famine and economics

8/12/2019 Famine and Economics

http://slidepdf.com/reader/full/famine-and-economics 1/39

American Economic Association

Famines and EconomicsAuthor(s): Martin RavallionSource: Journal of Economic Literature, Vol. 35, No. 3 (Sep., 1997), pp. 1205-1242Published by: American Economic AssociationStable URL: http://www.jstor.org/stable/2729976 .Accessed: 10/12/2013 01:29

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to Journalof Economic Literature.

http://www.jstor.org

This content downloaded from 158 .121.247.60 on Tue, 10 Dec 20 13 01:29:19 AMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=aea

http://www.jstor.org/stable/2729976?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp





http://www.jstor.org/stable/2729976?origin=JSTOR-pdf

http://www.jstor.org/action/showPublisher?publisherCode=aea



Journal of Economic LiteratureVol. XXXV (September 1997), pp. 1205-1242

amines n d conomics

MARTIN RAVALLIONWorld Bank

For their helpful comments on this paper, the author thanks Harold Alderman, Jazmes Boyce,Gaurav Datt, Karla Hoff, Peter Lanjouw, Michael Lipton, Pat McGregor, Munir Quddus,Arnartya Sen, Dominique van de Walle, Etienne van de Walle, Nicolas van de Walle, DennisYang, and the Journal's referees. These are the views of the author, and should not be attrib-uted to the World Bank or its member countries.

I. Introduction

SOME DOUBTFUL STARTING pointsfor the analysis of famine can easily

get embedded in its definition. Commonusage allows two distinct definitions.One is that famine entails an extremeand general scarcity of food while theother defines it as widespread, unusuallylife-threatening, hunger. This paper fol-lows the latter definition; I will say that ageographic area experiences faminewhen unusually high mortality risk is as-sociated with an unusually severe threatto the food consumption of at least somepeople in the area. This definition doesnot require that there be a contraction inthe aggregate availability of food-oreven in its aggregate consumption-forfamine to have occurred. But life-threat-ening starvation is present (even thoughchecks and balances may eventuallycome into play to forestall mass starva-tion). Nor does this definition requirethat the set of people who face death isthe same as the set of people who experi-ence the threat to their food consump-tion; some may die due to diseases thatspread during famines.

By this definition, or something likeit, the twentieth century has seen fam-ines in most parts of the world. In Asia,

famine occurred in Bengal (then part ofBritish India) in 1943-44, and again in1974-75 (in what is now Bangladesh).China had a famine in 1959-61. In Af-rica, there have been famines in Ethio-pia, Sudan, Mozambique, Nigeria, Ni-ger, Angola, Zaire, Uganda, Somalia,and Liberia, and a number of these hap-pened since 1980. The former Soviet

Union had three famines this century.In western Europe, Holland had a fam-ine in 1944-45. A great many peopledied in these famines, though we willprobably never know with much accu-racy how many.' And the lost lives went

1 For example, estimates for the China faminerange from 15 to 30 million people (this faminewas clearly the worst this century). (The upper es-timate is that of Basil Ashton et al. 1984; the lowerone is based on official mortality rates; see CarlRiskin 1990). The variance of estimates is as greatfor the Soviet Union's famines; it is estimated thatbetween 5 and 9 million people died during the1921-22 famine, and 5-11 million died in the1932-33 famine, and 2-5 million in the 1946-47famine (William Dando 1981). (Dana Dalrymple,1964, quotes an even wider range of mortality esti-mates lor the 1932-33 famine, namely from 1-10million.) The Ethiopian famine of 1984-85 wasclearly the world's worst since the 1970s, withabout 8 million people deemed to have been af-fected directly of whom 1 million are estimated tohave died (Helmut Kloos and Bernt Lindtj0rn1993). But nobody seems to have much confidencein mortality estimates for this famine either. On

the difficulties in estimating famine mortality seeJohn Caldwell and Pat Caldwell (1992).1205







1206 Journal of Economic Literature, Vol. XXXV (September 1997)

hand in hand with lasting miseries ofmaterial and other deprivations forthose lucky enough to survive. Yet

surely (it has been said before) famineshave been more avoidable this centurythan ever before.

Famines continue to raise deep ques-tions about the performance of eco-nomic and political institutions. Didthose institutions help protect peoplefrom starvation, or did they make mat-ters worse? Economists have sometimestried to answer such questions and toinfluence public policies towards fam-ines. For example, the writings of AdamSmith, Thomas Malthus, and John Stu-art Mill were used to support a laissez-faire policy with respect to food mar-kets during the many famines in theBritish Empire during the nineteenthcentury.2

From about 1980, a new literature onfamines emerged, also premised on aview that the tools of economic analysiscan throw light on why famines keephappening, and what can be done toprevent or relieve them. The new litera-ture has revisited many of the classicnineteenth century debates about fam-ines, such as the extent and nature ofappropriate governmental interventionsin markets. Substantial progress hasbeen made in developing a richer con-ceptual framework for understandingfamines. Progress has also been made inadvancing empirical knowledge. Someof the arguments advanced in past de-bates have been rigorously formulatedand tested for the first time, thoughmany remain untested, either becauseof lack of data or lack of effort. The em-pirical study of famines has posed anumber of challenges. Traditional types

of data and other forms of "fair-weatherresearch" (as Caldwell, P. H. Reddy,and Caldwell 1986, p. 696, nicely put it)

may be quite uninformative about theseevents. For example, sample surveysduring famines are rare, and aggregatedata can be quite unreliable at thesetimes. Studies of famines have relied ona wider range of types of data than nor-mally found in applied economics; forexample, accounts from direct ob-servers, such as found in local newspa-per reports, have been an importantsource of data, when used carefully.

This article offers an overview of thisnew literature on famines, and what is-sues endure. In addition to looking atwhat economic analysis can teach usabout famines, the article tries to saysomething about what economists canlearn from famines, including from thelarge body of work on this topic by non-economists. It is argued that the newliterature suggests that famines canhelp economists and policy makers un-derstand the tragic extremes to whichotherwise adequate political and eco-nomic institutions can be driven whenexposed to certain shocks. To under-stand famines one must understand hownormal institutions work under stressesthey do not normally confront. The arti-cle first examines famines from a microperspective, emphasizing the multipledeterminants of starvation and thelikely nonlinearities. This is the topic ofSection II. The article then looks athow the various markets and institu-tions which coordinate individualchoices performed during famines. Sec-tion III takes up these issues. It arguesthat famines arise from severe aperiodicmarket and institutional failures ineconomies under stress. Arrangementscollapse which had previously workedadequately. Understanding why that can

happen helps us understand famines,and to understand the functioning of

2 See, for example, Cecil Woodham Smith(1962) on the famine in Ireland in 1846-47, and S.Ambirajan (1978), Salim Rashid (1980) and Jean

Dreze (1990) on the numerous famines in BritishIndia during the nineteenth century.







Ravallion: Famines and Economics 1207

those institutions. Governments areamong the institutions that often failduring famines, though recent litera-

ture also reports some real successes inintervention. Section IV tries to drawout some lessons for policy. Conclusionscan be found in Section V.

II. Microfoundations

Attempts to understand the causes offamines, and what to do about them,have traditionally focused on a rathersmall set of economy-wide parameters,

notably aggregate food availability andthe rate of population growth.3 Themain distinguishing feature of the neweconomic literature on famines is itsemphasis on understanding the circum-stances of individuals in famine-vulner-able settings, and how those circum-stances interact with economy-widevariables.

1. The Entitlements Approach and its

CriticsFamine is often blamed on some ag-

gregate exogenous shock. This simplecausal model can be questioned from anumber of points of view. A shock ofsome sort can always be identified; afamine does not seem to have ever hap-pened by "spontaneous combustion."The most common shocks are spells ofunusually bad weather and wars. It is

not always obvious that the "shock"should be considered exogenous to thepopulation in which the famine occurs;for example, a war is the outcome ofcertain peoples' choices for which fam-ine can be a predictable or even delib-

erate outcome. But even similar exoge-nous shocks can produce quite dissimi-lar outcomes, depending on initial in-

equalities in physical and human assets,the way the economy works given thoseinequalities, and the policies pursued inresponse to the shock. Some people suf-fer badly, while others may even gain.Yes, a shock of some sort can invariablybe identified at the start of the chain ofevents leading to a specific famine, butto properly understand-and prevent-the famine one must understand thatchain of events.

A better approach is to work from theperspective of those who suffer. That isthe enduring lesson from the first andmost influential contribution in the newliterature on famines, namely Sen's(1981) book, Poverty and Famines. Thisbook was very much an economic analy-sis of famines, though it succeeded inattracting the interest of both econo-mists and non-economists. Many econo-mists were introduced to an importantbut somewhat neglected set of eco-nomic issues, and it offered many non-economists an insightful new perspec-tive on those issues.

The central concept in Sen's ap-proach is an individual's "entitlementset," defined as all the commodity bun-dles that can be obtained from all theresources at the individual's commandin a given society, subject to the laws ofthat society. Starvation is then seen toarise from an "entitlement failure,"meaning that an individual's entitle-ment set no longer includes enoughfood to stay alive. The failure can takeany one of a number of forms, reflect-ing the fact that "people establish com-mand over food in many different ways"(Sen 1990, p. 34). The entitlement fail-ure could be due to a loss of endow-ments, or to a change in one or more of

the various ways-through production,trade, or transfers-in which endow-

3 This article will not address the demography offamines, except to note that the Malthusian viewthat famines act as a check on population growthhas been discredited (Susan Watkins and JaneMenken 1985; Caldwell and Caldwell 1992;Robert Fogel 1992). For an overview of the de-

bates on demographic causes and consequences offamine see Siddiq Osmani (1996).








ments are transformed into entitle-ments. If a sufficiently large number ofindividuals experience an entitlement

failure then a famine occurs.One possible starting point for achain of events leading to entitlementfailure is a crop failure, such as due to adrought or flood. It is certainly truethat some famines have been associatedwith a sharp drop in domestic food out-put.4 However, Sen (1981) rejected thisas a universal explanation. A number ofempirical studies have confirmed hisconclusion that famines have quiteoften happened without a decline incurrent aggregate food availability.5 Insome cases (including some severe fam-ines) there was a food-availability de-cline (FAD) but it was relatively mod-est-more like a 10 percent drop incurrent food output than anything onecould reasonably call a "crop failure."6In still other cases there was a smallFAD, but after the rise in mortality.7The link from aggregate food availabil-ity to starvation involves numerous eco-nomic and political factors. FAD need

not cause mass entitlement failure; in-deed some severe crop failures in poorcountries have not resulted in famines.8

And there are other possible startingpoints, including wars which disrupt theproduction or flow of food and othercommodities within an economy, or aspeculative crisis in food markets trig-gered by rumors of impending short-ages.

Sen's approach to famines carried amessage that economics has a lot to of-fer both in understanding famines andin preventing them. That message wasnot lost on many reviewers, who clearlysaw the attraction of encompassing thenumerous ad hoc "single-cause" expla-nations for famines within a coherentframework; see, for example, the re-views by Kenneth Arrow (1982),Frances Stewart (1982), Shlomo Reut-linger (1984), and Robert Solow (1991,reviewing Dreze and Sen 1989). Butothers were unconvinced. Indeed, anoften vociferous debate was instigatedby Sen's writings on famines.9 Somecritics argued that Sen attaches toomuch importance to food, and too littleto other factors such as disease. For ex-ample, drawing on field work in Dafur,Sudan, Alexander de Waal (1989) ar-gues that it was not starvation but anunsafe health environment while mi-grating that caused death. Others ar-gued that Sen undervalues the impor-

tance of aggregate food availability,and

4 See, for example, Sen (1981) on the famine inthe Sahel in 1972-74, Michelle McAlpin (1983) onthe famines in Western India in the late nine-teenth century, Peter Nolan (1993) on the Sovietfamines in 1931-33, and the Chinese famine of1959-61, and Justin Lin and Dennis Yang (1995)on the latter famine.

5 Among others, see Fogel (1992) on famines inEurope between 1500 and 1800, Ajit Ghose (1982)on the famines in nineteenth century India, Paul

Greenough (1992)on the Great Bengal Famine,

Mohiuddin Alamgir (1980) on the Bangladeshfamine of 1974-75; and Christopher Locke andFredoun Ahmadi-Esfahani (1993) on the famine inSudan in the mid-1980s.

6 Examples include the Great Bengal Famine of1943 in which food output was only 5 percent be-low its average of the preceding five years (Sen1981, p. 58), and famines in Ethiopia and Sudanduring 1983-84 in which food output was 11 per-cent and 13 percent (respectively) below its levelin 1979-81 (Dreze 1990).

7 For example, the harvest after the worstmonths of the 1974-75 Bangladesh famine wasabout five percent lower than the pre-famine har-vest, which was above average (based on JamesBoyce 1987).

8 For example, Dreze (1990) points out that, atthe same time as the severe famines in Ethiopiaand Sudan in 1983-84, far larger food output de-clines (around a 40 percent loss of output) hadoccurred in Cape Verde and Zimbabwe yet therewas no famine in those countries-indeed, mortal-ity declined. Also see Sen's (1981, ch. 8) discus-sion of the diverse outcomes in the Sahel in 1972-74, despite common shocks.

9 An excellent overview of this debate can befound in Osmani (1995). Also see Meghnad De-sai's (1988) and Stephen Devereux's (1993b, ch. 6)discussions of the various critiques of the entitle-ments approach.








that in doing so he risks misinformingfamine policy and thus worsening thesituation (see, for example, Peter Bow-brick 1986; and Devereux 1988).10 Andsome critics have questioned whetheronly people facing current entitlementfailure will go hungry, pointing to evi-dence that poor people with ample enti-tlements may prefer to go hungry atcertain times rather than sell their as-sets.11 While agreeing with the basicmessage, other critics have said there isnothing new in the "entitlements" ap-proach, claiming instead that it is a

long-standing explanation of faminesdressed in new garb (T. N. Srinivasan1983; Amrita Rangasami 1985; EdwardClay 1991).

In retrospect, I do not think onecould reasonably say that all of this de-bate has been insightful or interesting.Some has been based on misunder-standings of Sen's approach. A commonmisunderstanding is the claim that Senproposes entitlement failure as a non-nested alternative explanation to FAD;for example, Devereux (1988, 1993b)interprets FAD as a "supply side" expla-nation of famine, while Sen's is a "de-mand side" explanation. Such an inter-pretation is not well founded in Sen(1981) where he makes clear that theentitlements approach should be seenas an encompassing framework, withinwhich food availability is only one pa-rameter.

Some critics have seen the entitle-

ments approach as too "static," pointingto anthropological and other (anecdotalbut credible) evidence that avoidingcurrent hunger may not be the mainmotive for coping efforts (Jodha 1975;Corbett 1988; de Waal 1989). The enti-tlements approach can, however, bereadily be extended to accommodate in-ter-temporal choice by recognizing thatpeople may choose a degree of hungernow in order to avoid starving in the fu-ture. 12 In an inter-temporal consump-tion model with borrowing constraintsand random income fluctuations, a

"stock-out" will eventually occur suchthat all remaining assets are consumed,at which point the household willclearly be highly vulnerable to a bad in-come draw (Angus Deaton 1989; Alder-man 1996). The threat of a stock-outcan explain the observation that somefamine-vulnerable households initiallyforgo consumption rather than depleteassets. As will be argued at a number ofpoints in this paper, there are ways inwhich richer dynamic models can helpunderstand famines. But these modelsare perfectly consistent with the entitle-ments approach.

Some of the criticisms of the entitle-ments approach have also been tangen-tial at best to the main point. Forexample, the fact that there wereantecedents of the entitlements ap-proach, particularly in the literature onfamines in nineteenth century India,does not appear to be at issue (Sen1990; Dreze 1990). The nineteenth cen-tury literature on famines in India wasstill, however, firmly anchored to theview that crop failure due to drought orflooding was the ultimate cause of fam-ines, though recognizing that variousfactors intervened in determining thehousehold-level impact, including ac-

10 Problems of measuring food availability, andconflicting estimates, have fueled some critics,though at least one appears to be driven by faithalone; Peter Cutler (1993, pp. 72-73) asserts that"Food availability decline is always an element offamine, . . . even if [it] is not measurable at na-tional or regional level."

11While such behavior has often been noted(see Narpat Jodha 1975; and Jane Corbett 1988,among others), its interpretation as somethingdeemed to be inconsistent with the entitlements

approach appears to be due to de Waal (1989).Also see Devereux (1993a).

12This point was anticipated by Sen (1981, p.

50) and answers de Waal's (1989) and Devereux's(1993a) criticisms of Sen's approach.








cess to other employment opportunities(Osmani 1995). The term "entitlement"has also generated some confusion.

The term carries a normative connota-tion, although the intended meaning inthis context is entirely positive; Sen(1981, p. 2) was careful to point thisout, and specifically to distinguish hisusage of the word from the normativemeaning found in, for example, RobertNozick's (1974) "entitlement" theory ofjustice.

Some of the issues that have beenraised in the debates over the entitle-ments approach are more substantive.One of these is the fact that what con-stitutes an "entitlement" is conditionalon a specific legal system, as this de-fines property rights and (hence) per-sonal endowments. If famines entail thecollapse of law and order, it will bemeaningless to search for an explana-tion in terms of legal entitlements. Thisproblem too was recognized by Sen(1981, p. 49), as well as commentatorssince (Arjun Appadurai 1984; CharlesGore 1993). The extent to which it un-dermines the economic analysis of fam-ines is an empirical question. Here oneshould be careful to distinguish a col-lapse of quasi-cooperative, "informal,"arrangements for risk-sharing and assis-tance for the poor from a collapse ofthe legal apparatus which defines andenforces property rights. Endowments

remain well defined in the former case,but not the latter; and while there isevidence (reviewed later) that the for-mer often collapses during famines, itdoes not appear to be the case thatthe latter commonly does.13 Starvationand its avoidance are for the most partlegal.

Another substantive issue is the rolehealth plays. Sen's (1981, Appendix A)

characterization of entitlement failureas a cause of starvation assumes the ex-istence of a consumption "floor," above

which one lives, but below which onedies.'4 Failure to reach some well de-fined food consumption level is clearlya rather simple view of what determinesmortality during a famine. Food is notthe only thing that matters. For exam-ple, it has often been observed that animportant proximate cause of death infamines is exposure to disease associ-ated with a poor health environment,especially among those who migrateduring famines (Sen 1981, Appendix D;Kurt Jansson, Michael Harris, and An-gela Penrose 1987; de Waal 1989; TimDyson 1991; John Seaman 1993; HelenYoung and Susanne Jaspers 1995).There is clearly latent inter-personalvariability and, hence, uncertaintyabout survival prospects.

Some critics of the entitlements ap-proach have downplayed command overfood as a factor in explaining faminemortality, arguing that the health envi-ronment is far more important (see, forexample, de Waal 1989). These are not,however, independent causes. Thehealth environment is determined inpart by the same variables determiningconsumptions; for example, the expo-sure to disease of migrating people dur-ing a famine is not exogenous, but (itcan be argued) an outcome of the same

entitlement failures which led to migra-tion in search of food. And while starva-tion is not often identified as the proxi-

13 Though there are clear exceptions, such asthe famine in Somalia in the early 1990s.

14 For convenience, I shall treat this consump-tion floor as a well-defined quantity of a homoge-neous good. More generally it will be a vector. SeeSen (1981, Appendix A) for a general statement.Also see the recent models of competitive equilib-rium which have included sufficient conditions forhuman survival in equilibrium (Jeffrey Coles andPeter Hammond 1995; McGregor 1996); thesemodels have also assumed that there exists a well-defined lower bound on individual consumptionwhich is necessary and sufficient for individualsurvival.








mate cause of death during famines, itdoes appear to have a strong potentiat-ing effect on the incidence and severityof infectious diseases (Nevin Scrimshaw1990; C. E. Taylor 1985; John Post1990; D. L. Pelletier et al. 1995).15 Ithas also become clear that in under-standing the synergies among undernu-trition and susceptibility to infectionone should consider more that the bio-logical relationship (as might be identi-fied in a controlled experiment in whichone measures the infection rate as foodis progressively withdrawn from a single

person). Behavior can generate a strongsynergy even if biology does not. Oneclearly wants a theory which allows forother factors influencing famine mortal-ity, both directly and via their impacton the effect of changes in consump-tion. However, that need not mean thatcommand over food is any less impor-tant than one would have otherwisethought.

What issues endure, more than 15

years after the publication of Povertyand Famines? We can surely agree thatFAD has limited power to explain fam-ines, which have happened with andwithout FAD. We can agree that theproximate causes of famine have muchmore to do with entitlements and(hence) economics. But that still leavesmany questions begging. Is starvationonly a matter of entitlement failure?What determines vulnerability to fam-ines? Can similar shocks yield very dif-ferent outcomes? What causes the enti-tlement failure? Why is it covariate over

so many people? Do markets and (gov-ernmental and non-governmental) insti-tutions help or hinder the way aggre-gate shocks impact on entitlements?This article will review what the newliterature on famines (within and out-side economics) has had to say aboutthese questions. The rest of this sectionwill look at theory and evidence on thelink between entitlements and mortal-ity; later sections will take up issuesconcerning the causes of mass entitle-ment failure, and the implications forpolicy.

2. Micro-level Determinants of Mortality

Borrowing from another Sen concept,starvation is fundamentally a capability-

failure, rather than a lack of commandover commodities per se.16 The abilityto avoid starvation depends in part oncurrent consumptions, but also on thehealth-relevant aspects of the individ-ual's environment, and various idiosyn-

cratic attributes of the individual whichmay depend on past health and con-sumptions. Building on work in relatedfields, including health economics andnutrition science, recent literature onfamines has begun to investigate theselinks, though a number of issues remainpoorly resolved.

There have been a few recent theo-retical studies of aspects of economicbehavior in settings in which survivalprospects are endogenous (Mark Ger-sovitz 1983; Ravallion 1987; GerhardGlomm and Michael Palumbo 1993;Ludovico Carraro 1996). Followingstandard practice in much micro-economic analysis of the determinantsof health, one can postulate a "healthproduction function" relating individualhealth attainments to individual con-

15 The same comment applies to some othercauses of death during a famine. For example, sui-cide may be the immediate cause of death, but thehunger of that person or someone close to her isnot far behind the scene; drawing on field work inBangladesh during the 1979 drought, BangladeshRural Advancement Committee (1979, p. 11)writes that: "One woman in Rowmari became sim-

ly unable to stand the cries of her hungry chil-aren and, leaving them uncared, hanged herself."

16 On the meaning of "capabilities" and their re-

lationship to other conceptualizations of well-be-ing see Sen (1985).








sumptions, personal attributes (inter-pretable as "personal constitution"), andcharacteristics of the individual's imme-

diate health environment. 17 Assumealso that there is a minimum level ofhealth needed to survive. The prob-ability that a person will survive in anyperiod is then the probability that his orher personal constitution assures thathealth is above this floor, given con-sumptions and the health environment.The mapping from the consumptionspace to survival probabilities can betermed the "survival function." This is

not a purely biological function, be-cause it depends on behavior (includingthe impact of individual consumptionchoices on health), and the socio-eco-nomic environment (as it determinesthe health environment facing the indi-vidual).

If the health production function isquasi-concave in consumption and per-sonal constitution, and the probabilitydensity function of personal constitu-tion is unimodal, then it can be shownthat beyond a critical value needed toassure at least a non-negligible chanceof living, extra consumption will have adecreasing marginal impact on theprobability of survival i.e., the survivalchance will be concave from above inconsumption (Ravallion 1987, ch. 2).There will still be a "live or die" pointin consumption space, if only because acertain amount of food energy is essen-tial to maintain bodily functions at rest.However, in discussing famine causa-tion and relief policy the more relevantsubset of the consumption space iswhere the marginal effect of a changein consumption on survival chance willtend to increase as consumption falls.This has a number of implications both

for understanding famines, and for poli-cies to avoid or relieve famines. Beforeexamining those implications, let us

first look at the empirical evidence onthe determinants of famine mortality.There can be little hope of rigorously

testing the relationships econometri-cally on suitable micro data collectedunder famine conditions. However, twosources of data still offer hope of throw-ing new light on these issues. The firstis micro data for famine-vulnerable orsimilar settings under normal condi-tions. There is inequality even in poor

economies; so there is scope for usingcross-sectional variances and covari-ances at normal times to infer thingsabout the determinants of mortality atabnormal times. There remains, how-ever, the real concern that the underly-ing relationships may well be so sharplynonlinear that observations from normaltimes may be deceptive. I will return tothis point.

Supportive evidence from micro dataof a nonlinear relationship betweenmortality and consumptions that can befound in studies by nutritionists andothers indicating that mortality riskrises sharply at low nutritional status asmeasured by anthropometric indicators(Lincoln Chen, A. K. M. Chowdhury,and S. L. Huffman 1980; Peter Hey-wood 1982; Seaman and John Rivers1988; Partha Dasgupta 1993, ch. 14;Pelletier et al. 1995; Young and Jaspers1995).

There have also been a number ofmicroeconometric investigations of thedeterminants of nutritional status,though invariably based on data pro-duced under non-famine conditions. Animportant strand of this micro-litera-ture in development economics hasaimed to quantify the effects of incomechanges on nutritional status. There has

beena debate on the

magnitudeof the

income effect on nutrition in poor

17For an exposition of this approach see JereBehrman and Anil Deolalikar (1988). For an exam-

ple, see Mark Pitt, Rosenzweig, and Hassan(1990).








countries.18 In normal times, the aver-age income elasticity of food energyintake may well be quite low even inpoor rural settings; a number of studieshave found elasticities around 0.05-0.1(see the survey by Howarth Bouis andLawrence Haddad 1992). However,this could easily rise to 0.2-0.4 amongpoor people. If one also factors in thenonlinearity between nutritional statusand food-energy intake, one can findquite strong income effects on nutri-tional status among the very poor;elsewhere I have argued that the in-come elasticity of the incidence ofsevere under-nutrition in a poor econ-omy is around unity, even though indi-vidual food-energy intake at the meanhas a low income elasticity (Ravallion1990a).

The reverse causation has also re-ceived attention. Here too a nonlinear-ity is plausible; conservation of energyin a steady state makes it impossible tosupport productive work unless con-

sumption is sufficient to cover thebody's food-energy needs at rest (Das-gupta 1993, ch. 14), but at higher levelsof consumption further gains are un-likely to help raise productivity; this isthe nutrition-productivity relationshipbuilt into the well-known "efficiencywage hypothesis."'19 There is now sup-portive evidence from a number of mi-cro studies that nutritional status canalso affect consumption via its impacts

on productivity (see the survey byStrauss and Thomas 1995). Again these

studies have been at normal times. Butif anything one would expect the effectsto be even stronger during famines.Later I will discuss possible implica-tions for understanding famines.

Another lesson from micro datacollected during non-famine times isthat the welfare impacts of price andexogenous income changes can behighly diverse, even among poor peo-ple. Clearly behavior is geared in partto protecting living standards,20 thoughsome people will obviously be betterable to do so than others. Householdswith chronically poor endowments, orwhose endowments have been run downby a series of shocks, will be particu-larly vulnerable. Net trading position infood markets also varies among thepoor. Peasants with enough land to benet proOducers of food will gain fromhigher food prices, but other peasantsand landless laborers will probably lose,though even among the latter group,some will be protected by longer-term

contracts. All this helps explain anotherempirical observation about famines:the victims often come disproportion-ately from certain strata of the poor,such as artisans and casual agriculturallaborers (Alamgir 1980; Sen 1981; De-sai 1989; Abdur Razzaque et al. 1990;Patricia Bidinger et al. 1991; Foster1995).

The second source of relevant infor-mation is more aggregate, and possibly

less reliable, data collected during fam-ines. Evidence on the importance ofprices and incomes to current consump-tions during famines has come from di-8 See Jere Behrman and B. Deolalikar (1987),

Ravallion (1990a), John Strauss and DuncanThomas (1995), Shankar Subramanian and Deaton(1996), and Aman Ullah and Nilanjana Roy (forth-coming). For a recent overview of this debate seeMichael Lipton and Ravallion (1995, Section 4.4).

19This has implications for (inter alia) the wayslabor markets work in poor economies (JamesMirrlees 1975; Joseph Stiglitz 1976; ChristopherBliss and Nicholas Stern 1978; Dasgupta and De-braj Ray 1986; Dasgupta 1993). Section III willdiscuss this topic further in the context of famines.

20 There is a large literature on consumptionsmoothing and risk-sharing in developing econo-mies; recent contributions include Deaton (1989,1992), Mark Rosenzweig and Hans Binswanger(1993), Rosenzweig and Kenneth Wolpin (1993),Christina Paxson (1993), and Robert Townsend(1994). Also see Andrew Foster (1995) on the ef-fects of borrowing constraints on child nutritionalstatus.








verse sources. Both (anecdotal) qualita-tive and (more limited) quantitativestudies of specific famines have identi-

fied a large increase in the price of foodas an important proximate cause of foodentitlement collapse. Discussions offamines in South Asia have emphasizedthe role of foodgrain prices (B. M. Bha-tia 1967; Alamgir 1980; Sen 1981;Greenough 1982; Ravallion 1987;Sugata Bose 1990; Dyson 1991). Simi-larly, analyses of the recent famines inthe pastoral economies of rural Africa(including Ethiopia, Sudan, and the

countries of the Sahel) have identifiedthe prices of food staples (particularlyrelative to livestock prices) as key vari-ables influencing mortality (Sen 1981;Dirck Stryker 1993; Paul Ulrich 1993;Mahmud Khan 1994; Joachim vonBraun, T. Teklu, and Patrick Webb1994). And the price of food staples hasbeen cited as one of the factors influ-encing the fluctuations in mortality ob-served in Europe during early modern

times (Edward Wrigley and RogerSchofield 1981; P. R. Galloway 1985;John Walter and Schofield 1989; Post1990).

While there has been no shortage ofevidence confirming that entitlementfailures have underlain specific famines,the evidence has been less informativeabout some important aspects of theprecise ways in which entitlementchanges impact on mortality. Key hereis the nature and extent of the non-linearities in this relationship.

The nonlinearity with regard to theprice of food staples is an important is-sue for both understanding famine cau-sation and for relief policy. The argu-ments and evidence reviewed abovesuggest that survival chances will beconcave in consumption. Micro data onthe characteristics of the poor in fam-ine-vulnerable settings also suggeststhat vulnerable groups rely heavily on

current foodgrain markets.2' These con-ditions do not, however, imply that sur-vival chance will be concave in food

price, because food demand functionsare likely to be convex in price. In dis-cussing this question it is useful to workwith the survival function defined onprices and incomes (by solving outquantities consumed), although for thepurposes of the argument one need notassume that individual consumptionchoices are survival maximizing.

Small price increases may entail largeincreases in mortality among subgroups

of the poor if survival chances are in-creasing and sufficiently concave func-tions of income (Ravallion 1987). And ifthese conditions hold then price vari-ability over time will increase expectedfamine mortality. The critical value ofthe elasticity of slope will depend onproperties of the food demand function;high price and income elasticities of de-mand, and low budget shares for thestored good, tend to raise the critical

level of concavity needed to benefitfrom price stabilization. For example, ifincome is fixed and all of it is spent onfood then survival chance will be con-cave in the price of food if and only ifthe elasticity of slope with respect toconsumption exceeds two.22

This is an empirical question. Econo-metric investigations of the relationshipbetween mortality and foodgrain prices

21 For a survey of evidence on the poverty pro-file in low-income countries see Lipton and Raval-lion (1995).

22 To see why, let s(x) be the probability of sur-viving the current period when consuming anamount x of food. When all income is spent onfood, x = m/p where m is money income and p isthe price of food. Because by assumption) moneyincome is unaffected by a change n the price offood, on differentiating wice with respect to pone finds that s(x) is concave n p if and only f-x.s"(x)/s'(x) > 2. For a more general formulation,relaxing ome of these assumptions, ee Ravallion(1987, ch. 2). Also see David Newbery and Stiglitz

(1981) for a general discussion of the welfare ef-fects of price stabilization.








over periods encompassing two severefamines in South Asia (South India in1877 and Bangladesh 1974-75) indicatethat high and unstable prices contrib-uted substantially to the excess mortal-ity; on regressing the death rate againsta suitable nonlinear function of theprice of the main food staple, mortalityin both famines was found to be an in-creasing concave function of price(Ravallion 1987, ch. 2).

Arguably the single most importantrelative price determining the prospectsof starvation in most famine-vulnerable

economies is the foodgrain purchasingpower of the wage rate for unskilled la-bor, which is more often than not thewage for casual agricultural labor. Fig-ure 1 plots the number of deaths bymonth in one rural area during a periodencompassing the Bangladesh famine of1974-75 against the rice purchasingpower of the agricultural wage rate inthat area. Counts of deaths during afamine are notoriously unreliable, but

the series used in this Figure is likely tobe quite accurate.23 There is a strongnegative relationship, and an indicationthat the absolute marginal impact ofhigher real wages tends to fall as wagesrise. The fitted line in Figure 1 is theregression of the log of the number ofdeaths against the log of the real wagerate; the regression coefficient (inter-pretable as an elasticity) is -0.55 whichis significantly different from zero at

the one percent level (the standard er-ror is 0.09).24 Notice also that thestrong overall relationship in Figure 1 ismuch weaker in normal, non-famine,times; on excluding the famine period,the regression coefficient drops to-0.29, and is not significantly differentfrom zero at the five percent level,though it is at ten percent level (thestandard error is 0.16). A nonlinearity isagain evident in mortality responses toshifts in entitlements.

Other variables mentioned in thefamines literature as factors influencing

mortality include the health environ-ment and access to health care. Thoughplausible, there is little hard evidenceon the impacts of these factors underfamine conditions. Again micro datafrom poor economies during normaltimes has thrown some light on the im-pacts of these factors, though there aresome serious concerns about endo-geneity in regressing (say) child mortal-ity on (inter alia) access to health

care.25 Aggregate cross-country data dosuggest that differences in public healthspending matter more to mortalityamong the poor than non-poor (BenuBidani and Ravallion 1997). It can alsobe hypothesized that health-relatedvariables enter non-additively in thatsickness not only lowers survival chanceat a given consumption level but mayalso add to the adverse marginal impacton mortality of a drop in food intake (as

argued by Young and Jaspers 1995).Plausibly, some of these other variableswill also have lagged impacts, and somortality will be a dynamic variablewhich depends on past circumstances.So the long-run response of mortality

23 The area is Matlab Thana, in which the Inter-national Centre for Diarrhoeal Disease ResearchBangladesh (ICDDRB) happened to be doingfield research which allowed a careful monitoringof deaths during and after the famine. The begin-ning of the rise in deaths can be pinpointed clearlyfrom the ICDDRB data at around June 1974, andthe death rate was not to return to its pre-faminelevel for a full two years. I have collated theICDDRB data with independent monthly data onwage rates for casual agricultural abor and courserice prices collected by the Bangladesh Bureau of

Statistics for Comilla district Tin which MatlabThana is located).

24 The R2 is 0.60, and the Durbin-Watson statis-tic is 1.40. The regression comfortably passed LMtests for first-order serial correlation, functionalform, normality, and heteroscedasticity.

25 For further discussion and an attempt to re-solve these problems see Mark Pitt, Rosenzweig,and Donna Gibbons (1995).








Deaths by month in Matlab Thana 1974-76

700-

600 -

500 - S \ * Log deaths/ regressed on

0 0 log wages

0

400 _-z0~~~~~~~~~~~~~~

200 - l l l l l0.3 0.4 0.5 0.6 0.7 0.8 0.9

Agricultural wage (maund of rice/10 days)

Figure 1. Mortality and Wages in a Rural Area of BangladeshNote: The solid squares are for the famnine eriod (defined as Sept. 1974 to April 1975). The hollow squares are forbefore and after the famine.

among vulnerable groups to lower foodconsumption may be higher than theshort-run response.

From the limited evidence available,it can be argued that either an increasein the price of food staples or an in-come loss will put upward pressure onmortality, but that these variables onlyhave strong (or possibly even notice-able) effects when prices are initiallyhigh or incomes are low. A marked non-linearity in the response is to be ex-pected. It can also be argued that atinitially lower incomes (higher foodprices) there will be a higher mortalityresponse from food price increases (in-come losses). Though some importantclues have emerged from both microdata collected mainly under non-famineconditions and more aggregate data col-lected during famines, there is stillmuch to do in testing and quantifyingthese relationships.

3. Implications for UnderstandingFamines

Nonlinearity in the relationship be-tween mortality and consumption holdsa number of implications for under-standing famine causation. It impliesthat there will be long-term survivalgains from stabilizing the consumptionof a given person over time. Equalizingconsumption over time will not in g en-eral be optimal because the marginalimpact of consumption on survivalchances will vary with other time vary-ing factors such as the health environ-ment. Conversely, sufficient destabiliza-tion of consumption over time willworsen or even produce a famine. Simi-larly, certain unequalizing redistribu-tions of consumption will increase ex-pected mortality. However (because thesurvival function may vary between peo-ple), an equal distribution of a fixed








supply of food need not necessarilyminimize expected mortality. Nonethe-less, inequality increasing transfers ofsufficient size will clearly add to thedeath toll. Concavity of the survivalfunction also implies that a small dropin food consumption for a large numberof vulnerable people can entail a largeincrease in aggregate mortality. A sharpincrease in mortality may thus be pre-ceded by a steady (even slow) deteriora-tion in the food consumption of vulner-able groups. This nonlinearity can beexacerbated by shifts in the survival

function associated with a worseninghealth environment. Average food con-sumption of the poor in Bangladesh wason a downward trend for a decade or soprior to the 1974-75 famine (Ravallion1995). Though the evidence is less con-clusive, a similar process appears tohave been at work prior to recent fam-ines in Ethiopia, Sudan, and elsewherein Sub-Saharan Africa (Jansson, Harris,and Penrose 1987; B. Gopalakrishna

Kumar 1990; Webb, von Braun, andYisehac Yohannes 1992; von Braun,Teklu, and Webb 1994).26

The lesson here for understandingfamines is not to look only for a pasthistory of decline (for that is clearly nota necessary condition), but rather toquestion the still widespread tendencyto look for a sudden and sizeable shockto food entitlements as an essentialprimitive factor in famine causation.Such shocks are certainly common, buttheir impact may crucially depend onthe recent consumption history of vul-nerable groups. Nor does this mean thatthe onset of famine is easily predicted.The point at which a sharp escalation in

mortality will occur is still quite uncer-tain.

Concavity of the individual survivalfunction only holds above some point.Survival for more than a brief period, orany form of activity, is possible only ifnutrition is adequate for the body'sfunctions at rest. Such "nonconvexities"might hold insights for understandinghunger and behavior with greatly con-tracted food entitlements. With suffi-ciently low aggregate command overfood in some group (family, village, orrelief camp) it will be optimal from the

point of view of aggregate mortality todistribute food quite unequally. Thismay happen at normal times in verypoor families (Lipton 1983). Noncon-vexities suggest one possible explana-tion of the above average mortalityrates observed in certain subgroups(children and the elderly) during fam-ines; possibly these reflect corner solu-tions to the problem of minimizing ag-gregate mortality in a family or village.

However, this is not much more than aconjecture because there has been littleresearch into testing for such corner so-lutions. And there are other explana-tions, including non-cooperative behav-ior within the relevant groups, or thecollapse of institutional arrangementsfor redistribution and insurance, be-cause at low aggregate availability thepressure for the group to break up canalso be strong.

The nonlinearity of mortality to foodprice changes that has been found forfamines in which it has been possible todo the necessary tests also holds impli-cations for famine causation and reliefpolicy. Simulations using the econo-metric models of the effects of foodprice changes on mortality estimatedfor the aforementioned famines inSouth India and Bangladesh indicatedthat foodgrain price stabilization wouldhave reduced famine mortality (Raval-

26Also see Rangasami (1985) who emphasizesthe importance of understanding the "impoverish-ment process" which leads up to the observed es-calation in mortality, and Bruce Currey (1992)

who questions the common view of famines as"discrete events."








lion 1987, ch. 2). This is so even if meanprice had remained unchanged; moreplausibly, aggregate demand functions

for food staples are convex in price, soprice stabilization at a fixed aggregatesupply would also have entailed a lowermean price, with further reductions inmortality. (Section IV will return to thepolicy issues on how best to stabilizeprices.) The concavity in price also sug-gests that the underlying survival func-tion is concave in income; indeed, theincome elasticity of the income deriva-tive of the survival function must be

presumed to be high, undoubtedly ex-ceeding two. This suggests that faminemortality may be quite sensitive tochanges in interpersonal inequality.

Possibly the most serious limitationof the above discussion is that it hasbeen largely static. By this view, life-threatening entitlement failures arisefrom current events only. This limitsour understanding of famines.

A simple model illustrates the poten-

tial insights from introducing richer dy-namics. Following the standard assump-tion of the efficiency wage hypothesis,let current productivity depend on cur-rent food consumption in a nonlinearway, such that some positive consump-tion level is essential for any work ef-fort, but that beyond some level of con-sumption the marginal gains from extraconsumption start to fall. Suppose, how-ever, that there is a lag in the consump-tion impact of a change in productivity,and (in particular) that current con-sumption is proportional to last period'sproductivity. Then we can write con-sumption Ct at date t as a nonlinearfunction f(Cti1) of consumption at datet-1 as depicted in Figure 2 which illus-trates a case with three steady-stateequilibria (in which Ct = Ct-i), only twoof which entail positive consumption ofwhich only the higher one, at C*, is sta-ble. Suppose the initial equilibrium is at

Ct Ct= Ct 1

Ct =f(Ct l)

AI

0 CB CA CI Ct_l

Figure 2. Similar Shocks, but Dissimilar Outcomes

C*, but that there is a shock to currentconsumption driving t down to CA. Thedynamic adjustments will entail thatconsumption eventually returns to C*.However, at only a slightly arger shock,enough to drive current consumptiondown to CB, the outcome will be verydifferent; the dynamics will then drive

consumption down to zero. In one casethere is a short-lived contraction, n theother there is a sustained fall in con-sumption.

The key assumption here is that thereis a lag in the consumption impact ofproductivity changes. How might thisarise? Carraro 1996) provides an inter-esting model in which the dynamicsarises from the (plausible) assumptionthat current agricultural output de-pends on lagged labor inputs. Whencombined with the efficiency-wage hy-pothesis, and the nonlinear survivalfunction of Ravallion (1987), Carraroshows how mortality rates can respondsharply to production shocks, with dif-ferentially larger impacts among land-less workers.

The nonlinear dynamics can arise un-der even weaker assumptions. The usualstatic productivity-nutrition elationshipcan be interpreted as the steady state








solution of a dynamic of energy-conser-vation equation; the static relation-ship assumes that there is no change in

the body's stock of stored energy (Das-gupta 1993, ch. 14). However, for thepurposes at hand, one would not wantto make that assumption; people starv-ing in a famine will be running downtheir energy stores. Thus productivity,and hence consumption, in any periodwill depend in part on past consump-tion.

So nonlinear dynamics appear to beplausible in this setting. However, I

know of no empirical work testing formultiple steady states in poor peoples'consumptions. Given that one or moreof the equilibria will be unstable, thenecessary data could well be hard tofind. Arguably the multiple equilibriaare always there, but a famine is one ofthe few real-world setting in whichtheir potentially devastating conse-quences can be seen clearly.

III. Markets and Institutions

The last section pointed to ways inwhich even a seemingly small entitle-ment shock to poor people can inducelarge changes in their survival pros-pects, and also how similar shocks canhave dissimilar outcomes. But whatcauses the changes in entitlements?Various markets and institutions deter-mine the factors-the distribution ofendowments, the structure of prices,and the pattern of transfers-whichbear on the likelihood that an exoge-nous shock will turn into mass entitle-ment failure and hence famine. By andlarge these are the same markets andinstitutions which operate in normaltimes. Observers often point to theirbreakdown during famines. To under-stand famines we must understand the

atypical performanceof markets

andin-

stitutions in poor countries.

1. Food Markets

Food markets are often blamed forcausing or exacerbating famines.27 Fam-ines in market economies do not implyinefficiencies by the usual Pareto crite-rion; a reallocation that lets more peo-ple live may make others worse off. En-titlement failures leading to massstarvation are not inconsistent with effi-cient resource allocation in a suitablymodified version of the Walrasianmodel of a competitive market economy(Coles and Hammond 1995).28 Faminesneed not imply "market failure."

However, this observation will dorather little to subdue the concerns ofthe market's critics. Even if efficient,one can abhor the market allocationduring famines because of its sheerinequity; as is well known, Pareto effi-ciency is consistent with a highly in-equitable and (by any reasonable assess-ment) socially sub-optimal allocation.And concerns about the market's effi-

ciency during a famine persist even ifthe seeds of market failure during fam-ines were there before the famine. Themarkets that actually exist in famine-prone traditional agricultural economiesare almost certainly incomplete mar-kets, so there will normally be scope forPareto-improvements .29

One source of market failure is thatmarkets generally do not exist to permit

27See, for example, Bhatia's ( 1963) descriptionof "price famines" in India during the nineteenthcentury. Also see Greenough's (1982) study of thefamine in Bengal 1943-44, and the discussion inRashid (1980). Similarly, there were reports ofhoarding by "grain barons" in Sudan in 1985 (J.Steele 1985).

28The standard model of competitive equilib-rium assumes that everyone can survive withouttrade. Responding in part to conjectures in Desai(1989), Coles and Hammond (1995) show thatPareto optimality of competitive markets can coex-ist with some individuals dying in equilibrium be-cause they do not attain some minimal consump-tion level.

29For a review of this topic, in a developmentcontext, see, for example, Stiglitz (1988).








trade in contracts for future grain deliv-ery. Thus, current trading decisions areconditional on expectations about the

prices which will prevail in future spotmarkets. And so expectations formationwill matter for the allocation of con-sumption over time and hence mortal-ity. Those facing starvation now willnaturally have a high rate of discount.Yet even greater starvation in the aggre-gate may result without storage. In amarket economy, this will depend onhow well stock holders predict futurescarcities. This is not a new observation,but (as with so much of the modern dis-cussion of famines) had clear antece-dents in the eighteenth and nineteenthcentury debates over famine relief pol-icy.30

In normal times we may tolerate adegree of market inequity or failure,because the cost of avoiding it exceedsthe gain. The more interesting questionin the present context is whether thosesame features of markets can magnifyadverse external shocks. Anecdotesabound of "panic buying" of food in re-sponse to news of impending famine.For example, there were many storiesof such behavior in newspapers duringthe lead up to the 1974-75 famine inBangladesh (Ravallion 1987, ch. 3). Cansuch behavior turn the threat of famineinto a reality?

Data have severely constrained ef-

forts to address this questionin famine-

vulnerable settings. Bangladesh hasbeen an important exception. Surveydata on rice price expectations ofwholesale traders and stock holders inBangladesh rice markets shows littlesign of informational efficiency (Raval-

lion 1987, ch. 6). The restrictions im-plied by the unbiasedness of one-day-ahead price forecasts are convincingly

rejected at the individual level, with atendency to overestimate price changesindicated for all sampled traders. Therandom walk model of price forecastsfor a terminal date is also rejected as ageneralization of traders' forecastingbehavior. Overall, the rational expecta-tions hypothesis is a poor charac-terization of this market.

We do not have the benefit of sur-veyed expectations data during a faminenor even reliable data on importantquantities, such as stocks. But we canstudy the effect on current prices ofnew information about future scarcities.Newspaper reports of flood damage tostanding crops resulted in higher cur-rent price rises during the 1974-75famine in Bangladesh, and this effectexisted independently of future prices(Ravallion 1987, ch. 3). Price forecast-ing errors appear to have been posi-tively correlated with readily availableinformation on damage to the futureharvest. Thus, rice hoarding prior to an-ticipated food-availability decline wasexcessive when compared to the likelyoutcome under competitive conditionswith informationally efficient expecta-tions. This appears to be the main rea-son why rice prices rose so dramaticallyin the lean months before the main(1974-75) winter harvest; rice pricesmore than doubled between March andOctober of 1974, leading to a devastat-ing contraction in command over foodamong those with little endowment tofall back on. As is typical of the panicstage after a speculative mania, riceprices fell almost as sharply in marketsduring the week or two before the nextharvest started to arrive in November.And in the end it turned out that the

harvest was depleted by less that fivepercent over its previous level, which

30Adam Smith's (1961, p. 41) classic defense ofthe holding of food stocks during a famine explic-itly assumed that stockholders would correctlypredict future prices. For further discussion of thehistorical debates over famine relief policy seeRashid (1980).








had been unusually high; it is evenquestionable whether output fell belowtrend.3' Similar shocks to aggregatefood availability in Bangladesh beforeand since that famine, had nothing likethe same disastrous consequences.

The policies of the government andprincipal aid donors appear to haveworsened matters. Rice prices in Ban-gladesh are influenced by the govern-ment's foodgrain stock, as determinedby previous imports (including foreignaid) and internal procurement efforts.Thus, expectations of the effect of pre-

harvest crop damage on future pricesrequire an assumption about the re-sponse of the domestic government andforeign governments to shared informa-tion. The most plausible conclusion isthat the stockholders' high price expec-tations during the 1974-75 famine werepremised on a belief that the govern-ment would be unable to implement asuitable stabilizing response to the re-ported damage to the future crop. Do-

nor threats to cut food aid on foreignpolicy grounds in early 1974 helped fuelthat belief. It has also been argued thataccelerated growth in the money supplyhelped fuel inflationary expectations(Bankim Chadha and Ranjit Teja 1992).The failings of foodgrain markets thusappear to have stemmed in part from(domestic and international) public-ac-tion failures.

Informational inefficiencies need notdestabilize markets, and considerableinstability is possible in an informa-tionally efficient market. Quddus andCharles Becker (1994) show that therice price formation process in Bangla-desh is consistent with the possibility of

a speculative bubble even with rationalexpectations. However, it is clear thatthe way rice markets worked in Bangla-desh meant that the combined effect ofthe crop damage from flooding and theexternal threats to food aid led to con-siderable de-stabilization of prices andconsumption. In the end, there wasclearly enough rice to avoid famine ifonly it had been better distributed forusage over time.

Concerns have also been raised aboutthe spatial performance of markets ingetting food from surplus to deficit ar-

eas. Dynamic analyses of spatial pricedifferentials suggest significant impedi-ments to market integration during fam-ines; see Ravallion (1987, ch. 5) on theBangladesh famine during 1974-75, andvon Braun, Teklu, and Webb (1994, ch.5) on the famines in Ethiopia and Su-dan in the mid-1980s. The causes ap-pear to include unusual transport costsand a variety of higher risks associatedwith the famine conditions (which most

local transporters would have no oppor-tunity to insure against); these stemmedin part from disruptions to trade due tothe famine or its proximate causes, andwere exacerbated by faulty policy re-sponses, such as deliberately impedingthe flow of food to protect better offregions.

2. Labor and Other Markets

Food markets have been the focus ofmuch of the literature. Yet the marketsfor the things that poor people sell willbe just as important to the outcomes.The capacity for work is the main assetof poor people. Naturally, when there islittle inequality in access to land, theagricultural labor market tends to bethin. In Asia, and increasingly in Africa,there is sufficient land inequality forthe labor market to be crucial to the im-pact of aggregate shocks on the ruralpoor. For example, the real agricultural

31 There are a number of problems in assessingagricultural output for Bangladesh; the statementsmade here are based on the careful work of Boyce(1987). However, they are also consistent with

more commonly used sources, as used by Alamgir(1980) and Sen (1981, ch. 9).








wage rate in terms of the food staplefell dramatically during the 1974-75famine in Bangladesh (Figure 1), and

landless agricultural workers, part-timefarmers and village artisans were thefamine's main victims (Alamgir 1980;Sen 1981). There is evidence of a sig-nificant structural break in the short-run wage adjustment process in Bangla-desh during the 1974-75 famine,without which the real wage rate inunits of rice would have been almosttwice as high during the worst monthsof the famine (Ravallion 1987, ch. 4).The famine appears to have been asso-ciated with a sizeable shock to rural la-bor markets, originating in the contrac-tion in agricultural employment thatcame with the flooding.

A standard competitive model of thelabor market would predict a lower realwage rate when an aggregate shock re-duces demand for labor. However, sucha model sits uncomfortably with otherobservations of seasonal underemploy-ment in Bangladesh, as in other under-developed agrarian economies.32 Onepossible explanation for wage stickinessdownwards is tacit collusion on the sup-ply side within the village. This will bedifficult to maintain during a famine inwhich the gains to defecting from sucharrangements become high, and newworkers appear on the local labor mar-ket who have migrated from flood af-fected areas. The sharp fall in the realwage rate during a famine may reflect afailure of an aspect of the village "moraleconomy" (discussed further below).

We do not know whether the wagefell enough during the famine to elimi-nate the famine unemployment. In allthe academic literature and newspaperreports in Bangladesh on this famine, Ihave never read a claim that workers

during the famine no longer had troublefinding work, which one would havethought highly notable if true, given

how much seasonal unemploymentthere appears to be in normal times. Itshould not be presumed that even sucha sharp fall in the real wage rate wouldbe enough to eliminate unemploymentin this setting. A worker's productivityat a low food wage may be so low thatemployers resist further wage cuts, as inthe efficiency wage hypothesis. Thatwill depend on the worker's othersources of consumption. Those amongthe poor with least non-labor endow-ments will be the ones left to starve, forthey will find that their best wage offersare undercut by other people withendowments which can be consumedduring the famine so as to raise theirproductivity to a competitive level(Dasgupta and Ray 1986; McGregor1990; Dasgupta 1993).33 Labor marketresponses during a famine may thus en-tail a "double whammy" for very poorpeople-their consumption falls di-rectly because of the higher price offood, and it falls indirectly through theincome effect (both on wages and em-ployment) of their diminished produc-tivity. This type of labor-market re-sponse can readily generate the type ofnonlinear dynamics in the evolution ofpoor people's consumptions and multi-ple equilibria discussed in Section II. Avariety of dynamic behaviors for con-sumption are then possible in responseto a labor-demand shock, includingdownward spirally contractions. The dy-

32 Rizwanul Islam (1986) reviews evidence onthis issue for Bangladesh.

33While there is evidence from micro studiesthat undernutrition reduces productivity, the em-pirical relevance of any downward wage-stickinessarising from such nutrition-based productivity ef-fects has been questioned at normal times even inpoor rural economies (Pranab Bardhan 1984, ch.4; Anand Swamy, forthcoming; Subramanian andDeaton 1996). Nonetheless, it can be conjecturedthat these effects are likely to be more relevant

during situations of unusually low wages, as oftenobserved during famines.








namic entitlement failure can entailboth higher unemployment and lowerreal wage rates (Carraro 1996).

Other markets are likely to matter tothe transmission of aggregate shocks tothe household level. With the pressureon informal credit and risk-sharing ar-rangements, and the highly covariatenature of the income shocks duringfamines, any existing formal credit mar-kets will probably see rising interestrates or rationing. However, this israrely mentioned in the literature,probably due to the fact that formal

credit markets are already thin in fam-ine-vulnerable economies, with borrow-ing constraints (particularly for thepoor) being the rule rather than the ex-ception.

With incomplete risk markets (entail-ing interdependence of production andconsumption decisions), a productioninput such as livestock can be an impor-tant asset which helps protect the con-sumption of poor people. This is welldocumented for the pastoral economiesof the Horn of Africa (Webb, vonBraun, and Yohannes 1992; von Braun,Teklu, and Webb 1994) and the Sahel(M. Khan 1994). Rosenzweig andWolpin (1993) find evidence that live-stock is used for consumption smooth-ing in semi-arid areas of India. Thehighly covariate nature of the shockduring a famine means that large num-bers of people must simultaneously selltheir livestock to buy foodgrains, with aconsequent (often sharp) fall in live-stock prices relative to foodgrains andother goods (Sen 1981; Marion Kelly1992; Paul Ulrich 1993; von Braun,Teklu, and Webb 1994; M. Khan 1994).This may be exacerbated by panic sell-ing of livestock in anticipation of fallingprices.

There is still much we do not knowabout (food and nonfood) market per-formance during famines. We should

clearly not assume that free marketswill help buffer an initial shock's impacton the poor. The extent to which mar-

kets help or hurt will depend on the na-ture of the market failures in normaltimes, and on how these interact withinitial distribution and with the natureof the initial shock. What is neededmost is further careful empirical workon how markets actually worked inthese circumstances.

3. The "Moral Economy" and Famines

There are undoubtedly a great manypeople who would not have died in afamine if only they could have bor-rowed without rationing, at rates of in-terest which equated demand and sup-ply for credit. Credit and risk marketsin famine-vulnerable settings appear tobe less than perfect at the best of times.And it is unlikely that they would workany better during a famine; indeed,there is (largely anecdotal) evidencethat existing market and non-market in-stitutions for credit and insurance per-form less well during famines; patron-age becomes scarce and "anti-socialbehavior" increases, such as renegingon familial and village obligations;34"Patron-client relationships, alreadyweak, are found virtually to snap duringperiods of severe food stress" (BinaAgarwal 1991, p. 207).35

34 The contraction or collapse of the variousforms of traditional credit and risk-sharing (in-cluding patron-client) arrangements during fam-ines or other food crises has been noted by Bang-ladesh Rural Advancement Committee (1979),Robert Dirks (1980), Alamgir (1980), Greenough(1982), Caldwell, Reddy, and Caldwell (1986),Megan Vaughan (1987), Frances D'Souza (1988),Agarwal (1991), John Field (1993a), and AlayneAdams (1993), among others.

35 It has also been argued that the indigenousmoral economy is eroded by the developing mar-ket economy, and that these are typically settingsin which no formal (public) safety net exists to re-place the moral economy; so longer-term vulner-

ability to famine can be increasing (see, for exam-ple, Sen's 1981, ch. 8, discussion of the erosion of








It is not surprising that famines canput considerable stress on normally sus-tainable-albeit imperfect-social ar-

rangements for credit and insurance inpoor economies. Village societies cancope to some extent with idiosyncraticrisks through risk-pooling arrange-ments. Such arrangements will probablybe less prone to moral hazard and ad-verse selection in traditional village andextended-family settings in which par-ticipants are well known to each other.Indeed, there have been claims frommicroeconometric investigations thatindigenous risk-sharing arrangements inpoor agrarian economies work well inpractice, though there is conflictingevidence from the same data of sizableexposure to income shocks.36 These ar-rangements cannot, however, be ex-pected to cope with covariate shocks tothe risk-sharing group, as are likely tobe common in a famine. And even whenthe risk-sharing group is large, or thereis a sizable idiosyncratic component tothe shock, in famine-prone settingsthese arrangements must still be imple-mentable without binding, legally en-forceable contracts. This fact constrainsperformance for the poor, particularlyin spells of transient poverty, or whenthe threat of destitution reduces theprobability of continued participation inthe social insurance game. Options out-side the household and village may be-

come more attractive.So there can be no guarantee that

routine, mutually beneficial, coopera-tive behavior will survive a famine-that will depend on the initial condi-

tions (including preferences) and thesize of the shock. Some strands of thetraditional network of support willbreak more easily than others; for exam-ple, Greenough (1982) notes thatlonger-term attachments were lesslikely to be broken during the Bengalfamine. Nonetheless, simulations sug-gest that seemingly small perturbationsto the parameters (the extent of risk-aversion and the discount rate) of a re-peated informal insurance game amongpurely self-interested parties can entaila large change in the realizable gains(Stephen Coate and Ravallion 1993).

The disruption of community and fa-milial reciprocity or benevolence isoften (though not always) associatedwith migration (Jodha 1975; D'Souza1988). It is not clear what is the causeand what is the effect here. Whengroup-based insurance fails, migrationis certainly one response. But migrationcan also help cause the breakdown, be-cause the village-support game relies onrepeated interaction among people whocome to know and trust each other; thelarge scale migrations observed duringmany famines may mean, however, thatnew players come on the scene who arelikely to treat it as a one-shot game.Again a rapid collapse of informal

risk-sharing arrangements can be ex-pected.This point also illustrates the perva-

sive, but often neglected, externalitiesinvolved in famines. In various ways, in-dividually rational responses to thethreat of starvation (of which migrationis one example) can aggregate into anenhancement of that threat. One way isthrough the disruption of socially basedsupport mechanisms premised on a lim-

ited capacity for collective action; an-other way is through enhanced exposure

traditional support systems in the Sahelian econ-omy). This topic takes us beyond the scope of thepresent paper. Further discussion can be found inAgarwal (1991), Jean-Phillipe Platteau (1991), andOsmani (1996).

36 Townsend (1994) argues that data for threeIndian villages are consistent with near-perfectrisk-sharing. However, on applying alternative es-timation methods to the same primary data set,Ravallion and Shubham Chaudhuri d1997) find

that consumptions are quite sensitive to incomechanges.








to infectious diseases in crowded reliefcamps. Individual efforts to cope withlife-threatening hunger can escalateinto an even greater aggregate threat.The view that mass starvation is just asimple sum of many individuals' inde-pendent hungers is questionable. How-ever there is still little hard evidence onthese issues.

4. Failures of Public Action

Even with well informed and well in-tentioned policies, governments andother agencies in poor countries canface severe infrastructural, administra-tive, and logistical constraints on theircapacity for action. This has been acommon problem in famines in under-developed or war-torn economies. In-adequate rural infrastructure and com-munications has often been identifiedas a contributing factor to famines, anda severe constraint on their relief.37Complex problems such as famines mayso strain infrastructure and administra-tive capabilities that a sharp deteriora-tion in performance becomes inevita-ble. This may well reflect past failuresof public action (such as neglecting ru-ral infrastructure).

There have also been serious shortterm public action failures. First thereare policy failures resulting from faultytheories or misinformation. There havebeen many examples. The British Gov-

ernment's faith during the nineteenthcentury in non-intervention in foodmarkets during famines almost certainlymade matters worse, by deterring pub-lic action to improve on the distri-butional outcomes of unfettered mar-kets (Ambirajan 1978; Rashid 1980;Dreze 1990). At the other extreme, the

forced collectivization of agriculture,and associated food procurement poli-cies implemented under the SovietUnion's first five-year plan resulted in asevere famine in rural Ukraine in 1930s(Dalrymple 1964; Dando 1981). Similarpolicies in Ethiopia from the mid-1970sare believed to have contributed to thefamine of the mid-1980s (Devereux1993b, ch. 10). The collapse of food en-titlements in China during 1959-61 isbelieved to have been instigated by thedestructive effects on food productionof the rapid industrialization policies

implemented as part of the "Great LeapForward," compounded by an urbanbias in food distribution and failures ofstatistical reporting (Ashton et al. 1984;Riskin 1990; Lin and Yang 1995). Simi-larly, wartime policies enforcing rice-export quotas and the production ofnon-food crops contributed to the fam-ine in Vietnam in 1943-45 (Bose 1990).Milder interventions have also gonebadly wrong at times. There have been

a number of descriptions of how gov-ernmental interventions in agriculturalmarkets enhanced vulnerability to fam-ines in Africa (Jansson, Harris, andPenrose 1987; Vaughan 1987; SimonMaxwell 1991; Stryker 1993). Othercommon examples of this type of pub-lic-action failure are the attempts madeto prevent food or people moving in re-sponse to famines (Quddus and Rashid1991; Devereux 1993b, ch. 10).

Secondly, there are public-action fail-ures that arise when those in power maynot share the objective of avoiding orrelieving famines. In some famine-prone economies, the government's sur-vival appears to have little to do with itsability to secure the basic consumptionneeds of those governed. Indeed,higher poverty during famines mayoften come side by side with absolute,as well as relative, gains to more power-ful subgroups of the non-poor. Thus the

37See, for example, Dreze (1990) on famines inBritish India, Leonard Berry and Thomas Down-ing (1993) on droughts in Africa during the 1980s,

and Jansson, Harris, and Penrose (1987) and Ku-mar (1990) on the Ethiopian famine of 1984-85.








incentives facing political leaders canwork against effective famine avoidanceor response.

An extreme, but not uncommon, formof this type of public-action failure iswhen a government knowingly usesfamine as a weapon against its (externaland internal) enemies (Dando 1981;David Keen 1991; Joanna Macrae andAnthony Zwi 1992; de Waal 1993). Buteven when the famine is not itself theweapon, it can be an entirely predict-able by-product of governmental poli-cies, such as when militarization de-

stroys the ability of an economy andsociety to cope with drought (Devereux1993b, ch. 10). Public-action failurethrough denial or obfuscation of thesigns of famine has been associated withsome of the worst famines in terms oflives lost, including the Ukraine famineof 1932-33 (Dalrymple 1964), and thatin China during 1959-61 (Ashton et al.1984; Riskin 1990); in both cases thegovernment refused to acknowledge theproblem, or to seek external assistance,and the outside world knew very littleabout it until many years later.

Within this category of deliberatepublic action failures are also examplesin which the threat of famine was recog-nized, but for other reasons. It has beenclaimed that the host government dur-ing the Ethiopian famine in 1983-85 ex-ploited perceived donor sympathy forthe victims to lobby for famine aid in-tended for other purposes, or to free upforeign exchange for other purposes in-cluding buying weapons (Cutler 1993;Kloos and Lindtj0rn 1993). And theproblem need not be confined to thehost governments. One major donorgovernment is claimed to have usedfamine aid to Ethiopia as an instrumentof foreign policy, with the aim of over-turning a government it did not like onideological grounds (Jack Shepherd1993).

IV. Famine Policies

Famines demonstrate that marketand non-market institutions cannot berelied upon to always operate in wayswhich will adequately buffer an externalshock's impact at the individual level.Indeed, under certain conditions, con-siderable magnification of the shock ispossible in the chain of events leadingto famine. What are the lessons for pol-icy, and research on policy?

1. Better Governance

The solution to some of the public-action failures identified in Section IIIis clear enough: base the policy re-sponse on a realistic and undogmatic as-sessment of how the specific economyworks and build the physical and insti-tutional capacity for making the re-sponse work. Examples of how this canbe done are discussed later.

The solution to deliberate public ac-tion failures is less clear. One factor

identified in recent literature is the ex-tent to which information can flowfreely and public pressure for action beexpressed without recrimination. In ad-dition to open formal political institu-tions, an independent press has beenseen to have an important role in avoid-ing famines (Dreze and Sen 1989; N.Ram 1990). And the negative effects agovernment having the ability to sup-press news and criticism have also beenidentified; for example, this appears tohave been a contributing factor to theunusual severity of the Chinese famineof 1959-61 (Sen 1983) and the Ethio-pian famine of 1984-85 (Peter Gill1986; Jansson, Harris, and Penrose1987). Also attempts to suppress infor-mation often come hand in hand withdeliberate misinformation, such as un-convincing promises, which can evenmake matters worse by further destabi-lizing markets (Ravallion 1987; Quddus








and Rashid 1991). There is also evi-dence that democracy is instrumentallyimportant to avoiding famines; in Africa

since about 1980, more democraticcountries have been less likely to have afamine (von Braun, Teklu, and Webb1994), though that correlation is basedon very few degrees of freedom (until1990 only two or three countries inSub-Saharan Africa could reasonably becalled democratic).

However, while it is plausible thatopen political institutions and a freepress constrain the scope for this type

of deliberate public action failure, it isnot clear what the policy instrumentsare here. The extent to which informa-tion and criticism can flow freely isoften under the control of the sameagents whose preferences inhibit fam-ine relief.

External levers might be identified, ifdonors and international agencies werewilling to apply appropriate conditional-ity to their aid and concessional lend-ing. Multi-lateral aid agencies and (non-governmental and private) voluntaryagencies can be trusted to have a firmercommitment to saving lives during afamine than some governments, thoughhere too governments can try to ma-nipulate or suppress these agencies, ashas happened in relatively illiberal po-litical environments (Adams 1993).Some of these agencies appear unawareof-or reticent to exploit-their ownbargaining power to improve host-gov-ernment policies, preferring instead todisburse famine aid as rapidly as possi-ble without conditions. It should not beassumed that unquestioning handouts of"humanitarian aid" will best serve hu-manitarian ends.

Clearly more research is needed onthe political and institutional dimen-sions of famine causation and relief. For

now, it is believable that the increasingpolitical liberalization we are seeing in

famine-prone economies will at leastconstrain the possibilities for these de-liberate public action failures. Other

constraints are also emerging; the"CNN factor" is one, whereby the inter-national visual media can now zoom inrapidly on trouble spots even when thelocal regime is trying to suppress infor-mation. If these recent trends continue,then it is likely that political leaders willhave a greater demand for sound eco-nomic analysis in formulating anti-fam-ine policies. The rest of this section re-views issues pertaining to specific

policies aimed at relieving and prevent-ing future famines.

2. Early Warning and Rapid Response

Aggregate food availability has tradi-tionally been the main indicator usedfor predicting famines. The recent lit-erature has suggested that similar initialconditions of aggregate food supply canentail very different outcomes. Even ahighly sophisticated forecasting model,and accurate data, may produce unreli-able predictions of how much FAD willimpact on the welfare of poor people infamine-vulnerable settings. What canbe done?

Arguably the best place to look forsigns of impending famine is in the vari-ables which directly determine the enti-tlements of poor people (prices, wages,employment, and agricultural yields).Here the entitlements approach has in-fluenced practical early-warning andmonitoring methods, such as used byOxfam (Young 1992). Much has beenlearnt about early-warning indicators bystudying behavioral responses to con-tractions in food entitlements (W. I.Torry 1988; Corbett 1988; Kelly 1992;M. Khan 1994). Such research has re-vealed promising indicators in specificsettings, although the degree of hetero-

geneity in behavioral responses evenwithin one well defined area and socio-








economic group (let alone problems ofdata quality and administrative feasibil-ity) has constrained efforts to find ro-

bust "universal" early warning indica-tors (Corbett 1988; Susanna Davies1993).

The public response to early warningsmust also be timely. Late responses arecostly in lives and money. A greater re-duction in mortality can generally beachieved with given aid if it is distrib-uted more evenly over time (given thelikely concavity of the survival function,as discussed in Section II). Adverse ef-fects on producers' future price expec-tations and, hence, future food avail-ability can also be reduced with a moretimely response. Often, however, thecosts of a delayed response are not in-curred by those making the decision ofwhen to respond. For example, variousdistress signals (consumption of wildfoods, unusual migration, selling of live-stock to buy food, and high food prices)were evident for at least 12 monthsprior to the main external interventionsin Ethiopia in late 1984 (Gill 1986;Jansson, Harris, and Penrose 1987; Ku-mar 1990).

There are examples of how rapid andeffective public responses to signs offamine have almost certainly averteddisaster. These include Kenya and Bo-tswana in the mid-1980s, and Ban-gladesh in 1979 and 1984. But among

famine-vulnerable countries, post-Inde-pendence India has provided the mostwell documented examples of effectiveearly warning and public response. TheFamine Codes developed in the nine-teenth century originally relied largelyon monitoring foodgrain supply indica-tors (such as failed rains), though thisbroadened to include food prices andunusual population movements. TheMadras Famine Code of 1883 recom-

mended monitoring grain prices as anindicator of famine (Rangasami 1985),

and the Bengal Famine Code of 1918outlined an elaborate early warning sys-tem (Currey 1984). These Famine

Codes formed the germ of the post-In-dependence early warning system, butthe public response system was devel-oped greatly (Government of India1989). Local administrators report signsof famine; their efforts are monitored,and an open press is rarely reticent toplay a "watchdog" role (Ram 1990). Lo-cal officials maintain the capability forrapid response. The administrative inte-gration of warning and response func-tions, combined with political will and afree press, have undoubtedly been im-portant to India's success at avoidingfamines since Independence, despitenumerous droughts (Dreze 1990). How-ever, as a number of observers havenoted, India's success at fightingchronic poverty and hunger has beenmore disappointing.

3. On the Role of Aggregate FoodAvailability

It has been argued that an increase inaggregate food availability is not calledfor to stop a famine which was not itselfcaused by a decline in aggregate avail-ability (Bowbrick 1986). And the beliefseems to have occasionally been putinto practice. For example, the sluggishresponse of some aid agencies to earlywarnings of the 1983-85 famine inEthiopia has been attributed to their as-sumption that private food surpluses inthe country were sufficient to maintainaggregate consumption, in spite of thedrought (Gill 1986).38 Of course, the aidagency's ex ante assessment of food sup-ply is bound to be imprecise and it nowappears likely that availability was seri-

38 Another well documented example is theGreat Bengal Famine of 1943 in which the slug-gish response from the authorities appears to havebeen due in part to the fact that they could findlittle sign of FAD (Sen 1981, ch. 6).








ously low in Ethiopia in 1984-85 (Ku-mar 1990). But that is not the issuehere. Even if aggregate domestic avail-

ability is not threatened, can there be acase for emergency interventions aimedat increasing the supply of food to thedomestic economy in response to signsof an increase in starvation?

A number of arguments can be madein favor of increasing aggregate foodavailability in response to a famine,even if it not caused by FAD.39 Con-sider the stylized case of starvation aris-ing from an adverse shift in food distri-

bution, as a result of which the poorend up with a lower share of an un-changed aggregate. In principle, the ab-solute consumption levels of the poorcould be restored either by restoringthe previous food distribution function,or by increasing aggregate availability.It is not difficult to imagine situationsin which the first option is unfeasible inthe (crucial) short run.

To give another illustration, supposethat private storage is deemed to be ex-cessive. A government can try to dealwith this by sending the police out tobash up "hoarders," as happened, forexample, during the Bangladesh famine.This may work, but one suspects that itis more effective as a public relationsexercise than as a means of increasingcurrent food consumption. And onemust also appreciate the importance ofthe government's behavior to the for-mation of stockholders' price expecta-tions. In these circumstances, the im-mediate goal should be to take actionswhich lead stockholders to revise down-wards their expectations of futureprices. A believable undertaking to im-port food would almost certainly dothis.

Is there also a case for promoting do-

mestic food production as part of alonger-term strategy for preventingfamines? The role of commercialization

of agriculture and the promotion of ex-port crops (both food and non-food) hasbeen much debated. It is sometimes ar-gued that this will divert resources awayfrom the domestic production of food,and expose poor farmers to even greaterrisk of food entitlement failure (see, forexample, Frances Lappe and JosephCollins 1977). There have been cases inwhich a new export crop has increasedexposure to risk; groundnut in the West

African Sahel is an example (Devereux1993b, ch. 9). However, as a generaliza-tion, the view that export-crop produc-tion creates famines is questionable; ifby producing a cash crop instead the ru-ral poor can afford to buy more food,and the markets and infrastructure areadequate for getting the food to themfrom elsewhere, then that will reducetheir vulnerability to famine.40 Similarcomments apply to the idea that thehighest priority for famine-vulnerableareas is to promote food production soas to be self-sufficient in food; in re-gions where the land is much more suit-able to cash crops that can be a formu-lae for a severe food entitlementfailure. Nor does self-sufficiency infood appear to be an ecologically soundprinciple for tropical agriculture (P. F.Stewart 1988).

Agriculture and rural developmentcan play an important role in reducingvulnerability to famine, both in provid-ing extra insurance (such as throughagricultural diversification, and non-farm jobs in lean seasons) and longer-term development. Unfortunately, pastdevelopment policies (both pricingand spending) have typically been bi-

39Contrary to the claims of Bowbrick, that isentirely consistent with Sen's (1981) arguments.

40 For evidence on the effects of agriculturalcommercialization on poverty and hunger see the

various papers in von Braun and Eileen Kennedy(1994).








ased against this sector (Lipton 1977;Anne Krueger, Maurice Schiff, and Al-berto Valdez 1988). The policy lesson is

clear.

4. Distribution Policies

Although the case is often strong forincreasing aggregate food availabilityduring a famine, food handouts neednot be the best form of interventionfrom the point of view of minimizingmortality. Cash or coupon payments topotential famine victims can provide

more effective relief than the usual pol-icy of importing and distributing food(Reutlinger 1988; Dreze and Sen 1989;Coate 1989; Sen 1990; Maxwell and G.Templer 1994). Under competitive con-ditions, the case in favor of cash relief isstrong if either food prices in the af-fected region are below world prices atthe border (because, of course, recipi-ents will then be able to purchase morefood locally with the cash aid than theaid agency can obtain with the samemoney on world markets) or if privatetraders can deliver the food more effi-ciently than the aid agency. Some ob-servers have seen that traders get thefood to affected areas quicker than gov-ernments or aid agencies.41 The case infavor of cash relief is less obvious if lo-cal food markets perform poorly,though even then the policy switch maybe desirable; for example, cash reliefcan result in lower mortality than directfood aid even if the local food trade ismonopolized, provided that (amongother conditions) traders can still de-liver food at lower cost than the aidagency (Coate 1989). Savings in storageand handling costs can mean that cashhas a higher transfer benefit to recipi-ents. It can also help avoid the need forrecipients to congregate in crowded and

unhealthy relief camps. The rapidmonetization of food aid to finance en-titlement-protection programs has

helped some famine-prone countriesbetter avoid famines.42 However, donorrestrictions on monetizing food aid re-main common.

Whether in the form of food or cashrelief, the effect of aid on mortalitywill depend greatly on how it is distrib-uted over time and between people.Governments and agencies involved infamine relief face the problem of howto allocate a fixed quantity of emer-gency food aid. Indicators of aggre-gate food availability in specific regionsare sometimes used, though (for thesame reasons that FAD is questionabletheory of famine causation) this may notmean that the transfers go to those ingreatest need, and there is at least onedocumented case in which focusing onlocal foodgrain output shortfalls led theaid agency to miss those in greatestneed (Webb and Thomas Reardon,1992, on the 1983/84 drought in Bur-kina Faso).

Food allocation according to moredirect indicators of "need" is often ad-vocated and used, and guidelines fordoing so have been laid down by someNGOs and UN agencies.43 In practicethis often means that handouts arebased on readily observed individual orgroup attributes such as assessments

of age, height, and weight. This wouldseem more likely to achieve better out-comes than targeting according to foodoutput shortfalls, though finer targetingcan also entail higher costs; we know lit-

41 See Caldwell and Caldwell (1992), referringto the drought in the Sahel in the early 1970s.

42 See Dreze and Sen's (1989) discussion ofCape Verde's policies in which food aid is sold onopen markets and used to finance cash transfersfor protecting the command over food of poorpeople in the presence of severe droughts.

43 For example, the Oxfam Health Unit (1984)advocates the use of need indicators such asweight-for-height indicators. Also see Seaman andRivers (1988) and Young (1992).








tle about the precise tradeoffs involvedduring famines.

A thorny question arises in this con-text. As a general rule, the mortalityminimizing allocation of a given supplyof food over a population will equalizethe food consumption slope of the sur-vival function across all persons. Theoptimal allocation to each person willbe implicitly a function of relevant per-sonal attributes, interpretable as eachperson's "needs" (Ravallion 1987, ch.7). Implementing an (unconstrained)optimal allocation of given aid may be

very far from being feasible in practice.However, it is not even clear that an op-timal allocation from the point of viewof aggregate mortality would look muchlike the guidelines typically laid down.There can be no general presumptionthat the mortality minimizing handoutof food or cash is an increasing functionof need. Indeed, the opposite will holdwhenever persons in greater need havelower consumption slopes of their sur-

vival functions.44 On moral grounds onemay prefer to give more food to thosein greater need, but one may also haveto come to terms morally with the factthat this could well involve more peopledying than strictly necessary with thegiven amount of food available.

Direct intervention to improvehealth-such as through disease treat-ment and control, water purification,and improved sanitation-will not onlyreduce mortality directly, but may en-tail that other policies which enhancefood entitlements will have larger mar-ginal impacts on survival prospects. Ifso, then the converse will also hold; invarious ways greater command over

food will enhance the impact of health-based interventions.45 Or effective (cashor food) transfers may obviate the needfor health-based intervention, such aswhen relief reaches villagers beforemigration begins, thus reducing theneed for crowded relief camps whereinfection becomes more likely. Theliterature contains discussions of suchsynergies between health and more con-ventional food- or cash-based interven-tions (see for example Young and Jas-pers 1995). But as yet the relationshipsare not well understood, or quantified

to a point where one can give firm guid-ance on the priorities that should be at-tached to each mode of interventionwhen aiming to minimize famine mor-tality.

The administrative capability for im-plementing indicator-based targeting ofinterventions during famines is limitedin many settings. This has led to exten-sive use of one of the oldest forms offamine relief policy, namely the provi-

sion of unskilled employment to anyonewho wants it at low wages on publicworks projects.46 The attraction of thispolicy has not been the value of the out-puts produced, but its incentive effects,notably the potential for "self-targeting"(given that those not in need are un-likely to want to do such work at thewages offered) and that incentives forescaping poverty by other means arepreserved (Timothy Besley and Coate1992). Furthermore, in famines arisingfrom a collapse of employment, the

44This does not require nonconvexities leadingto corner solutions, in which some are left to die.The solution may be an interior one, but with het-erogeneity in marginal impacts of consumption on

survival prospects. For an analysis of a relatedproblem in targeting see Michael Keen (1992).

45 Non-food goods may also matter; for example,Rivers (1988) argues that better clothing and shel-ter will reduce nutritional requirements.

46 Though intervention in food markets was dis-couraged, relief works schemes were prominent inthe "Famine Codes" in British India (Bhatia 1963;Dreze 1990). The aim was to provide local em-ployment at a subsistence wage to all who wantedit. On the use of this policy in famines in Africasee Dreze (1990) and von Braun, Teklu, and

Webb (1994, ch. 8). For a survey of the literatureon these schemes see Ravallion (1991).








forgone incomes of relief-work partici-pants will be low, and hence the reliefcan be cost-effective. By allowing a

separation of those able to work fromthose not, and by targeting on other in-dicators of need among the lattergroup, this type of policy can be an im-portant component of a comprehensiverelief strategy (Dreze and Sen 1989).

The body of experience and evidenceon public works schemes offers compel-ling support for the belief that they canhave an important role in famine relief.That does not mean that they will al-

ways be the most cost effective option;if the intervention is delayed too longthen the poorest may be least fit towork; and under certain conditions for-gone incomes and non-wage costs maymean that such schemes will only domi-nate informationally feasible alterna-tives if the assets created have suffi-cient value to the poor, such as byhelping to prevent the deterioration ofrural infrastructure and other assetsduring the famine (Ravallion and Datt1995). There have been remarkably fewrigorous evaluations of the gains to thepoor from specific safety net interven-tions in poor countries.

Recent literature has also emphasizedthe limitations of analyzing relief policyoptions in partial equilibrium terms; anincome transfer is often assumed toleave all prices and other sources of in-come unchanged. This is unlikely tohold when considering widespreadtransfers of income in market econo-mies. In distorted economies, transferscan have significant multiplier effectson the incomes of the poor. For exam-ple, in a dualistic economy with mobil-ity between a rural sector and an urbansector with unemployment, the multi-plier effect from transfers to the ruralpoor will be large when the wage elas-

ticity of demand for labor in agricultureis low (Ravallion 1990b). There appears

to be potential for exploiting such gen-eral equilibrium multiplier effects inpolicy design aimed at achieving the

most cost effective relief aid. For exam-ple, relief work is generally confined toalleviating poverty from rural unem-ployment in lean periods. This may re-duce the transfer benefit to the poor,once market responses are considered.Indeed, under this restriction, the pol-icy cannot be expected to achieve themost cost effective aid-in the sense ofmaximizing the recipient's transferbenefit for a given outlay on the policy.

5. Stabilization Policies

Public buffer stocks have been apopular stabilization policy. Stocks areraised (lowered) when aggregate foodavailability is high (low). The size of thestock needed, and hence the cost of thescheme, will depend on the degree ofstabilization desired and how privatestorage decisions respond (Newbery

and Stiglitz 1981). By stabilizing prices,such policies reduce incentives for pri-vate speculation. And so private storagewill be displaced. The extent of this willdepend on the way foodgrain marketswork. At one extreme, buffer stocks willgenerally be ineffective as price stabi-lizers in competitive markets in whichagents hold rational expectations ofprice movements. But plausible marketimperfections may or may not enhance

the case for stabilization through a con-ventional buffer stock policy. For exam-ple, the specific informational ineffi-ciencies suggested by the evidence forBangladesh imply clear limits to the useof this type of policy due to the poten-tial for speculative attack during fam-ines.47 The lesson is clear: it is crucial

47 The usual policy of selling from buffer stocksat times of high prices, may fail at sufficiently lowlevels of public storage. Indeed, at times of high

prices, if public stocks are believed to be so low asto generate highly inflated expectations of future








6. Tradeoffs?

Comprehensive poverty reductionstrategies for low-income countries

rightly emphasize the role played by so-cial safety nets even in growing econo-mies.51 Policy makers may be very will-ing to forgo other objectives ofdevelopment to assure greater securityagainst famines. However, it is far fromobvious that effective policies againstfamine entail a positive cost to othercommonly espoused objectives of devel-opment, including economic growth andenvironmental protection.

The literature on famines reviewedhere has suggested that failures of bothmarket and non-market institutions lieat the heart of famine causation; so itcan be argued that famines can be ame-liorated by longer-term developmentpolicies which strengthen the social andeconomic institutions (both governmen-tal and non-governmental) which helpprotect poor people from economy-wide shocks.

Evidence in the famines literatureand elsewhere also suggests that an ef-fective social safety net for protectingpoor households from severe shocks isconsistent with longer-term goals ofeconomic growth and environmentalprotection. Arguments that greater cur-rent security against famine can pro-mote growth have come from two mainsources. First, it can be argued thatcredit constraints make it difficult forsurvivors to restore their productive as-sets after the famine, entailing irre-versible lasting damage to livelihoods(Jodha 1975; Agarwal 1991; Osmani1996). Yet without an adequate safetynet some will be forced into drawingdown their productive assets. There issome evidence that famines have re-

sulted in an increase in wealth inequal-ity, through increased concentration ofland ownership (Azizur Khan 1977;Alamgir 1980; both on Bangladesh) andlivestock in pastoral regions of Africa(Osmani 1996). Thus, after the faminethe economy will have a higher numberof credit-constrained households whoare unable to take up productive invest-ments in physical and human capital.There is some evidence from cross-country studies that higher initial in-equality entails lower subsequentgrowth rates (see the Michael Bruno,

Ravallion, and Lyn Squire 1995 survey).Putting these pieces together, it canbe conjectured that temporary faminescan create more permanent distri-butional shifts which entail lower longterm growth rates. By the same reason-ing, a safety net which can prevent thedepletion of productive assets by poorpeople will entail a higher longer termrate of both growth and poverty reduc-tion.

The second source of evidence comesfrom normal (non-famine) times in set-tings like those in which famines haveoccurred. There has been recent com-pelling evidence from microecono-metric investigations that exposure touninsured risk influences poor peoples'investment decisions and, hence, pros-pects of escaping poverty (Rosenzweigand Binswanger 1993; Rosenzweig andWolpin 1993; Jonathan Morduch 1993,1994; Chaudhuri 1994). For example,even though expected income may behigher from a new seed variety, extraexposure to uninsured risk may preventadoption.

While the above arguments and em-pirical results are suggestive, more re-search is needed on the ways in whichuninsured risks and borrowing con-straints handicap the longer escapefrom poverty. What we seem to lackmost at present is a credible assessment

51 See, for example, World Bank (1990, ch. 6),Lipton and Ravallion (1995), and Binswanger andPierre Landell-Mills (1995).








of the likely dynamic gains from safety-net interventions, so as to better informpublic choices about how much to in-vest in such policies. Even if we foundno longer term gains, we would stillwant to prevent famines. However, it isnot implausible that a number of thepolicies discussed above for achievingbetter protection for poor people-aspart of a famine relief strategy-will en-hance the future prospects of escapingpoverty. Whether that argument is rightin practice, and what quantitative gainscan be expected from specific interven-

tions, will of course depend on how ef-fective the specific interventions are inprotecting poor people, as well as theextent of the dynamic gains from extrainsurance. Further research on thistopic could have important implicationsfor assessments of the case for safety-net policies.

There are potential synergies withother aspects of development policy.Improved infrastructure (both better

roads and marketing arrangements) hasalso been identified as a key factor inthe success stories in reducing vulner-ability to famines.52 Having good ruralroads will certainly not guarantee thatthe available food will get to those inneed, or even that it will move in theright direction, but without them theremay be little chance of it doing so. Be-sides promoting food-market integra-tion, there is evidence to suggest thatimproved rural transport allows a diver-sification of rural income sources, re-ducing risk and promoting both farm

and non-farm growth.53 A similar com-ment applies to certain policies for pro-moting agricultural and rural develop-ment, such as reducing the (direct andindirect) taxation of agriculture, andcertain human resource interventions.54There probably will be some changes inpriorities if one aims to prevent fam-ines. But because past policies do notappear to have been growth maximiz-ing, the growth cost of the requiredchange in policies need not even bepositive.

What about the environmental conse-

quences? It has been argued that vul-nerability to famines is increased by thelack of well-defined property rights,and the failures of other means of assur-ing quasi-cooperative solutions to theproblem of allocating common-propertyresources (Dasgupta and Karl Meler1995). In pastoral economies, overgraz-ing can result, and the open-accessproblem often also underlies the pro-cess of deforestation. When these pro-

cesses have continued over a long pe-riod, the ecological system has little

52Walter and Schofield (1989) argue that in En-gland by the late seventeenth century improvedfarming methods and better spatial integration offood markets due to improved transport and mar-keting had resulted in higher farm yields and morediversified food production in the country as awhole, and greater protection from localized andcrop-specific harvest failures. In parts of continen-tal Europe, however, progress on all these fronts

had been slower, and vulnerability to famines per-sisted.

53On the positive effects of infrastructure onagricultural output in India see Binswanger,Khandker, and Rosenzweig (1993). Datt andRavallion (forthcoming) find that states of Indiawith better rural infrastructure saw higher longer-term rates of rural poverty reduction, controllingfor other differences in initial conditions and ex-ogenous time-varying factors. Also see McAlpin(1983) who argues more informally that the dimin-ished vulnerability to droughts in India during thefirst 20 years of this century was due in part torural infrastructure development (including theexpansion of the railways). (Though also see thecomments by Appadurai 1984; and Rangasami1990.) A not dissimilar story can be told about In-donesian economic development since the 1970s;Indonesia was probably as vulnerable to faminesas Bangladesh in the 1960s, but this has changedgreatly, and rural development (particularly inJava) appears to have had an important role(Ravallion 1995).

54On the taxation of agriculture see Krueger,Schiff, and Valdes (1988). On human resourcepolicies see Behrman's (1990) survey. For more

general discussions see World Bank (1990), Das-gupta (1993), and Lipton and Ravallion (1995).








resistance. If individual and social cop-ing strategies also fail then the peoplewho depend on that ecological system

can be extremely vulnerable to even amodest drought. The ensuant faminecan itself entail ecological upheaval; forexample, mass migration can have dev-astating environmental effects. Thesearguments suggest that environmental-protection policies can be viewed ascomplementary to reducing famine vul-nerability in the longer term.

There are a number of other poten-tial synergies between policies to pro-

tect from extremes such as famines, andlonger-term progress in other aspects ofdevelopment. Some of the more effec-tive social safety nets for alleviatingpoverty in non-famine times evolvedout of experience in famines. Examplesinclude the "Employment GuaranteeScheme" in the Indian state of Ma-harashtra which evolved out of the fam-ine relief efforts in the early 1970s, andthe targeted food distribution and cashtransfers found in Botswana (Dreze andSen 1989). Recent discussions of disas-ter-relief policy have also emphasizedthe synergy with longer-term develop-ment goals (Mary Anderson and PeterWoodrow 1989).

With limited resources, many sharptradeoffs will undoubtedly be faced indetailed policy choices. But it shouldnot be presumed that effective policiesfor reducing vulnerability to famineswill come at a cost to other develop-ment objectives. While a strict separa-tion of "relief' and "development" isclearly untenable, a far deeper under-standing is still needed of the extentand nature of the synergies. Public ac-tion to prevent famines need not waitfor the results of that research. But it isresearch that could matter greatly to aproper assessment of the extent of theworld's

publicresources that should be

devoted to the task.

V. Conclusions

Famines have defied simple explana-

tions as well as geographic boundaries.They have happened under both social-ist and capitalist economic systems.They have happened with and withoutdeclines in food supply, and with andwithout wars or unusual political or so-cial instability. An objective and holisticview of the determinants of individualsurvival prospects in economies understress offers the greatest hope of under-standing why famines happen, and of

usefully informing domestic and inter-national public action on how best toavoid or relieve them. That is the maininsight that economic analysis can offerto replace the heavy reliance in the paston often doubtful "single-cause" expla-nations for famines.

Economic analysis can help under-stand famines. But economists can alsolearn from famines. While they are dra-matic events, they can start quite un-

dramatically. Under certain conditions,the threat of mass starvation canemerge from seemingly small shocks toan economy, or from a steady-evenslow-decline in average living stan-dards. And similar, even large, shocks insimilar settings can have very differentconsequences. Market and non-market(including governmental) institutionswhich work quite adequately, thoughnot perfectly, in normal times can easilyturn even a moderate aggregate shockinto a devastating blow to some poorpeople.

Famine can thus be viewed as a tragicmagnification of normal market andgovernmental failures. The nonlineari-ties, cumulative effects, and aperiodicbehavior, that can transform a shockinto mass starvation appear to be intrin-sic features of quite normal econo-mies-rather than peculiar features ofhighly distorted or badly damaged








economies. They are there always, butnormally hidden from view. And theycan surface in any number of ways.

While famines do not lend themselvesto easy prediction, there are some com-mon elements. Entitlement failure isone. While famines have happened withand without crop failures, every famineappears to have entailed a collapse inthe legal command over food of at leastsome subgroups within society, eitherthrough a loss of endowment or a con-traction in the amount of food that canbe acquired from given endowments.

Looking deeper into the causes of en-titlement failure we also see some com-mon elements albeit at a more aggre-gate level. Weak social and physicalinfrastructure, weak unprepared gov-ernment, and a relatively closed politi-cal regime, all enhance vulnerability tofamine. Arguably the same factors con-strain longer-term poverty reduction.Recognizing the contingent nature ofthe market and institutional failures

that underlie famines points to implica-tions for both their relief and preven-tion. And it is doubtful that effectivepolicies for preventing famines comeat any significant, or even positive, cost interms of other development objectives.

REFERENCES

ADAM, HUSSEIN M. "Building Capacity in theCountryside," in JOHN FIELD, ed. 1993a, pp.172-204.

ADAMS, ALAYNE. "Food Insecurity in Mali: Ex-

ploring the Role of the Moral Economy," IDSBulletin, Oct. 1993, 24(4), pp. 41-51.AGARWAL, BINA. "Social Security and the Family:

Coping with Seasonality and Calamity in RuralIndia," in Social security in developing coun-tries. Eds.: EHTISHAM AHMAD ET AL. Oxford:Oxford U. Press, 1991, pp. 171-244.

ALAMGIR, MOHIUDDIN. Famine in south Asia: Po-litical economy of mass starvation. Cambridge:Oelgeschlager, Gunn & Hain, 1980.

ALDERMAN, HAROLD. "Saving and EconomicShocks in Rural Pakistan," J. Devel. Econ., Dec.1996, 51(2), pp. 343-66.

AMBIRAJAN, S . Classical political economy andBritish policy in India. Cambridge: CambridgeU. Press, 1978.

ANDERSON, MARY B. AND WOODROW, PETER J.Rising from the ashes: Development strategiesin times of disaster. Boulder, CO: WestviewPress, 1989.

APPADURAI, ARJUN. "How Moral Is South Asia's

Economy? A Review Article," J. Asian Stud.,May 1984, 43(3), pp. 481-97.ARROW, KENNETH J. "Review of 'Poverty and

Famines' by A. K. Sen," The New York Reviewof Books, 1982, 29, pp. 24-26.

ASHTON, BASIL ET AL. "Famine in China, 1958-61," Population Devel. Rev., Dec. 1984, 10(4),pp. 613-45.

BANGLADESH RURAL ADVANCEMENT COMMIT-TEE (BRAC). Peasant perceptions, famine.Dacca: Bangladesh Rural Advancement Com-mittee, 1979.

BARDHAN, PRANAB K. Land, labor and rural pov-erty. New York: Columbia U. Press, 1984.

BASU, KAUSHIK; PATTANAIK, PRASANTA ANDSUZUMURA, KOTARO, eds. Choice, welfare, anddevelopment. Oxford: Oxford U. Press, 1995.

BEHRMAN, JERE R. "The Action of Human Re-sources and Poverty on One Another. What WeHave Yet to Learn." Living Standards Measure-ment, Study.Working Paper 74. Washington,DC: World Bank, 1990.

BEHRMAN, JERE R. AND DEOLALIKAR, ANIL B."Will Developing Country Nutrition Improvewith Income?, J. Polit. Econ., June 1987, 95(3),pp. 492-507,

"Health and Nutrition," in Handbook ofdevelopment economics. Vol. 1. Eds.: HOLLISCHENERY AND T. N. SRINIVASAN. Amsterdam:Elsevier, North-Holland, 1988, ch. 14.

BEHRMAN, JERE AND SRINIVASAN, T. N., eds.Handbook of development economics. Vol. 3.Amsterdam: Elsevier, North-Holland, 1995.

BERRY, LEONARD AND DOWNING, THOMAS E."Drought and Famine in Africa, 1981-86," inJOHN FIELD, ed. 1993a, pp. 35-58.

BESLEY, TIMOTHY AND COATE, STEPHEN. "Work-fare versus Welfare: Incentive Arguments forWork Requirements in Poverty-Alleviation Pro-grams," Amer. Econ. Rev., Mar. 1992, 82(1), pp.249-61.

BHATIA, B. M. Famines in India: A study of someaspects of the economic history of India, 1860-

1964. London: Asia Publishing House, 1963.BIDANI, BENU AND RAVALLION, MARTIN. "De-composing Social Indicators Using Distri-butional Data," J. Econometrics, Mar. 1997,77(1), pp. 125-40.

BIDINGER, PATRICIA D. ET AL. "Consequences ofMid-1980s Drought: Longitudinal Evidencefrom Mahbubnagar," Economic and PoliticalWeekly, Sept. 28 1991, 26(39), pp. A105-14.

BINSWANGER, HANS P.; KHANDKER, SHAHIDURR. AND ROSENZWEIG, MARK R. "How Infra-structure and Financial Institutions Affect Agri-cultural Output and Investment in India," J.Devel. Econ., Aug. 1993, 41(2), pp. 337-66.

BINSWANGER, HANS P. AND LANDELL-MILLS,PIERRE. The World Bank's strategy for reduc-








ing poverty and hunger. A report to the interna-tional community. Washington,, DC: WorldBank, 1995.

BLISS, CHRISTOPHER AND STERN, NICHOLAS."Productivity, Wages and Nutrition, 1: Theory,"J. Devel. Econ., Dec. 1978, 5(4), PP. 331-62.

BOSE, SUGATA. "Starvation Ami st Plenty: theMaking of Famine in Bengal, Honan andTonkin, 1942-45," Modern Asian Studies, Oct.1990, 24(4), pp. 699-727.

BouIS, HOWARTH E. AND HADDAD, LAWRENCEJ. "Are Estimates of Calorie-Income ElasticitiesToo High? A Recalibration of the PlausibleRange," J. Devel. Econ., Oct. 1992, 39(2), pp.333-64.

BOWBRICK, PETER. "The Causes of Famine: aRefutation of Professor Sen's Theory," FoodPolicy, May 1986, 11, pp. 105-32.

BOYCE, JAMES K. Agrarian impasse in Bengal: In-stitutional constraints to technological change.Oxford: Oxford U. Press, 1987.

VON BRAUN, JOACHIM AND KENNEDY, EILEEN,eds. Agricultural commercialization, economicdevelopment, and nutrition. Baltimore: JohnsHopkins U. Press, 1994.

VON BRAUN, JOACHIM; TEKLU T. AND WEBB,PATRICK. Famines in Africa: Causes, mitiga-tion prevention. Washington, DC: InternationalFood Policy Research Institute, 1994.

BRUNO, MICHAEL; RAVALLION, MARTIN ANDSQUIRE, LYN. "Equity and Growth in Develop-ing Countries: Old and New Perspectives onthe Policy Issues." Policy Research Working Pa-

per 1563, World Bank, Washington, DC, 1995.CALDWELL, JOHN C. AND CALDWELL, PAT."Famine in Africa: a Global Perspective," inMortality and society in sub-saharan Africa.Eds.: ETIENNE VAN DE WALLE, GILLES PISON,AND MPEMBELE SALA-DIAKANDA. Oxford:Clarendon Press, 1992, pp. 361-83.

CALDWELL, JOHN C.; REDDY, P. H. ANDCALDWELL, PAT. "Periodic High Risk as aCause of Fertility Decline in a Changing RuralEnvironment: Survival Strategies in the 1980-1983 South Indian Drought," Econ. Devel. Cult.Change, July 1986, 34(4), pp. 677-701.

CARRARO, LUDOVICO. `Undcerstanding Famine: aTheoretical Dynamic Model." DevelopmentStudies Working Paper No. 94, Centro SudiLuca d'Agliano, Terion, Italy, 1996.

CHADHA, BANKIM AND TEJA, RANJIT S. "Macro-economics and Famine." Mimeo, InternationalMonetary Fund, Washington, DC, 1992.

CHAUDHURI, SHUBHAM. "Crop Choice, FertilizerUse, and Credit Constraints: an EmpiricalAnalysis." Mimeo, Department of Economics,Columbia U., 1994.

CHEN, LINCOLN C.; CHOWDHURY, A. K. M. ANDHUFFMAN, S. L. "Anthropometric Assessmentof Energy-Protein Malnutrition and SubsequentRisk of Mortality among Preschool Aged Chil-dren," American J. Clinical Nutrition, 1980, 33,pp. 1836-45.

CLAY, EDWARD. "Famine, Food Security, Poverty

and Public Action," Devel. Policy Rev., 1991, 9,pp. 307-12.

COATE, STEPHEN. "Cash Versus Direct Food Re-lief,"J. Devel. Econ., Apr. 1989, 30(2), pp. 199-224.

COATE, STEPHEN AND RAVALLION, MARTIN."Reciprocity Without Commitment: Charac-terization and Performance of Informal Insur-ance Arrangements," J. Devel. Econ., Feb.1993, 40(1), pp. 1-24.

COLES, JEFFREY L. AND HAMMOND, PETER J."Walrasian Equilibrium Without Survival: Exis-tence, Efficiency and Remedial Policy," inKAUSHIK BASU, PRASANTA PATTANAIK, ANDKOTARO SUZUMURA, eds. 1995.

CORBETT, JANE. "Famine and Household CopingStrategies," World Devel., Sept. 1988, 16(9),pp. 1099-1112.

CROW, BEN. "Warnings of Famine in Bangla-desh," Economic and Political Weekly, Oct. 6,1984, 19, pp. 1754-58.

CURREY, BRUCE. "Coping with Complexity inFood Crisis Management," in Famine as a geo-graphical phenomenon. Eds.: BRUCE CURREYAND GRAEME HUGO. Dordrecht: D. Reidel,1984, pp. 183-202.

. "Is Famine a Discrete Event?" Disasters,June 1992, 16(2), pp. 138-44.

CUTLER, PETER. "Responses to Famine. WhyThey Are Allowed to Happen," 1993 in JOHNFIELD, ed. 1993a, pp. 72-87.

DALRYMPLE, DANA G. "The Soviet Famine of1932-1934," Soviet Studies, Jan. 1964, 15, pp.

250-84.DANDO, WILLIAM A. "Man-Made Famines: SomeGeographical Insights from an ExploratoryStudy of a Millennium of Russian Famines," inFamine: Its causes, effects and management.Ed.: JOHN R. K. ROBSON. New York: Gordon &Breach, 1981, pp. 139-54.

DASGUPTA, PARTHA. An inquiry into well-beingand destitution. Oxford: Oxford U. Press, 1993.

DASGUPTA, PARTHA AND MALER, KARL G. "Pov-erty, Institutions, and the Environmental Re-source-Base," in JERE R. BEHRMAN AND T. N.SRINIVASAN, ds. 1995, ch. 39.

DASGUPTA, PARTHA AND RAY, DEBRAJ. "Inequal-ity as a Determinant of Malnutrition and Unem-ployment: Theory," Econ. J., Dec. 1986,96(384), pp. 1011-34.

DATT, GAURAV AND RAVALLION, MARTIN. "WhyHave Some Indian States Done Better thanOthers at Reducing Rural Poverty?" Economica,forthcoming

DAVIES, SUSANNA. "Are Coping Strategies a CopOut?" IDS Bulletin, 1993, 24, pp. 60-72.

DEATON, ANGUS. "Saving in Developing Coun-tries: Theory and Review," in Proceedings of theWorld Bank annual conference on developmenteconomics. Eds.: STANLEY FISCHER AND DEN-NIS DE TRAY. WASHINGTON, DC: WORLDBANK, 1989, PP. 611-96.

. Understanding consumption. Oxford: Ox-ford U. Press, 1992.








DESAI, MEGHNAD. "The Economics of Famine,"in G. AINSWORTH HARRISON, ed. 1988, pp.107-38.

. "Rice and Fish: Asymmetric Preferencesand

Entitlement Failuresin

Food GrowingEconomies with Non-Food Producers," Europ.J. Polit. Econ., 1989, 5, pp. 429-40.

DEVEREUX, STEPHEN. "Entitlements, Availabilityand Famine: a Revisionist View," Food Policy,1988, 13, pp. 270-82.

. "Goats Before Ploughs: Dilemmas ofHousehold Response Sequencing During FoodShortages," IDS Bulletin, Oct. 1993a, 24(4), pp.52-59.

. Theories of famine. New York: HarvesterWheatsheaf, 1993b.

DIRKS, ROBERT. "Social Response During SevereFood Shortages and Famine," Current Anthro-pology, Feb. 1980, 21(1), pp. 21- 44.

DREZE, JEAN. "Famine Prevention in India,"1990, in JEAN DRtZE AND AMARTYA SEN eds.Vol. 2. 1990, pp. 13-122.

DREtZE, JEAN AND SEN, AMARTYA. Hunger andpublic action. Oxford: Oxford U. Press, 1989.Vs eds. The political economy of hunger:Vols. 1, 2, and 3. Oxford: Oxford U. Press, 1990.

D'SOUZA, FRANCES. "Famine: Social Security andan Analysis of Vulnerability," n G. AINSWORTHHARRISON, ed. 1988, pp. 1-56.

DYSON, TIM. "On the Demography of South AsianFamines. Parts 1 and 2," Population Stud., Mar.& July 1991, 45(1 & 2), pp 5-25 and 279-97.

FIELD, JOHN O., ed. The challeng offamine: Re-

cent experience, lessons learned. West Hartford,CT: Kumarian Press, 1993a.Understanding Famine," 1993b, in JOHN

FIELD, ed. 1993a, pp. 11-29.. "Beyond Refief. Toward Improved Man-

agement of Famines," 1993c, in JOHN FIELD,ed. 1993a, pp. 253-74..

FOGEL, ROBERT W. "Second Thoughts on theEuropean Escape from Hunger: Famines,Chronic Malnutrition, and Mortality Rates," inNutrition and poverty. Ed.: S. R. OSMANI. Ox-ford: Oxford U. Press, 1992, pp. 243-86.

FOSTER, ANDREw D. "Prices, Credit Markets andChild Growth in Low-income Rural Areas,"Econ. J., May 1995, 105(430), pp. 551-70.

GALLOWAY, . R. "Annual Variations n Deaths byAge, Deaths by Cause, Prices, and Weather inLondon 1670-1830," Population Stud., Nov.1985, 39(3), pp. 487-505.

GERSOVITZ, MARK. "Savings and Nutrition at LowIncomes," J. Polit. Econ., Oct. 1983, 91(5), pp.841-55.

GHOSE, AJIT K. "Food Supply and Starvation: aStudy of Famines with Reference to the IndianSub-Continent," Oxford Econ. Pap., July 1982,34(2), pp. 368-89.

GILL, PETER. A year in the death of Africa: Poli-tics, bureaucracy and thefamine. London: Pala-din, 1986.

GLOMM, GERHARD AND PALUMBO, MICHAEL G."Optimal Intertemporal Consumption Deci-

sions under the Threat of Starvation," J. Devel.Econ., Dec. 1993, 42(2), pp. 271-91.

GORE, CHARLES. "Entitlement Relations and 'Un-ruly' Social Practices: A Comment [on the Work

of Amartya Sen]," J. Devel. Studies, Apr. 1993,29(3), pp. 429-60.GOVERNMENT OF INDIA. The drought of 1987:

Response and management. Vol. 1. National ef-forts. New Delhi, Ministry of Agriculture, Gov-ernment of India, 1989.

GREENOUGH, PAUL R. Prosperity and misery inmodern Bengal: the Famine of 1943-1944. NewYork: Oxford U. Press, 1982.

GROSSMAN, SANFORD J. AND STIGLITZ, JOSEPHE. "On the Impossibility of Informationally Ef-ficient Markets," Amer. Econ. Rev. June 1980,70(3), pp. 393-408.

HARRISON, G. AINSWORTH, ed. Famine. Oxford:Oxford U. Press, 1988.

HEYWOOD, PETER. "The Functional Significanceof Undernutrition Growth and Prospective Riskof Death in the Highlands of Papua NewGuinea," J. Food and Nutrition, 1982, 39, pp.13-19.

HUDDLESTON BARBARA ET AL. International fi-nance for food security. Baltimore: Johns Hop-kins U. Press, 1984.

ISLAM, RIZWANUL. "Rural Unemployment andUnderemployment: a Review," in Bangladesh:Selected issues in employment and development.Eds: RIZWANUL ISLAM AND M. MUQTADA.New Delhi: International Labour Organisation,1986, pp. 11-20.

JANSSON, KURT; HARRIS, MICHAEL AND PEN-ROSE, ANGELA. The Ethiopian famine. London:Zed Books, 1987.

JODHA, NARPAT S. "Famine and Famine Policies:Some Empirical Evidence," Economic and Po-litical Weekly, Oct. 11, 1975, 10(4), pp. 1609-17.

JONES, STEPHEN. "Food Security Reserve Policyin Ethiopia: A Case Study of Experience andImplications," Disasters, June 1994, 18(2), pp.140-51.

KEEN, DAVID. "Targeting Emergency Food Aid:the Case of Dafur in 1985," in SIMON MAX-WELL, ed. 1991.

KEEN, MICHAEL C. "Needs and Targeting," Econ.J., Jan. 1992, 102(410), pp. 67-79.

KELLY, MARION. "Entitlements, Coping Mecha-nisms and Indicators of Access to Food: WolloRegion, Ethiopia 1987-88," Disasters, Dec.1992, 16(4), pp. 322-38.

KHAN, AZIZUR R. "Poverty and Inequality in RuralBangladesh," n Poverty and landlessness in ru-ral Asia. Geneva: International Labour Orga-nisation, 1977, pp. 137-60.

KHAN, M. MAHMUD. "Market-Based Early Warn-ing Indicators of Famine for the PastoralHouseholds of the Sahel," World Devel., Feb.1994, 22(2), pp. 189-99.

KLOOS, HELMUT AND LINDTJ0RN, BERNDT."Famine and Malnutrition," in The ecology ofhealth and disease in Ethiopia. Eds.: H. KLOOS








AND Z. AHMED ZEIN. Boulder, CO: WestviewPress, 1993.

KRUEGER, ANNE 0.; SCHIFF, MAURICE ANDVALDEtS, ALBERTO. "Agricultural Incentives inDeveloping Countries: Measuring the Effect ofSectoral and Economy-Wide Policies," WorldBank Econ. Rev., Sept. 1988, 2(3), pp. 255-71.

KUMAR, B. GOPALAKRISHNA. "Ethiopian Famines1973-1985: A Case Study," in JEAN DRtZE AND

AMARTYA SEN, eds., 1990, pp. 173-216.LAPPE, FRANCES MOORE AND COLLINS, JOSEPH.

Food first: Beyond the myth of scarcity. Boston:Houghton Mifflin, 1977.

LIN, JUSTIN Y. AND YANG, DENNIS T. "FoodAvailability, Entitlements and the ChineseFamine of 1959-1961." Research Paper inAsian/Pacific Studies, Asian/Pacific Studies In-stitute, Duke U., 1995.

LIPTON, MICHAEL. Why poor people stay poor:Urban bias and world development. London:Temple Smith, 1977.

. "Poverty, Undernutrition and Hunger."World Bank Staff Working Paper 597, WorldBank, Washington, DC, 1983.

LIPTON, MICHAEL AND RAVALLION, MARTIN."Poverty and Policy," in JERE BEHRMAN ANDT. N. SRINIVASAN 1995, pp. 2551-2657.

LOCKE, CHRISTOPHER G., AND AHMADI-ESFA-HANI, FREDOUN Z. "Famine Analysis: a Studyof Entitlements in Sudan, 1984-85," Econ.Devel. Cult. Change, Jan. 1993, 41(2), pp. 363-76.

MACRAE, JOANNA AND Zwi, ANTHONY B. "Foodas an Instrument of War in Contemporary Afri-can Famines: a Review of the Evidence," Disas-ters, Dec. 1992,16(4), pp. 299-321.

MAXWELL, SIMON, ed. To cure all hunger. Foodpolicy andfood security in Sudan. London: In-termediate Technology Publications, 1991.

MAXWELL, SIMON AND TEMPLER, G. "TheMonetization of Project and Emergency FoodAid: Project-Level Efficiency First " Food Pol-icy, Feb. 1994, pp. 9-15.

MCALPIN, MICHELLE B. Subject to famine: Foodcrises and economic change in western India,1860-1920. Princeton: Princeton U. Press,1983.

MCGREGOR, PAT. "A Model of Crisis in a PeasantEconomy," Oxford Econ. Pap., Oct. 1990, 4, pp.793- 811.

. "Famine: a Simple General EquilibriumModel." Mimeo, School of Public Po icy, Eco-nomics and Law, U. of Ulster, Northern Ire-land, 1996.

MIRRLEES, JAMES. "A Pure Theory of Underde-veloped Economies," in Agriculture in develop-ment theory. Ed.: LLOYD G. REYNOLDS. NewHaven: Yale U. Press, 1975, pp. 84-106.

MORDUCH, JONATHAN. "Risk, Production andSaving: Theory and Evidence from IndianHouseholds." Mimeo, Harvard U., 1993.

. "Poverty and Vulnerability," Amer. Econ.Rev., May 1994, 84(2), pp. 221-25.NEWBERY, DAVID M. G. AND STIGLITZ, JOSEPH

E. The theory of commodity price stabilization.Oxford: Oxford U. Press, 1981.

NEWMAN, LUCILE F., ed. Hunger in history: Foodshortage, poverty, and deprivation. Oxford:Blackwell, 1990.

NOLAN, PETER. "The Causation and Prevention ofFamines: a Critique of A. K. Sen," J. PeasantStudies, Oct. 1993, 21(1), pp. 1-28.

NOZICK, ROBERT. Anarchy, state, and utopia. Ox-ford: Blackwell, 1974.

OSMANI, SIDDIQ R. "The Food Problems of Ban-gladesh," In JEAN DRtZE AND AMARTYA SEN,eds. 1990.

. "Comments [on Alex De Waal's Re-assess-ment of Entitlement Theory in the Light of Re-cent Famines in Africa]," Development andChange, July 1991, 22(2), pp. 587-96.

. "The Entitlement Approach to Famine: an

Assessment," in KAUSHIK BASU, PRASANTAPATTANAIK, AND KOTARO SUZUMURA, eds.1995.

. "Famine, Demography and Endemic Pov-erty," J. Int. Devel., Sept./Oct. 1996, 8(5), pp.597-623.

OXFAM HEALTH UNIT. Oxfam's practical guide toselective feeding programmes. Oxford: Oxfam,1984.

PAXSON, CHRISTINA H. "Consumption and In-come Seasonality in Thailand," J. Polit. Econ.,Feb. 1993, 101(1), pp. 39-72.

PELLETIER, D. L. ET AL. "The Effects of Malnu-trition on Child Mortality in Developing Coun-tries," Bulletin of the World Health Organiza-tion, 1995, 73, pp. 443-48.

PITT, MARK M.; ROSENZWEIG, MARK R. ANDGIBBONS, DONNA M. "The Determinants andConsequences of the Placement of GovernmentPrograms n Indonesia," in Public spending andthe poor: Theory and evidence. Eds.: D. VAN DEWALLE AND K. NEAD. Baltimore: Johns Hop-kins U. Press, 1995.

PITT, MARK M.; ROSENZWEIG, MARK R. ANDHASSAM, MD. NAZMUL. "Productivity, Health,and Inequality in the Intrahousehold Distri-bution of Food in Low-Income Countries,"Amer. Econ. Rev., Dec. 1990, 80(5), pp. 1139-56.

PLATTEAU, JEAN-PHILIPPE. "Traditional Systemsof Social Security and Hunger Insurance: PastAchievements and Modern Challenges," in So-cial security in developing countries. Eds.:EHTISHAM AHMAD ET AL. Oxford: Oxford U.Press, 1991, pp. 112-70.

POST, JOHN D. "Nutritional Status and Mortalityin Eighteenth Century Europe," in LUCILE F.NEWMAN, ed. 1990, pp. 241-80.

QUDDUS, MUNIR AND BECKER, CHARLES."Speculative Price Bubbles in the Rice Marketand the 1974 Bangladesh Famine." Mimeo, De-partment of Economics, U. of Southern Indi-ana, 1994.

QUDDUS, MUNIR AND RASHID, SALIM. "SomeMyths about Disaster Management in Bangla-desh," Asian Affairs, 1991, 18, pp. 51-58.








RAM, N. "An Independent Press and Anti-HungerStrategies: the Indian Experience," in JEANDREZE AND AMARTYA SEN, eds. Vol. 1. 1990,pp. 146-90.

RANGASAMI, AMRITA. "'Failure of Exchange Enti-tlements' Theory of Famine: a Response," Eco-nomic and Political Weekly, Oct. 12, 1985,20(41), pp. 1947-52; Oct. 19, 1985 20(42), pp.1797-1801.

. "The McAlpin Capers," Economic and Po-litical Weekly, Sept. 1, 1990.

RASHID, SALIM. "The Policy of Laissez-Faire Dur-ing Scarcities," Econ. J., Sept. 1980, 90(359),pp. 493-503.

RAVALLION, MARTIN. Markets and famines. Ox-ford: Oxford U. Press, 1987.

. "Income Effects on Undernutrition,"Econ. Devel. Cult. Change, Apr. 1990a, 38(3),pp. 489-515.

. "Market Responses to Anti-hunger Poli-cies: Wages, Prices, and Employment," 1990b,in JEAN DREZE AND AMARTYA EN, eds. 1990,pp. 241-78.

. "Reaching the Rural Poor Through PublicEmployment: Arguments, Experience and Les-sons from South Asia," World Bank ResearchObserver, 1991, 6, pp. 153-75.

"Household Vulnerability to AggregateShocks: Differing Fortunes of the Poor in Ban-gladesh and Indonesia," in KAUSHIK BASU,PRASANTA PATTANAIK, AND KOTARO SUZU-MURA, eds. 1995.

RAVALLION, MARTIN AND CHAUDHURI, SHUB-

HAM. "Risk and Insurance in Village India: aComment," Econometrica, Jan. 1997, 65, pp.171-84.

RAVALLION, MARTIN AND DATT, GAURAV. "IsTargeting Through a Work RequirementEfficient? Some Evidence for Rural India," inPublic spending and the poor: Theory andevidence. Eds.: D. VAN DE WALLE AND K.NEAD. Baltimore: Johns Hopkins U. Press,1995.

RAZZAQUE, ABDUR ET AL. "Sustained Effects ofthe 1974-5 Famine on Infant and Child Mortal-ity in a Rural Area of Bangladesh," PopulationStud., Mar. 1990, 44(1), pp. 145-54.

REUTLINGER, SHLOMO. "Policies for Food Secu-rity in Food-importing Developing Countries,"in Food security: Theory, policy and perspec-tives from Asia and the Pacific Rim. Eds.: A. H.CHISHOLM AND RODNEY TYERS. Lexington,MA: Lexington Books, 1982.

"Review of Sen (1981)," Econ. Devel.Cult. Change, 1984, 32, pp. 881-86.

. "Efficient Alleviation of Poverty and Hun-ger: a New International Assistance Facility,"Food Policy, Feb. 1988, pp. 56-66.

RISKIN, CARL. "Feeding China: the Experiencesince 1949," in JEAN DREZE AND AMARTYASEN, eds. Vol. 3, 1990.

RIVERS, JOHN P. W. "The Nutritional Biology ofFamine," in G. AINSWORTH HARRISON, ed.1988.

ROSENZWEIG, MARK R. AND BINSWANGER, HANSP. "Wealth, Weather Risk and the Compositionand Profitability of Agricultural Investments,"Econ. J., Jan. 1993,103(416), pp. 56-78.

ROSENZWEIG,MARK

R.AND

WOLPIN,KENNETH

I. "Credit Market Constraints, ConsumptionSmoothing, and the Accumulation of DurableProduction Assets in Low-Income Countries:Investments in Bullocks in India," J. Polit.Econ., Apr. 1993,101(2), pp. 223-44.

SCRIMSHAW, NEVIN. "World Nutritional Prob-lems," in Hunger in history. Food shortage,poverty, and deprivation. Ed. LUCILE F. NEW-MAN. Oxford: Blackwell, 1990, pp. 353-73.

SEAMAN, JOHN. "Famine Mortality n Africa," IDSBulletin, 1993, 24, pp. 27-32.

SEAMAN, JOHN AND RIVERS, JOHN. "Strategies forthe Distribution of Relief Food," J. Royal Sta-tistical Society A, 1988, 151, pp. 464-72.

SEN, AMARTYA. Poverty andfamines: An essay onentitlement and deprivation. Oxford: Oxford U.Press, 1981.

. "Development: Which Way Now?" Econ.J., Dec. 1983, 93(372), pp. 742-62.

. Commodities and capabilities. Amster-dam: North-Holland, 1985.

. "Food, Economics, and Entitlements," inJEAN DREZE AND AMARTYA SEN, eds. Vol. 1.1990, pp. 34-52.

SHEPHERD, JACK. "Some Tragic Errors. AmericanPolicy and the Ethiopian Famine, 1981-85,"1990 in JOHN FIELD, ed. 1993a.

SMITH, ADAM. An inquiry into the nature and

causes of the wealth of nations. London:Methuen, [1776] 1961.SOLOW, ROBERT. "How to Stop Hunger," New

York Review of Books, Dec. 5, 1991.SRINIVASAN, T. N. "Review of Sen's 'Poverty and

Famines," Amer. J. Agr. Econ., 1983, 65, pp.200-01.

STEELE, J. "Millionaire 'Grain Barons' HoardingSupplies in Sudan," The Guardian Weekly, Oct.20, 1985, 133, pp. 8.

STEWART, FRANCES. "Review of Sen Poverty andFamines (1981)," Disasters, 1982, 6(2), 1982,pp. 145-46.

STEWART, . F. "The Ecology of Famine," in G.AINSWORTH HARRISON, ed. 1988, pp. 139-61.

STIGLITZ, JOSEPH E. "The Efficiency Wage Hy-pothesis, Surplus Labour and the Distributionof Income in LDCs," Oxford Econ. Pap., July1976, 28(2), pp. 185-207.

. "Economic Organization, Information,and Development," in Handbook of develop-ment economics. Vol. 1. Eds.: HOLLIS CHEN-ERY AND T. N. SRINIVASAN. Amsterdam: El-sevier, North-Holland, 1988, ch. 5.

STRAUSS, JOHN AND THOMAS, DUNCAN. "Health,Nutrition and Economic Development." Work-ing Paper 95-23, Labor and Population Pro-gram. Santa Monica, CA: RAND, 1995.

STRYKER, J. DIRCK. "Drought and Famine in thePastoral Zone of the Sahel: the 1984 Drought inNiger," 1993 in JOHN FIELD, ed. 1993a.








SUBRAMANIAN, SHANKAR AND DEATON, ANGUS."The Demand for Food and Calories," J. Polit.Econ., Feb. 1996,104(1), pp. 133-62.

SWAMY, ANAND. "A Simple Test of the Nutrition-based Efficiency Wage Model," J. Devel. Econ.,forthcoming.

SWIFT, J. "Understanding and Preventing Famineand Famine Mortality," IDS Bulletin, 1993,24(4), pp. 1-16.

TAYLOR, . E. "Synergy among Mass Infections,Famines, and Poverty," in Hunger and history:The impact of changing food production andconsumption patterns on society. Eds.: ROBERTI. ROTBERG AND THEODORE K. RABB. Cam-bridge: Cambridge U. Press, 1985.

TORRY, W. I. "Famine Early Warning Systems:The Need for an Anthropological Dimension,"Human Organization, 1988, 47, pp. 273-81.

TOWNSEND, ROBERT M. "Risk and Insurance inVillage India," Econometrica, May 1994, 62(3),pp. 539-91.

ULLAH, AMAN AND RoY, NILANJANA. "Parametricand Nonparametric Analysis of Panel DataModels," in Handbook of applied economic sta-tistics. Eds.: DAVID GILES AND AMAN ULLAH.New York: Marcel Dekkar, forthcoming.

ULRICH, PAUL. "Using Market Prices as a Guideto Prevent, or Mitigate Famine in PastoralEconomies: A Case Study from Sudan, 1989-90," 1993 in JOHN FIELD, ed. 1993a, pp. 239-52.

VAUGHAN, MEGAN. The story of an Africanfamine: Gender and famine in twentieth-century Malawi. Cambridge: Cambridge U.Press, 1987.

DE WAAL, ALEXANDER. Famine that kills. Darfur,Sudan, 1984-1985. Oxford: Oxford U. Press,1989.

. "War and Famine in Africa," IDS Bulletin,1993, 24, pp. 33-40.

WALTER, JOHN AND SCHOFIELD, ROGER. "Fam-ine, Disease and Crisis Mortality in Early Mod-

ern Society," in Famine, disease and the socialorder in early modern society. Eds.: JOHN WAL-TER AND ROGER SCHOFIELD. Cambridge:Cambridge U. Press, 1989, pp. 1-73.

WATKINS, SUSAN C. AND MENKEN, JANE. "Fam-ine in Historical Perspective," Population De-vel. Rev., Dec. 1985, 11(4), pp. 647-75.

WEBB, PATRICK; VON BRAUN, JOACHIM AND YO-HANNES, YISEHAC. "Famine in Ethiopia: PolicyImplications of Coping with Failure at Nationaland Household Levels." Research Report 92.Washington, DC: International Food Policy Re-search Institute, 1992.

WEBB, PATRICK AND REARDON, THOMAS.

"Drought Impact and Household Response inEast and West Africa," Quart. J. Int. Agri.,1992, 31(3), pp. 230-46.

WOODHAM SMITH, CECIL. The great hunger: Ire-land 1845-1849. New York: Harper and Row,1962.

WORLD BANK. Poverty and hunger. Issues and op-tions for food security in developing countries.Washington, DC: World Bank, 1986.

. World development report. Poverty. NewYork: Oxford U. Press, 1990.

WRIGLEY, EDWARD A. AND SCHOFIELD, ROGERS.. The population history of England. London:Edward Arnold, 1981.

YOUNG, HELEN. Food scarcity and famine: As-sessment and response. Oxfam Practical HealthGuide No. 7, Oxfam, Oxford, 1992.

YOUNG, HELEN AND JASPERS, SUSANNE. "Nutri-tion, Disease and Death in Times of Famine,"Disasters, June 1995, 19(2), pp. 94-109.

famine and economics

Documents