existenceof droughts, oods, and other calamities is not independent of social and economicpolicies. In Sub-Saharan Africa, for example, problems of famine and hunger areoften seen as related to climate change, overlooking the inuence of societal factorson how people produce and consume food.

As Table 1.2 shows, though countries which experienced famine did also experiencea decline in food production, so did Botswana and Zimbabwe|both countries \with-out a noted famine during the 1970s, 1980s, or 1990s" (von Braun et al. 1998:32).Though this table alone cannot establish food production as a (causal) factor of coun-tries' famines, it is interesting to note that almost all the countries examined in thisreport experienced a decline in production between 1971, 1981, and 1991 (only Kenyamanaged to increase its food production after the 1980s). This chart, however, leavesmuch to be desired, including but not limited to production statistics for the periodsbetween evaluations.

To those in power, the identication of famine as primarily resulting from droughtor other natural causes can dramatically reframe lines of accountability, and may sig-nicantly alter the policy response. Indeed, the identication of famine conditions assuch at all \represents a choice and is therefore more political than technical: ratherthan being `found,' a denition of famine must be `agreed [upon]' " (Howe & Devereux2004:4). The action and outcome of making this decision has great implications, in-cluding who may be held responsible for providing relief, and to whom. When famineis blamed on natural causes, as with the Sahelian famines of the 1970s|supposed


Table 1.2: Indexes of per capita food production in famine-aected and not-aectedAfrican countries, 1961{95/96

Country 1961 1971 1981 1991 1995/96

Countries aected by famine at least once during the 1970s, 1980s, or 1990sAngola 126 136 96 79 88Burundi 100 104 102 91 82Chad 111 103 99 102 110Ethiopia 113 104 96 86 96Liberia 91 99 102 66 n.a.Madagascar 104 109 100 86 80Malawi 86 108 100 75 70Mozambique 117 112 99 77 82Niger 112 99 97 78 80Rwanda 80 93 102 84 80Sierra Leone 95 108 100 84 81Somalia 99 109 101 78 n.a.Sudan 98 97 109 80 82Uganda 121 126 102 98 102Zaire 117 107 99 94 82

Countries without a noted famine during 1970s, 1980s, or 1990sBenin 97 99 95 119 136Botswana 134 150 102 68 72Burkina Faso 112 116 100 119 127Cameroon 91 111 101 78 86Central African Republic 90 95 102 94 92Congo 117 107 101 92 84Co^te d'Ivoire 74 88 100 93 106Gabon 99 114 96 82 76The Gambia 160 163 116 90 58Ghana 117 129 99 116 134Guinea 101 101 101 90 95Guinea-Bissau 112 110 95 103 90Kenya 112 110 95 103 90Lesotho 136 127 96 70 75Mali 95 100 106 96 90Namibia 114 127 93 98 67Nigeria 129 135 97 123 135Senegal 153 139 124 98 107Swaziland 69 95 107 85 78Tanzania 97 93 97 78 74Togo 119 117 97 95 90Zambia 101 105 97 96 81Zimbabwe 101 127 119 78 67

NOTES: 1980 index = 100; n.a. = not available (von Braun et al. 1998:32).


products of cyclical \changes in heat coming from the sun" (Okudi 1992:4)|droughtand famine (and consequently the prevention or alleviation thereof) are removed fromthe scope of agentive action.

A similar argument that has existed since colonial times is the tendency to placeblame on pastoral society. Mainly espoused by \governments, liberal scholars andchange agents," the idea is that these societies are `backwards' and conservative,emphasizing \stock accumulation as opposed to quality control, pastoral mobility andcattle raiding" (Okudi 1992:4). Thus, it is not uncommon to see Africa's endemictwentieth century famines directly and solely attributed to the victims.

In the case of Uganda, Akol (1985) argues that development of agricultural pro-ductivity in aected areas was impeded by a `persistent rustling' of livestock, leadingto distress migration, and Alnwick (1985) attributes the country's 1980 famine to the`alarming' amount of raids after the fall of Amin's government in 1979, leading to aloss of animals and cropland. Akol also blames land fragmentation, stemming fromcustomary Karimojong inheritance practices, as having limited agricultural produc-tivity: By dividing the land into small plots among families, it is argued, the use ofmodern farming technologies is made impossible. Other scholars attribute food short-ages in the area to the Karimojong culture of ritual feasts, for which large amountsof food and drink are prepared.

However, these arguments are misleading, based on generalities, and present pas-toralism simply as ignorance, with external factors such as the weather and marketconditions left out of the picture. \The assumption behind food decline is that thereare extra mouths to eat and not extra hands to produce" (Okudi 1992:5). In focus-ing on aggregate supply and demand of food, though, the exploitative nature of themarket is ignored. While Sen's argument for considering entitlements follows, Okudiargues that such an approach is as \narrow in scope as it is limited to the immedi-ate causes of famine and its consequences" (Okudi 1992:5), and fails to expose therelationship between long-and short-term causes of famine.

1.3.1 Identication

Beyond identifying the cause(s) of a famine, there is still the question of what `famineconditions' entail, what distinguishes famine from mere `episodes of food insecurity'or `chronic hunger,' and, of course, how best to avert such an aiction. While `pop'conceptions of famine do tend to oversimplify, they can provide a helpful startingpoint for exploring what constitutes this complex occurrence.

To some, famine is identied primarily as a health crisis, manifested by signicantchanges in nutrition levels, or starvation. In this vein, Dreze and Sen (1989) distin-guish famine|\involving acute starvation and a sharp increase in mortality"|fromchronic hunger|\sustained nutritional deprivation on a persistent basis" (Dreze &Sen 1989:7)|as distinct yet related phenomena. While the former, they note, requiresspeed in intervention, often resulting in the use of existing distributional mechanisms,the nature of the latter is such that slower but more impactful policies may be en-acted. This distinction is also important in considering the experiences of dierentcountries. While India, for example, experiences regular hunger and endemic under-


nutrition, the Bengal famine of 1943 was the result of acute starvation escalating intolarge-scale mortality. On the other hand, countries may become adept at dealing withpersistent hunger but fall prey to considerable transient hunger, as in China during1958-61. The distinction between these countries is especially notable, Dreze and Sennote, due to the seeming inability of either to overcome both endemic and transienthunger: While India's success in famine prevention seems not to have reduced thecountry's level of chronic hunger, China has achieved success in improving its citizens'level of nutrition in normal times, though this has not translated to an ability to avertfamine8.

However, de Waal (2000) argues that this distinction|between transient `famine'and `chronic hunger'|is a fallacious one. As with most social scientic terms, heargues, there is generally a discrepancy between popular perceptions of such complexphenomena and arbitrary scientic bounds. Like a ne wine or hard-core pornog-raphy9, \Everyone can diagnose a famine when they encounter one, but identifyingits crucial elements is rather more dicult" (de Waal 2000:5). In part, this stemsfrom the unpredictable relationship between various statistics (mortality, food supply,etc.) and impressions of the events at hand|while the Sahelian famine of the 1970soccurred without measured excess mortality, for example, the Bengal famine of 1943is notable for having occurred without any decline in food availability.

It is important, then, to at least attempt the establishment of an operational def-inition of famine, so as to allow, among other reasons, a distinction between chronicimpoverishment and acute famine|both of which still satisfy Sen's `democracy pre-vents famine' hypothesis. While it is easy to identify extreme cases of famine, thereare at least as many borderline cases, especially in Africa.

Many African countries, de Waal notes, \have lived for years in a sort of economictwilight, suering extreme dislocation and poverty, and occasionally attracting inter-national media and relief attention that points to starvation and famine" (de Waal2000:5). Indeed, there are communities|such as Southern Sudan|who would arguethat they have suered famine continuously for some time.

In an attempt to construct an operational denition of famine, de Waal (2000)identies four main components, with real-world instances combining some if not allto varying degrees:

Hunger Includes \subjective feelings of severe and prolonged hunger, the socially-dened going without acceptable food, and the measurable fact of undernutri-tion."

Impoverishment Includes \loss of livelihood, income and assets, and other compo-nents of increased poverty. For most rural people threatened with famine, themost concrete fears arising from famine are those associated with the threat ofdestitution."

Social breakdown Social symptoms of famine typically include \distress migration,

8Dreze and Sen explore this further in Hunger and Public Action Chapter 11.9I refer of course to the famous line from US Supreme Court Justice Potter Stewart (d. 1985)

describing his threshold test for obscenity in Jacobellis vs. Ohio (1964).


splitting up of families, etc. For many famine-aected communities, these arenot `symptoms' of famine, but are intrinsic elements of the unpleasant experi-ence itself."

Mortality \Many famines are accompanied by increased levels of mortality, usuallyconcentrated among vulnerable groups such as children, the aged, and migrants.The highest death rates are invariably found in camps and other concentrationsof destitute people. When social disruption is such that there is mass migrationto camps, death rates tend to shoot up" (de Waal 2000:6)

De Waal also mentions a fth component to famine, which are coping strategies ofindividuals, families, and larger groups in response to any of the above occurrences.However, it is also possible, he notes, for famine to occur without any number ofthe above factors (though hunger is likely a universal element). Though uncommon,famine striking asset-rich societies is not unheard of (for example, the Dutch HungerWinter of 194410), while others have occurred without social breakdown, or evenwithout excess mortality. The nature and extent of a famine, then, depends greatlyon social and economic responses to pleas made by aected populations.

Rather than simply conceptualizing hunger and famine as on the same linear scale,famines may assume qualitatively dierent forms as they escalate. However, whilethis does help clarify famine conceptually, it does not make the identication of suchan occurrence any less imprecise.

1.3.2 Quantication

A second vector along which famines may be distinguished is severity|the degree towhich each factor is present.

Though `famine' may occur without excess mortality, this statistic is nonethelessrelevant in understanding the severity of such a crisis. However, for most famines, lit-tle is known about the scale of deaths. Estimates are always approximate, and reectthe wishes of the publisher, whether it is a government or governmental organiza-tion seeking to hide or inate the crisis, or humanitarian actors attempting to drawsupport. For methodological reasons, Devereux (2000) explains, demographers andnutritionists prefer to release `crude mortality rates'|deaths per thousand|of par-ticular population subgroups, though the media and public tend to desire aggregatetotals which are not typically known with condence.

Problems in estimating crude mortality rates may arise for a number of reasons.First, in poor countries, population data may be unreliable or unavailable. As aresult, one method for estimating crude mortality rates involves scaling up mortalityrates of observed population subgroups (e.g. malnourished children in refugee camps)to the national population, and comparing against `normal' mortality rates.

However, the movement of large populations is itself associated with elevated mor-tality, \because of loss of access to normal food sources and exposure to hazardous

10Though notable for occurring in a more developed and asset-rich country, this famine occurredduring and was generated in no small part by the Nazi occupation.


environments|new disease vectors, lack of shelter, use of unsafe water" (Devereux2000:5), et cetera. Thus, refugee camps, paradoxically, tend to experience the high-est mortality rates. Malnourishment experienced on the journey to the camp, aswell as overcrowded and unsanitary conditions at many camps, make them breedinggrounds for communicable diseases. Ultimately, an estimation based on such a casewill likely be exaggerated. At the same time, though, mortality does not capture thefull impact of famine, which \often doubles when lost births due to fertility declineare considered" (Devereux 2000:5).

Instead, Devereux (2000) disaggregates mortality by age and sex, revealing pat-terns which reect interactions between \biological and sociocultural vulnerability."The latter, like the former, is context-specic, so mortality rates specic to age andgender, for example, are not easily generalizable geographically or over time. Socialhierarchies, cultural practices, and economic priorities, as well as age- and gender-specic biological vulnerability will also aect household-level mortality patterns11.

Considering that the nature of famine may change considerably as it `escalates,'de Waal identies three qualitative degrees of famine severity:

1. Famines involving primarily hunger and impoverishment;

2. Famines in which there are elevated rates of mortality;

3. Famines in which there are spectacularly high death rates alongside severe socialdislocation and collapse (de Waal 2000:7).

While most dictionary denitions of famine refer to types two and three, many experi-ential denitions, de Waal notes, include the rst type as well. Though these degreesof severity also include changes in exchange entitlement mappings (e.g., the abilityto command food), they also consider antecedent and subsequent occurrences. If, forexample, a famine escalates from a `type one' to a `type two' famine, it may be dueto a coping strategy that involves large-scale migration and increases the incidenceof disease, consequently raising mortality rates.

Additionally, responses to famine may change over time as people learn from pastexperiences. Thus, in Sudan, for example, the failure of relief shelters during 1984-5led many rural people to avoid migration to towns during the country's 1990-1 droughtand famine. In this case, death rates were likely lower as a result, though such a claimimplies a degree of reliability among mortality gures which seems unlikely.

After identifying component elements of famine as well as a qualitative scale ofseverity, de Waal proposes a (simplied) typology of famines, based on the sector ofsociety aected and the primary causal elements:

11That being said, generally, \women are more resilient than men because females store more bodyfat than males, and a statistical eect also favors female survival because famines are associated withfertility declines, so maternal mortality rates fall for the duration of the food crisis. On the otherhand, women often face higher `social vulnerability' than men, in cultural contexts where intra-household allocation rules for food, health care, and other basic needs favor males over females"(Devereux 2000:11). This pattern of mortality was observed in Karamoja, Uganda.


Pastoral These famine aect herders primarily. While their short-term cause isusually drought leading to a lack of pasture and water for animals, in the long-term, they generally entail an \alienation of pastures for farms and plantations,and restrictions on nomadic movements. These are slow-onset famines. Becauseof the mobility of pastoralists, these famines tend to cover wide areas, but oftenthey are invisible outside the pastoral areas. They may be extremely protracted.As well as relief distributions, eective responses can include buying up livestockat guaranteed prices and providing credit."

Agrarian/Smallholder \These are the paradigmatic African famines, aecting sc-attered farming populations. Commonly, drought-related production failuresare the proximate cause, with deeper causes including exclusion from land, ex-ploitative economic relationships, etc. These are also usually slow-onset famines.They can often be highly localized, and the more severe famines are often akinto a series of interconnected localized famines in which each locality is unableto assist its neighbor. Often there is an `epicenter,' from which waves of grainprice rises and distress migrations move out. Only severe agrarian famines be-come visible outside the aected areas. Coping strategies are usually muchmore important than relief programs. A wide range of programs and policies,ranging from land preservation to prepositioning food stocks, can help preventsuch famines, and a range of responses including food relief and labor-basedrelief projects can help ameliorate the eects."

Class-based/Occupational \In this category, wage laborers are often the worst hitoccupation. These are the paradigmatic Asian famines, in which whole class ofpeople (farm laborers, artisans, sherman) are suddenly rendered destitute bya collapse in the demand for their labor or a rise in the price of a staple food.Some recent African famines have become closer to this type. These faminestend to be rapid onset and cover a wide area, selectively aecting certain groups.They can be highly visible, with aected people ooding towns, and townspeoplethemselves suering. Political visibility and newsworthiness are certainly factorsin the extent of political concern about these famines. In these famines, copingstrategies are less eective, and state intervention is far more necessary. Grainprice controls and employment guarantee schemes are the most eective reliefmeasures."

Wartime \Wartime famines are usually associated with a catastrophic collapse ofthe livelihood base, either by physical destruction or conscation, or by severerestrictions on movement and economic activity. The nature of the faminedepends very much on the nature of the war and the determination with whichthe belligerent parties pursue their famine-creating strategies. These faminescan be very rapid, or can take years to develop (perhaps aected societies thatare not normally famine-prone); they can be highly localized, or can cover a hugearea. Occasionally, coping strategies can be forcibly prevented by belligerents,leaving aected people wholly reliant on relief. Sometimes these famines are


almost completely invisible|deliberately kept that way by the belligerents" (deWaal 2000:9).

Despite these categories, many famines are, of course, compounds of the above types.Agrarian famines, for example, are commonly associated with pastoral famines; andchanges in grain prices caused by agrarian or wartime famines may cause class-basedfamines in adjacent areas. In Ethiopia in 1983-5, for example, \some of the highestmortality was recorded in areas which did not themselves suer a major productionfailure, but which were suddenly (and to their residents, inexplicably) struck by highfood prices and immigration of destitute laborers" (de Waal 2000:10). Very generally,de Waal notes that `characteristic Asian famines' have been class-based, rapid-onset,and high visibility, in response to which coping strategies are seen to be relativelyineective, making state action more important. On the other hand, most Africanfamines \have been more locality based, slow-onset, low visibility, and with greaterroles for coping strategies and less for public action" (de Waal 2000:10), contributingto a lesser political signicance.

A slightly more rigorous approach to quantifying famine is taken by Howe &Devereux (2004), who propose a bipartite scale to estimating famine severity. Inaddition the `magnitude' of a famine|the aggregate impact of the crisis|the authorsestablish `intensity' as a separate metric|reecting the severity of the crisis at a givenlocation and time. While there do already exist approaches to quantifying famine,most are descriptive, ambiguous, or lack clarity necessary for use as an operationalguide. Additionally, while qualitative denitions of famine are complex, in that theyinevitably describe a causal relation between elements of society (e.g. whether famineis a `natural disaster,' `social process,' `community crisis,' et cetera), a denition basedon numbers should, in theory, be immune to such biases (though the origins of thenumbers themselves may be biased).

During the 1980s and 1990s, there were a number of studies undertaken that ex-amined the behavioral responses of populations aected by food insecurity. Patternsand sequences inferred from these observations|`coping strategies'|were identied,leading some to believe that these behaviors could serve as an indicator of worsen-ing famine conditions. In this light, Walker (1989) identies four distinct `stages' ofcoping strategies:

Stage (a) includes strategies for overcoming `normal seasonal stress,' such as ra-tioning of food or diversifying of income.

Stage (b) is the result of prolonged food stress, during which time increasingly irre-versible strategies are employed, such as selling breeding livestock or mortgag-ing land, though these solutions simply substitute short-term gain for long-termtroubles.

Stage (c) is \characterized by dependence on external support" (Howe & Devereux2004:358) such as food aid, and, if all else fails,

Stage (d) starvation and death, will follow (Walker 1989).


Between stages (a) and (b), Walker explains, \the essential dierence... is that[strategies] in the second stage directly undermine the basis of the victim's means ofsurvival... [T]hey sacrice future security for present survival. If one wished to markthe true beginnings of famine, as opposed to seasonal food shortage, this might be theappropriate place" (Walker 1989:50). One important insight from coping strategiesliterature, Howe & Devereux note, is that \famines threaten livelihoods as well aslives, and that eective famine prevention requires early intervention to protect liveli-hoods, rather than mandating famine relief just to `save lives' " (Howe & Devereux2004:359). However, coping strategies are highly context-dependent, underminingtheir ability to serve as generalizable famine indicators. Other authors illustrate thecomplexities of isolating `stages' from each other, with some authors distinguishing`coping' from `adaptive' strategies, and others argue that coping strategies are notnecessarily adopted sequentially, but concurrently. Ultimately, however, these do noteectively justify the presence and nature of coping strategies as a viable dimensionby which famines may be distinguished.

Whereas coping strategies are wholly context-specic, nutrition-related indicators,Howe and Devereux argue, may be compared universally. However, this metric alsoposes denitional problems, such as what specic rate of malnutrition or mortalitymay indicate the beginning of a famine. Another problem with nutritional indicatorsis that they generally refer to children under ve years of age, as there is no agreed-upon denition of moderate and severe malnutrition among older population groups.Yet, children over ve and adults are demonstrated to be more aected by emergenciesthan are younger children, as adults will often reduce their intake to ensure that theirchildren have enough. Thus, child malnutrition may serve as a `trailing indicator,'failing to manifest until well after adult malnutrition has set in.

A third complication of nutrition-related metrics is the ambiguous nature of therelationship between nutritional status and food crises. This relationship is madeproblematic because \malnutrition outcomes can be the result of other factors besidesinadequate food intake, such as disease, an unsanitary public health environment orpoor child-care practices" (Howe & Devereux 2004:359). In particular, the point israised that nutrition should not be taken as an indicator of famine without simul-taneously considering wider food-security information, as \low rates of malnutritionmay mask a developing famine process: severe erosion of livelihoods and exhaustionof coping strategies" (Howe & Devereux 2004:360).

To provide greater operational clarity and accountability for famine prevention,Howe and Devereux propose adopting intensity and magnitude scales|while the for-mer refers to the time- and location-specic severity of a crisis, `magnitude' refersto the aggregate impact of a crisis on the aected population. While their scale isinspired and inuenced by prior literature, it diers from previous indexes in threeways. First, the scale is intended to be generalizable, such that situations of foodinsecurity may be compared across the globe while at the same time recognizing thateach crisis is grounded in a specic context. Second, the scales \capture a greatercomplexity of famine impacts than other approaches and denitions" (Howe & Dev-ereux 2004:360) by disaggregating the intensity and magnitude; and third, the scalesoer a much more nuanced and operationally-useful diagnosis of food insecurity than


do binary scales of `famine' or `no famine.'

Intensity

To estimate a measure of `intensity,' the authors combine anthropometric and mortal-ity indicators, as well as `food-security descriptors.' The \anthropometric/mortalityindicators provide cut-os for each level that can be compared across situations. Thefood-security descriptors capture the dynamic, self-reinforcing changes in the liveli-hood system associated with increasing degrees of food insecurity and famine, and canbe adapted to specic circumstances (for example, drought or conict) and diversecontexts." (Howe & Devereux 2004:361)12. Additionally, the authors establish a sys-tem of weighting anthropometric/mortality versus food-security indicators dependingon which is estimated to occupy a more causal role.

One issue with this scale, however, which has yet to be resolved, is the appropriateunit of analysis for determining the intensity level. The authors oer no solution,except to note that the intensity (localized) and magnitude (aggregate) measuresare designed to be complementary. By creating an ordinal scale of localized food-insecurity, the situation may be observed over time, and allows external stakeholdersto make more informed decisions regarding aid.

Magnitude

A complete assessment of a crisis' full impact, Howe & Devereux argue, can only bemade in retrospect. Thus, `magnitude' refers to \the scale of human suering causedby the entire crisis, as proxied by excess mortality" (Howe & Devereux 2004:364).The authors also note that mortality of the `magnitude scale' starts at zero, ratherthan one, because, as has been established, malnutrition need not imply deaths. Thisis, the authors note, one resolution to a long-standing debate as to whether faminesare characterized by excess mortality by denition. On the one hand, quantifying theimpact of famine by excess mortality makes sense, as death is \the most tragic humanconsequence of famine" (Howe & Devereux 2004:364). Though on the other hand, toacknowledge that `famine' may occur or have occurred even without excess mortalitybetter characterizes famine as a complex processes (marked chiey by hunger anddestitution).

As mentioned earlier, the intensity and magnitude scales are designed to work intandem, and thus interact in specic ways: \Any intensity level of 3 or above willregister as a famine on the magnitude scale, even if it occurs in a very localized area,and even if no deaths are recorded (this could be a `Category A' famine)" (Howe &Devereux 2004:364). However, the opposite is (clearly) not necessarily true: everyincident which involves death is not necessarily a famine. Additionally, deaths mayresult from a food crisis condition, but the crisis may not be considered a `famine'unless the intensity of the conditions in any given area matches or exceeds `level 3.'

12\The anthropometric/mortality indicators and food-security descriptors can be thought of asregistering the eects of a crisis on the `lives' and `livelihoods,' respectively, of the aected popula-tion" (Howe & Devereux 2004:361).


Table 1.3: Famine Intensity Scale

Level Phrase Designation Malnutrition & MortalityIndicators

Food-security Descriptors

0 Food-security CMR


Table 1.4: Famine Magnitude Scale

Category Phrase designation Mortality range

A Minor famine 0-999B Moderate famine 1,000-9,999C Major famine 10,000-99,999D Great famine 100,000-999,999E Catastrophic famine 1,000,000 & over

(Howe & Devereux 2004:365)

accountability.' Proportionality, it is argued, is important to establishing account-ability both in terms of the number of deaths caused, as well as intent.

However, while attempting to determine intent can increase accountability andbring perpetrators to justice, the authors warn against the realization of perverse in-centives, \such that governments and humanitarian agencies devote disproportionateresources and energy to ensuring simply that threshold malnutrition and mortalityrates on the famine scales are not crossed" (Howe & Devereux 2004:367). Addition-ally, on the part of a perpetrator|if such a role exists|these perverse incentives mayresult in eorts to mitigate (or exacerbate) famine conditions to the extent that thecrisis crosses one threshold or another. Instead, the authors advocate formulatingpolicy solutions that address the underlying causes of famine, though the eort to doso may be prompted or catalyzed by the identication of more discrete levels.

Chapter 2

Theoretical Argument

2.1 Democracy and Elections

While there certainly is no dearth of scholarship regarding identifying the presence,scope, and causes of famine, we turn now to the relationship between famine andpolitics. While events which catalyze famine may be of a natural or meteorologicalcharacter, another line of thought identies famine as resulting from a lack of ac-tion: Any society, with concerted eort, can prevent drought or other natural causesfrom escalating into famine. What circumstances, then, would alter a government'sresponse, or provoke a government to respond?

From Malthus (1798) comes a `demographic explanation' of famine: it is theresult of food supply restricting unsustainable increases in the population. As hasbeen addressed elsewhere, by this reasoning, famine is a natural phenomenon whichcannot be averted. Though in retrospect we may call neglectful the government thatallows famine to `run its course' (e.g. Britain prior to 1880), the logic behind this issuch that famine is outside the scope of possible agentive action.

From Sen (1981), we see an economic explanation of famine: it is the failure ofsome people's entitlement relations to provide them access to enough food to sur-vive. By such a conception, we may say that famines are allowed to occur when agovernment is insuciently concerned with protecting the entitlements of its citizens(i.e., preventing a relative degradation or decline of people's exchange entitlementmapping). A policy solution, then, would take the form of entitlement protection,seeking to prevent changes in exchange conditions or in endowments.

Finally, from Dreze and Sen (1989) and Sen (1999) come political reformulations ofthe entitlement argument: with competitive elections comes an incentive to appeasevoters as politicians struggle for reelection, while a free press guarantees `informationaccess to the public,' helping politicians gauge public opinion and compelling govern-ments to act. By this logic, famine may occur if and when one or both of the aboveinstitutions are in some way decient.

De Waal (1990, 2000) responds to Sen, and though he believes in entitlementtheory, he identies entitlements as part of a larger picture. Famines, it is argued,are composed of ve elements in varying proportions: hunger, impoverishment, social

30 Chapter 2. Theoretical Argument

breakdown, mortality, and coping strategies in response to the rst four. Thus, whilefamine prevention eorts should include entitlement protection measures, the presenceand severity of each other factor should be considered as well.

Sutter (2011), on the other hand, explores `statehood' as it mediates the con-nection between political institutions and famine. To disaggregate state quality,Sutter measures two forms of legitimacy: `horizontal'|the ethnic fragmentation ofa country|and `vertical'|the \proximity of the state|as a structure of politicalpower|to the society" (Sutter 2011:3). According to this conception of politicalfamine response, institutions that promote (vertical) accountability, such as a freepress and competitive elections, are made distinct from the political will necessary toutilize those mechanisms eectively1.

Another explanation, known as selectorate theory, comes from Bueno de Mesquitaet al. (2002), who argue that governments possess a nite amount of response capac-ity: \both democracies and autocracies face a trade-o between the cost of action andthe cost of inaction. The government is assumed to maximize its political support tostay in power." As a result, famine mortality may occur if governments \nd thatinaction is the support-maximizing strategy" (Sutter 2011). Key to this decision isevaluating the size of the selectorate (S )|\the set of people who have an institutionalsay in choosing leaders"|relative to the winning coalition (W )|\the minimal set ofpeople whose support the incumbent needs in order to remain in power" (Mesquitaet al. 2002). When W is small, leaders may focus on the particularistic transferof goods and services, at the expense of the provision of public goods, while a largewinning coalition encourages bandwagoning. While by this logic, when W is small,famine is somewhat Malthusian, this theory does account for the possibility of con-certed action overcoming such a catastrophe.

In all cases, legitimacy accompanies political authority and trust, which formsthe \basis of a better ability of institutions to protect citizens" (Sutter 2011), andincreases the state's power to act. A second aspect of legitimacy is the \respect ofsocial contract, reecting how close the state is to society," and third, state legitimacy\decreases the risk of shocks leading to famine, especially the likelihood of civil wars:trust in institutions build peace." While Mesquita et al. do not discuss the legiti-macy of the state per se, the same responsiveness may be thought of as a successfulbroadening of the winning coalition|the set of people whose support the incumbentneeds to remain in power.

Sutter proposes a few metrics for measuring state legitimacy. First is an exam-ination of public opinion, and second is a behavioral approach, with \contestationbehaviors [meaning] a lower state legitimacy." In other words, people casting votesfor non-incumbent candidates during elections indicates that at least some peoplewould prefer an alternative regime to the one currently in power. Her third approachfocuses on the \change of legitimation discourses of politicians as a way to denelegitimacy" (Sutter 2011).

1Her examination of thirty-six countries over 1980-2005 reveals two ndings: as institutionalquality improves, the likelihood of famine decreases; and even accounting for climatological andpolitical shocks, a higher aggregate state legitimacy corresponds to a lower probability of famine.

2.1. Democracy and Elections 31

Polity IV

In an attempt to more rigorously examine each country's political situation over time,I have included the use of the Polity IV dataset in my analysis. Begun as an attempt to\[code] the authority characteristics of states in the world system," the dataset's levelof analysis is a country's `polity': a \political or governmental organization; a societyor institution with an organized government; state; body politic" (Marshall et al.2011:1). Among other reasons, the dataset is particularly useful for its placement of`democracy' and `autocracy' as collinear|captured as a country's `Polity Score'|aswell as its disaggregation of `executive recruitment.'

Table 2.1 contains Polity scores of relevant countries. For the scores of all Africancountries, see Appendix A. In each case, the score listed is the country's average from1980 to 1989, with the exceptions of Sudan, whose 1985 regime change warrants adistinction, and Uganda, whose score is only that under Obote II (1980-5).

The rst score (POLITY) is simply the dierence between a country's democracy(DEMOC) and autocracy (AUTOC) scores on a scale of +10 (full democracy) to -10(full autocracy). Democracy is conceived as

accountability without democracy: lessons from african famines of the 1980s

Documents

african famines

aected african countries

african cases

polity scores

polity iv scores

press freedom scores

xrreg scores

xrcomp scores