reading review

World Development 105 (2018) 144–155

Contents lists available at ScienceDirect

World Development

journal homepage: www.elsevier.com/locate/worlddev

Famine systems: A new model for understanding the development of famines

https://doi.org/10.1016/j.worlddev.2017.12.028 0305-750X/� 2017 Elsevier Ltd. All rights reserved.

E-mail address: [email protected]

1 For the purposes of this paper, famine is defined using the criteri Integrated Food Security Phase Classification (IPC): 20 percent of househol area experiencing famine conditions; acute malnutrition exceeding 30 perc crude death rates above 2/10,000/day or under 5 death rates above 4/10,000/ 2017).

Paul Howe Afghanistan Country Office, World Food Program, Green Village, Off Jalabad Road, Kabul, Afghanistan

a r t i c l e i n f o

Article history: Accepted 20 December 2017 Available online 3 February 2018

Keywords: Famine Food security Nutrition Systems

a b s t r a c t

Humans have experienced famines throughout their history. Even today, the world faces the prospect of several of these crises occurring simultaneously. Yet despite their persistence, there is no agreed model for the development of famines, making it difficult to detect their emergence and to prevent their occur- rence. Examining a diverse range of historical and contemporary crises, this paper argues that the evolu- tion of famines can be identified by a set of recognizable elements: pressure, hold, self-reinforcing dynamics, famine system, and rebalancing. It suggests that severe pressure on a population, when held in place for sufficient time, leads to self-reinforcing dynamics that can eventually organize into a famine system that rapidly causes high levels of mortality, until it re-balances and collapses. It contends that this famine systems model can provide analytical insight into the development of most famines and can potentially be used to better identify and respond to these crises in the future.

� 2017 Elsevier Ltd. All rights reserved.

1. Introduction

Over time, specialists in famine have developed crucial insights into the formation and evolution of these crises – from descriptions of their basic stages to more detailed analyses of the sequences of coping strategies employed. However, in part due to the inherent complexity of the social, economic, political, and environmental dynamics involved, the insights have never been assembled into a widely agreed model for the development of famines. Yet the continued persistence of famine, including the recent concern over its potential occurrence in Nigeria, South Sudan, Yemen, and Soma- lia, makes finding such a model an urgent task.

This paper makes an initial attempt to fill this gap by proposing an analytical model consisting of five elements that seek to provide insight into the development of famines: pressure, hold, self- reinforcing dynamics, famine system, and rebalancing. Severe pressure on a population, when held in place for sufficient time, leads to self-reinforcing dynamics that can eventually organize into a famine system that rapidly causes high levels of mortality, until it re-balances and collapses. Although these elements take different forms depending on the circumstances, they appear to be applicable to both historical and contemporary crises. A better

understanding of these elements may assist with the detection and response to famines1 in the future.

The paper is divided into five sections. Following this introduc- tion, the second section reviews the literature that informs our cur- rent understanding of the evolution of famines, describing the wealth of existing insights but also pointing out some of the cur- rent limitations. The third section identifies the basic concepts of the famine systems model. The fourth section tests these concepts by applying them to several historical and contemporary famine situations to see if they can help provide insight into the develop- ment of the crises. The final section identifies possible lessons for early detection and response and makes suggestions for the way forward.

2. Review of relevant literature

Broadly speaking, three competing views have emerged to char- acterize the development of a famine, describing it as an event, a process, or a composite of a process and event (Howe, 2003). For many years, famine was understood as an event: a sudden rise in

a of the ds in the ent; and day (IPC,

P. Howe / World Development 105 (2018) 144–155 145

malnutrition and mortality, often attributed to a food ‘shortage’ or ‘scarcity’ (cf. Dando, 1980; Glantz, 1987; Ravallion, 1997). In Pov- erty and Famines, Amartya Sen (1981) challenged part of this for- mulation, arguing that it was not food shortages but an inability to access food through a breakdown of entitlements that led to famines. However, he seemed to affirm the sudden nature of the crisis, suggesting that a person could be ‘‘plunged into starvation” and that starvation arose from ‘‘a sudden collapse in the level of food consumption” (Sen, 1981, pp. 40–41).2

A number of academics criticized this view of famine, arguing that it decontextualized and depoliticized the crises, by separating the event from its causes. A new set of researchers suggested, instead, that famine should be viewed as a process (Walker, 1989). The precise interpretation of the process took various forms, but most posited the existence of various stages. Some took a longer-term view of the stages. Keen (1994), for example, analysed the historical, underlying, and immediate processes in the 1992 famine in what was then southern Sudan. Others focused mainly on the more immediate stages of the crisis. Rangasami (1985) sug- gested that the famine process consists of three periods: dearth, famishment, and morbidity. Still others contributed different mod- els that illuminated new dimensions of these crises, such as health- related mortality (de Waal, 1989), the relationship to nutrition (Young & Jaspars, 1995), and the process in conflict settings (Hendrickson, Armon, & Mearns, 1998).

These models were, in turn, critiqued because while they offered insight into the process that led up to the event, they failed to recognize the distinctive, time-bound nature of a famine (Rubin, 2016). By blurring the temporal dimensions of the crisis, they cre- ated confusion about its duration and made it difficult to meaning- fully analyse them. Partly as a result, a ‘composite’ view emerged that argued that the famine event was a result of a process. Several models have reflected this view. An MSF model identified three stages: food insecurity, food crisis, and famine, health crisis, and death (Oxfam, 2001). Walker (1989) suggested that, in the context of a larger process, a famine starts when affected populations undertake coping strategies that undermine their ‘‘future security for present survival” (p. 50). The pressure and release (PAR) model (Wisner, Blaikie, Cannon, & Davis, 2003) takes a similar composite view, arguing that a range of processes that increase the vulnera- bility of households, on one side, combined with natural hazards, on the other, can create the pressure that leads to crises, including famines.

While these models seek to identify the broad outlines of the crises, there is another literature that analyses the detailed sequences within these larger phases. The literature on coping strategies has described the overall pattern of actions that house- holds seem to consistently take in response to drought-induced food crises (Corbett, 1988; Walker, 1989; Watts, 1983). There have been important nuances of these findings, suggesting that some strategies are carried out in parallel (Devereux, 1993) or applying the concept to other contexts, such as urban settings (Curtis, 1995). More recent formulations have distinguished between intensification, diversification, and migration (Scoones, 1998), and have emphasized the role of social capital in surviving famines (Maxwell, Majid, Adan, Abdirahman, & Kim, 2016). Taken together, they have created a rich understanding of affected populations’ responses to these crises.

This paper will draw on many of these insights about the broader phases and the more detailed understanding of the affected populations’ responses. It will adopt what is essentially a composite approach. While recognizing that historical and under-

2 Interestingly, in some of his case studies, Sen did describe different temporal phases of the famines, suggesting a clear awareness of processes, too.

lying processes are critical to understanding the causation of fami- nes, it focuses on a time-bound set of stages. One of those stages is the ‘famine system,’ which represents the famine event, or simply the famine, itself. As we will see, the preceding stages may or may not lead to the development of a famine system, depending on a number of factors. At the same time, the paper will draw on the coping strategy literature that has informed the understanding of how these crises are experienced and responded to by affected populations.

However, in doing so, it also attempts to offer a new model that builds upon several critical insights that have not been fully incor- porated into previous approaches. First, expanding on Howe (2010), the paper emphasizes systems thinking, suggesting that there are synergistic interactions among different elements that lead to the rapid shifts that characterize famines: the price spirals, the rise in malnutrition and mortality, and the intensive responses, after which interest and support taper off. It suggests that it is the interaction of factors in a system, rather than a simple linear sequence of stages, that characterizes the evolution of a famine. While Howe (2010) describes some of the key features of self- reinforcing dynamics and famine systems, this paper explicitly names these elements, introduces the concepts of pressure, hold, and rebalancing, and creates a more comprehensive and general- ized model that integrates them into a coherent, overarching framework.

Second, it makes a distinction between two types of models. The first provides labels for different categories of crises, based on their level of severity. For food security crises, these have cur- rently been agreed as: minimal, stressed, crisis, emergency, and famine (IPC, 2017). Since most crises pass through these different levels on their way to becoming a famine, they can be viewed as different ‘stages’ or phases of a famine’s evolution. In this paper, by contrast, the emphasis is on explaining the pattern of elements that come together dynamically to generate these crises of various levels. In some cases, they will only create a ‘crisis;’ in others, if conditions are right and sustained, they will generate a ‘famine.’ It therefore does not focus on the ‘stages’ of severity, but rather on the ‘stages’ of the dynamics that produce that severity.

Third, by examining both historical and contemporary crises, it searches for a model that is applicable to all types of famine. If the literature of the 1980s emphasized nature-induced crises (Corbett, 1988; Sen, 1981; Walker, 1989), the seminal works of the 1990s focused on the prominent role of conflict (Duffield, 1994; Keen, 1994; Macrae & Zwi, 1994). In the 2000s, more attention has been given to the global and political natures of crises and the diverse circumstances in which they occur, identifying a wider range of famine types than previously understood (Devereux, 2007). A strong literature on individual crises in the Soviet Union (Dolot, 1985), Maoist China (Becker, 1996), Greece (Hionidou, 2006), North Korea (Natsios, 2001), Malawi (Vaughan, 1987), and other locations has also broadened our understanding of the different circumstances that might lead to a famine. This model attempts to capture and account for this diversity.

Finally, in the process, the paper indirectly engages with and potentially complicates the understanding of accountability in famines. A number of authors (Marcus, 2003; de Waal, 2017) have argued for the criminalization of famine, a task which is in part dependent on the ability to assign culpability. The model proposed in this paper, with its non-linearities and feedbacks, identifies a range of factors that need to be considered in understanding the formation of a famine system, possibly making attribution more difficult. However, since it also clearly recognizes that some acts can be intentional – and proportionally more significant and deci- sive – in a famine’s development, the model does not stand at odds to the larger project, but may add a degree of context and a layer of complexity to the discussions.

146 P. Howe / World Development 105 (2018) 144–155

3. Basic concepts

The model suggests that five basic elements characterize key aspects of a famine’s formation and evolution: pressure, hold, self-reinforcing dynamics, famine system, and rebalancing.3

3.1. Pressure

The initial stage of a famine’s formation occurs when a commu- nity experiences pressure on its food security and nutrition. In this context, pressure can be defined as the result of a set of factors combining to partially or systematically disrupt a vulnerable group’s ability to meet their food and nutrition needs. The degree of pressure is a function of the strength and comprehensiveness of the disruptive factors (whether shocks or stressors) and the vul- nerability (or, conversely, the resilience) of the communities. Strong and comprehensive disruptive factors combined with deep underlying vulnerability lead to a high degree of pressure, while weak and minimal disruption combined with limited vulnerability is associated with a low degree of pressure.

The disruptive factors can take a variety of forms: politically- induced, such as conflicts, blockades, and raiding; naturally- induced, such as droughts, floods, and other hazards; economically-induced, such as high food prices or currency depre- ciations; and socially-induced, including discrimination based on political, religious, or ethnic affiliation. They can be intentionally or unintentionally inflicted, or some combination of the two. In many cases, a number of interrelated factors – such as drought and conflict – combine simultaneously to disrupt a population’s ability to meet its food and nutrition needs. Among these various factors, one – for example, conflict – may have proportionally greater impact than the other – in this instance, drought. Disrup- tive factors can also arise in succession: a drought in one year may be compounded by a disease outbreak during the rainy season in the next.

Vulnerability can be defined as the susceptibility of certain pop- ulations to experiencing food security and nutrition crises. The vul- nerability, or lack of resilience, can take several forms. It may consist of underlying chronic food insecurity and undernutrition, which disruptive factors can accelerate rapidly into famine condi- tions. Alternatively, or in addition, it may be an expression of sus- ceptibility to potential disruptive factors. For instance, if a community is dependent on a single livelihood, such as rain-fed agriculture, its food security and nutrition is more easily disrupted by a drought, creating greater vulnerability. Finally, the vulnerabil- ity can be a function of a limited capacity, in terms of assets, to withstand a combination of disruptive factors.

It is the interaction of disruptive factors and vulnerability that creates the pressure that can lead to a famine. This interaction has a differential spatial impact. Although a number of communi- ties and households experience a shock, the impact will be differ- ent for two reasons. First, the shock is not equally intense in all areas, with some populations being more affected by, for example, a drought or conflict than others. Second, a community or house- hold’s vulnerability to the disruptive factors will also vary. As a result of these differential impacts, pressure ‘disturbances’ will appear at first in various locations where the shock is intense and the vulnerability is high, creating a ‘spotty’ pattern.

Within a given spatial distribution of pressure, there are differ- ential livelihood, social and biological patterns. For example, cer- tain populations, such as farmers, may experience greater pressure from a drought than, say, traders. But even in a rural dis-

3 These names of the elements will be used as shorthand descriptions of the broader concepts presented in this section.

trict consisting predominantly of farmers engaged in rain-fed agri- culture, there may be differences in assets and therefore in the pressure felt by different households. Within households experi- encing pressure, there may be increased social vulnerability based on gender, age, disability and other diversity characteristics. Some people, especially young children, the elderly, and sometimes males, may also experience greater biological vulnerability. It is critical to understand these ‘bands’ of particular intensity within the wider pattern of pressure.

The pressure can also have varying intensities over time. A one- off event, such as a drought or flood, can damage a harvest, but its impact will intensify over time as the community consumes its resources and enters the lean season. By contrast, on-going pres- sure, such as a conflict, may be present on a regular basis, but vary when the fighting ebbs away from the area. In other instances, a community may experience a series of shocks over the course of a number of years that gradually erodes its assets and coping capacities. When a subsequent shock occurs, it will impact a much more vulnerable community, intensifying the pressure. In other words, repeated shocks may ‘set up’ the conditions for intense pressure and a potential famine in the future if there is a hold in place.

3.2. Hold

The ‘hold’ refers to a condition in which communities affected by pressure do not receive sufficient release from it to prevent a deterioration of their food security and nutrition situation. Within the evolution of a famine, the hold’s primary function is to permit the pressure to lead to the creation of self-reinforcing dynamics. Holds can take many forms, including natural, policy, political, social, economic, and others; they can be put in place deliberately or unintentionally; and they can occur alone or in combination. When occurring in combination, they may have proportionally dif- ferent impacts and levels of intentionality, with political holds, for instance, often being more deliberate and challenging to break than natural holds.

An example of a natural hold occurs when a drought affects the harvest one year and the new harvest will not be ready until the next year. In this case, the timing of the seasons will put in place a natural hold, which will not release the community from the hungry season until the new harvest arrives. A policy hold refers to a situation in which there is an unwillingness or inability of a government or other entity to respond to the needs of a commu- nity (cf. Plumper & Neumayer, 2009; Rubin, 2009) in a manner commensurate with the degree of pressure they are experiencing. Political holds can take a variety of forms: an isolated regime refus- ing outside assistance to meet the needs of its people; conflict or siege, which is not only a source of pressure but also prevents access for humanitarians; blockades imposed as part of a war effort; and a political ideology pursued despite its impact on the population. Social holds can exist when certain communities, because of ethnicity, party affiliation, or other characteristic, are systematically denied access to needed assistance or opportunities when under pressure. Traders hoarding food and high food prices that depend on global or external trends are examples of economic holds.

It is important to recognize that holds have different degrees of permeability. That is, even though an area may be largely cut off due to fighting, there may be some trade or assistance that gets through this political hold. Similarly, in a politically isolated coun- try, some areas, located near neighbouring countries, may be able to benefit from cross-border exchange. Another feature of the hold is that it fluctuates in strength over time. A drought that caused a poor harvest may impose a natural hold that is temporarily allevi- ated when the new harvest comes in, but that is re-imposed if that

P. Howe / World Development 105 (2018) 144–155 147

harvest is poor. Similarly, fighting may lessen over time in a certain area but will intensify later, giving a pattern of strengthening and diminishing force to the hold that prevents humanitarian access. The most powerful holds have limited permeability and do not fluctuate significantly.

In some cases, the pressure and hold will have the same source. A conflict may create pressure by systematically disrupting access to necessary nutritious food for vulnerable populations. At the same time, the insecurity may represent a hold by preventing humanitarian assistance from reaching those in need. However, it is important to distinguish between their functions, because in many cases they will be distinct. For example, a drought may be the source of pressure in a rural agricultural community, but the holds will have to do with the timing of the next harvest and the government policies that do not respond adequately to the com- munity’s needs. As another example, drought and conflict may both be sources of pressure, while only the conflict contributes to the hold.

There are some situations in which the pressure is almost immediately alleviated through a response. For instance, insurance mechanisms or a well-functioning social safety net system may prevent the pressure from intensifying. However, when a hold is in place, the initial pressure ‘disturbances’ may worsen and spread to other locations. At a certain point, if the hold is maintained, it will lead to the development of self-reinforcing dynamics.

3.3. Self-reinforcing dynamics

The hold creates the conditions for self-reinforcing dynamics to occur in a community under pressure. A self-reinforcing dynamic is a pattern of accelerated and interrelated changes in key factors driving food insecurity and undernutrition. In a simplified exam- ple, a drought and poor harvest may exert pressure on a commu- nity dependent on rain-fed agriculture. Fearing that they will run out of food before the next harvest, households conserve their sup- plies and begin selling assets such as livestock to purchase grain. As numerous households engage in this behaviour, the price of grain is bid up by the increased demand and limited supply, which is affected by both their conservation and the poor harvest. It is compounded by traders, who recognize the desperate situation and hold back the grain from the market until prices rise further. At the same time, the price of assets falls as the market is flooded with goats, sheep and cattle. Their purchasing power and ability to acquire food therefore declines rapidly.

Increasingly desperate households seek alternatives through labour opportunities, but as a critical mass of households turn to this option, the increased supply of labour drives down wages, again worsening the terms of trade for food. As food becomes more difficult to obtain, men may migrate to far off areas to seek employment, and thefts may increase. Households restrict their consumption and collect famine foods. But the limited number of meals and deteriorating quality of food reduce their energy, mak- ing it difficult to pursue potential options or to think clearly. These dynamics are beyond an individual household’s ability to control and are both interlinked and mutually reinforcing.

The challenge is that at this stage in the crisis the changes can be observed and measured – for instance, in the prices or the num- ber of thefts – but it does not yet lead to extremely elevated under- nutrition rates and sudden spikes in mortality. To some extent, it corresponds to the ‘watch’ (Howe, 2010), when the dynamics have been set in motion, but the populations seem to be ‘coping’ because the most visible and concerning consequences have not yet materialized. Yet what is occurring is a masked acceleration of destitution and the rapid exhaustion of the means to stave off a famine. It should therefore be viewed with seriousness. This pat- tern is seen in the hungry season in many countries, but it may not

proceed to a famine system because the natural hold is broken by the harvest.

If the type of hold permits, the self-reinforcing dynamics may both intensify and start to spread to other communities. In the drought example, the intensification occurs in the most affected communities because the pressure is in place without release. Other communities may also be influenced in two ways. First, over time, the pressure intensifies for them as they exhaust their buffer. Second, people experiencing distress in already-affected communi- ties might, for instance, migrate to these areas to search for work, driving down wage rates, bidding up the price of food, and making requests for assistance. During this period, more communities – not necessarily adjoining but nevertheless linked by markets, livelihoods, or ethnic relations – fall into a self-reinforcing dynamic of distress. The ‘spotty’ pattern of pressure is intensified, with more spots appearing and existing spots deteriorating into more severe conditions.

Certain ‘bands’ of the population, based on economic, social, political, demographic, environmental, and biological characteris- tics, will continue to be more affected than others within these communities. The bands are impacted both because they are espe- cially vulnerable to the particular self-reinforcing dynamic (e.g. economically dependent on rain-fed agriculture in a drought) and because the self-reinforcing dynamics start to change the soci- ety in ways that further heighten and expose other forms of vul- nerability (e.g. social marginalization of the elderly). The breakdown of traditions of communal sharing, the rise in crime, and the emotional and physical changes associated with hunger ramify outwards and have implications for who is most vulnerable during the crisis and for the heightening of that vulnerability. This precipitating out of the most vulnerable is a function of the dynam- ics of the crisis. While almost every crisis produces vulnerable groups, their precise composition often differs based on the context.

3.4. Famine system

A famine system occurs when there is a tipping point and a growing sense of unstoppable momentum that results in a sudden and dramatic rise in undernutrition and mortality rates. It usually takes place when the dynamics driving the hunger can no longer be balanced by the coping strategies of the populations. At that point, the dynamics become even more strongly self-reinforcing and lead to the spikes in undernutrition and mortality, with rates crossing the agreed thresholds that indicate the presence of a famine (Howe & Devereux, 2004; IPC, 2017). The crisis develops around an ‘epicentre’ or ‘epicentres’ (cf. de Waal, 1989). The formation of the famine system and the crossing of associated thresholds defines the crisis as a ‘famine,’ and a famine is said to be occurring during the existence of the famine system. The use of the term ‘famine system’ emphasizes the synergistic nature of the dynamics. The mortality from the famine would include deaths that occurred during the famine system; mortality in the stages preceding the creation of the famine system and deaths that take place after its collapse would be associated with the famine process but would not be, strictly speaking, caused by the famine itself.

The characteristic of a famine system is a dynamic that is ‘out of control’ (cf. United Nations Administrative Committee on Coordination/Sub-Committee on Nutrition [UN ACC/SCN], 1997). The exact dynamic will vary depending on the situation, but it will have several essential features. First, for a critical mass of people, a diverse range of ‘coping strategies’ will become exhausted, allow- ing the self-reinforcing mechanism to become accelerated. In the example of the drought, the self-reinforcing dynamics were mak- ing it difficult for people to obtain sufficient food and nutrition, but were offset, to an extent, by the coping strategies employed.

148 P. Howe / World Development 105 (2018) 144–155

Once those coping strategies are exhausted, the sudden lack of food dramatically undermines the body’s (and mind’s) ability to func- tion, making it even more difficult to obtain food, and undermining capacity further, in a rapid spiral that leads to spikes in acute mal- nutrition and death (see Fig. 1), and to desperate measures such as outmigration.

Second, the system will be reinforcing and linked at multiple levels: global, regional, national, community, household, and indi- vidual. In the case of the drought, a policy hold may lead to inac- tion by global, regional, and national actors at least at the early stages, because there are other priorities and there is a reluctance to commit funds to a situation that is not yet deemed sufficiently serious. The famine system will arise in the affected communities, leading to the accelerating dynamics for a critical mass of house- holds who will experience the consequences at the individual bio- logical level in the form of undernutrition and the risk of death for the most vulnerable. In a globalized world, decisions made interna- tionally are connected to individual experiences. Sometimes, there may be efforts at the global level to respond to a crisis that are blocked at, say, the national level, but it is the overall interaction that helps to determine the biological outcomes at the individual level.

Third, it is synergistic and interconnected in space and time (Dyson, 2002). The famine system can be greatly intensified by synergistic interactions with new dynamics. For example, outmigration often leads to exposure to disease, which interacts synergistically with undernutrition to elevate deaths in some circumstances (de Waal, 1989; Young & Jaspars, 1995). At the same time, the famine systems have longer-term effects. The accelerated destitution that occurs may set the population up for a follow-on famine (Dyson, 2002), or may lead to demographic impacts, such as a reduction in pregnancies and births or movement to urban areas during the crisis. These demographic changes may be partially or fully compensated for after the crisis is over.

Famine systems can vary in their appearance depending on the circumstances and can, for simplicity, be characterized as open or closed. An open system (see Fig. 2) refers to a famine in which there are few restrictions on movements of the affected popula- tions and those wishing to assist them. It often arises because of a natural or policy hold. Open famine systems usually involve a process of ‘emptying,’ which can take two forms. First, people sometimes out-migrate from an area affected by famine to a place beyond the ‘blended periphery,’ where conditions are better. This may lead to the formation of camps, or aid magnets, but the effect is that the famine system site is depopulated at the centre and peo- ple are concentrated in locations where they can receive assis- tance. This concentration can in turn lead to higher mortality rates as a result of disease (de Waal, 1989). Second, those who are left in the centre of the famine system are often the most vul- nerable – the elderly, sick, disabled, children – who cannot escape the crisis. The system will experience an ‘emptying’ in the sense of high mortality rates.

A closed famine system can occur when the pressure and hold are focused in a clearly delimited area, as in the siege of a town, a blockade of a country, or a politically isolated society. The system is closed in the sense that populations are unable to move in and out of the area, and people from outside the area are unable to observe the famine itself or to provide assistance. In a closed sys- tem, there may be no visible outmigration, and ‘emptying’ in this sense, although there may be significant movement within the area contained by the ‘hold,’ which in some cases represents the boundary of the famine. There will, however, be an ‘emptying’ related to high levels of mortality, especially among the most vul- nerable. Some famines may combine characteristics of closed and open systems. If, for example, the hold is somewhat permeable, some people will find a way to flee the country (Natsios, 2001).

Famine systems also have varying intensities. Spatially, for the reasons described, the famine’s intensity is different depending on the conditions – the force of the pressure, the nature of the hold, the development of self-reinforcing dynamics – in a specific area. Studies of mortality tend to show varying levels, though usually concentrated around one or a few epicentres (cf. de Waal, 1989), areas where the famine was the most intense, or where people congregated for assistance but nevertheless experienced high rates of mortality. Temporally, famine systems can also change in inten- sity. A famine system that produces deaths during the winter in one year may abate during the spring when more wild foods become available, but then, if the hold remains in place and the pressure continues, the famine system may regain intensity during the next winter (cf. Dolot, 1985).

The system accelerates the suffering of the ‘bands’ of the most vulnerable, the ‘losers’ from the famine, but also creates ‘winners’ (Keen, 1994). The system acts on an existing society, consisting of certain social, political, economic, environmental, and demo- graphic patterns that are constantly in flux but nevertheless occur within an expected range and create a sense of ‘stability.’ The fam- ine system, even more than the self-reinforcing dynamics, is a powerful disruption to this stability and is at least temporarily transformative. As a result, some will suffer and be more likely to experience destitution, undernutrition, and death, for economic, social, political, environmental, demographic, and biological rea- sons. The bands depend on the specific context. However, in many but not all famine situations, it is young children, the elderly, and young men, who are the most vulnerable to death (O’Grada & Dyson, 2002). At the same time, some members of society may benefit from the new system. The winners will also depend on the context, but may range from traders (Sen, 1981) to farmers (Hionidou, 2006) to government officials (Keen, 1994; Natsios, 2001). This temporary advantage may or may not outlast the re- balancing of the system.

3.5. Rebalancing

A famine system collapses because of a rebalancing of the dynamics, which occurs when the conditions that have driven its creation are taken away. The rebalancing can often take place in two reinforcing ways. First, the famine system itself, by its very ability to cause deaths, begins to ‘use up’ the most vulnerable pop- ulations or to force them to migrate and is therefore not able to maintain, for a long period, extremely elevated levels of mortality in a location. In those cases, it may start to dissipate or, if there are self-reinforcing dynamics taking place in other areas, to ‘move’ to another location.

Second, in the face of such high mortality, governments, civil societies, and the international community attempt to collapse the system. These efforts have two components: the breaking of the hold and the altering of the self-reinforcing dynamics. The breaking of the hold could involve the arrival of a new harvest, a change in a policy position, an easing of the conflict during the rainy season, or an opening up of a politically isolated country to assistance. The altering of the dynamic occurs because more food is available on the market and within households, and there is an opportunity to return to homes and land. In effect, the self- reinforcing negative loops are offset by balancing loops, in a dynamic that plays out not only over time but space.

It is important to recognize, however, that the process is not lin- ear and does not involve an uninterrupted diminishment of the system. The process of rebalancing can lead to unexpected dynam- ics, which might be described as ‘eddies,’ that reinforce rather than counter the death toll. For example, when a harvest arrives, break- ing the natural hold that has permitted the formation of a famine system, a population may consume too much unripe food too

Fig. 1. Stylized representation of stages and mortality over time.

Fig. 2. Stylized representation of the evolution of an ‘open’ famine.

P. Howe / World Development 105 (2018) 144–155 149

quickly, causing intestinal problems and sometimes death. In other cases, new events – a re-imposition of a blockade (Hionidou, 2002), an epidemic outbreak, another poor harvest, or a refugee crisis (de Waal, 1989) – can temporarily worsen the overall trend of improvement. Sometimes, the famine leads to changes, such as movements to towns, which are not easily reversed and may rep- resent a transformative legacy of the crisis.

4. Applying the concepts

These basic concepts – pressure, hold, self-reinforcing dynamic, famine system, and rebalancing – appear to apply to a range of his- torical famines. In this section, we look at three: the Malawi Fam- ine of 1949–50, the Greek Famine of 1941–42, and the Somalia Famine of 2011–12. (See also Fig. 3 for a range of other examples.) As a counterpoint to these cases, and in order to clarify the con- cepts further, we also examine the crisis in Bosnia and Herzegov- ina4 in 1992–95, in which many of the elements of the model appeared but a famine did not ultimately materialize.

4.1. Malawi famine, 1949–50

Drawing on the multi-disciplinary work of Megan Vaughan (1987),5 we can identify each of these elements in the Malawi fam-

4 The paper will use ‘Bosnia’ for short in the rest of the text. 5 In the case studies, it should be assumed that all the factual information is

derived from the works cited at the start, unless otherwise indicated. The translation of that information into the elements of the framework is my own.

ine of 1949–50. Although she did not organize her study along these lines, the elements are clearly illustrated in her descriptions.

Pressure. In January 1949, normally the wettest month of the year in southern Malawi, the rains failed completely, disrupting the growing season and representing a severe shock to the liveli- hoods of the rural farming communities. The severe drought affected a population that was already vulnerable because of its dependence on rain-fed agriculture and from having suffered an outbreak of locust the previous year, which left ‘‘few full granaries” (Vaughan, 1987, p. 29). The pressure was, at this stage, a ‘‘patch- work one, varying in intensity over a small area,” both because of the resilience of the local crops in certain places, and ‘‘the resources available to [a given area] other than their own- produced crops” (Vaughan, 1987, p. 28). The crisis could vary not only among areas within a district but also among individuals, with some being viewed as more ‘‘improvident” than others (Vaughan, 1987, p. 28).

Hold. In Malawi, the hold consisted of two elements. The first was a natural hold. The failed rains had affected the harvest in 1949, but the next main harvest would not be until the following year. As a result of this natural hold, the agricultural community would not be able to receive a significant release from this pressure for almost a whole year. To some extent, this natural hold fluctu- ated: some areas harvested a small crop of sweet potatoes in May and June, and a harvest of mangoes offered a respite to the hold in mid-September. But they were described ‘‘as a drop in the ocean” (Vaughan, 1987, p. 41) and were therefore only minor fluctuations within the larger trend of increasingly severe pressure and worsening conditions during the course of 1949.

Fig. 3. The application of the model to other selected crises. (See above-mentioned references for further information.)

150 P. Howe / World Development 105 (2018) 144–155

The second was a policy hold. The British government did not fully recognize the seriousness of the situation and was reluctant to undertake decisive relief measures. Their principal intervention was ‘Operation Nyala,’ which ‘‘involved the replanting of gardens and the planting of root crops, enforced by agricultural officers and compliant village headmen and chiefs” (Vaughan, 1987, p. 38). Other measures included a controversial ban on beer brewing, an ultimately ineffective effort to support a Greek trader to increase the supply of fish, and allowing 500 families to find work with South African mining companies (Vaughan, 1987). But the efforts were not commensurate with the severity of the crisis and failed to stem the deterioration of the situation.

Self-reinforcing dynamics. With the holds in place, a set of interrelated self-reinforcing dynamics was generated. By February, the price of maize in urban centres rose from 1d. per lb. to 5d. per lb. at government-controlled stores. It had to be brought down to 3d. per lb. to make it affordable in urban areas. By March, the crime rate had increased with a surge in crop thefts. In Blantyre, the capital, there was an average of 130 cases a month in 1949. By March or April, there was an ‘‘extreme mobility of the popula- tion (especially the men) in search of food to buy, earn, beg, or gather” (Vaughan, 1987, p. 31), but also a breakdown of the system of mutual aid.

These dynamics were related: the pressure from the drought and vulnerability caused prices to rise and purchasing power to decrease. More people were forced to seek alternatives means of acquiring food, leading to an increase in migration, petty thefts, and wild food collection. As more people engaged in these alterna- tives, there was more competition for the limited resources and a breakdown in the system of mutual aid. Yet the reliance on these strategies gave the appearance that the communities were coping, corresponding to a ‘watch.’ The government was surprised at the community’s ability to survive: ‘‘The District Commissioner for Dedza District, where the lakeshore areas were badly affected, described his astonishment at the ability of people to stay alive and apparently healthy when there were no food stocks left in the village” (Vaughan, 1987, p. 31).

The self-reinforcing dynamics affected various ‘bands’ of the population differentially. While men by-and-large migrated in search of food, sometimes bringing assistance back, sometimes remaining in better-off areas, women and children were left on their own. One district commissioner noticed ‘‘‘how many children and young mothers were neglected and without any means of sup- port – one saw cases of children with no parents or relatives to turn to in times of distress, and of females deserted by their husbands, who had gone off to find food and employment for themselves”’ (as

P. Howe / World Development 105 (2018) 144–155 151

quoted in Vaughan, 1987, p. 35). In this case, since most of the rural population shared a similar livelihood, the most evident differ- ences in the impact of these dynamics were along the lines of age and gender.

Famine system. Despite some fluctuations in the holds, the self- reinforcing dynamics had left people unable to meet their needs: ‘‘By September it was obvious that the ‘internal distribution sys- tem’, the functioning of which had so surprised officials was finally breaking down in some areas” (Vaughan, 1987, p. 41). When the rains came in October, signs of malnutrition suddenly became evi- dent. ‘‘Survivors recount how physical appearances changed – legs became swollen with ‘famine oedema’, faces were haggard, and people were no longer able to speak” (Vaughan, 1987, p. 44). Distri- bution centres, which sold rather than gave maize to needy people, were set up in the rural areas, but it was already too late. The fam- ine system had taken hold and was producing malnutrition and deaths.

Reflecting the emptying in an open famine system, the commis- sioner for the Southern Provinces arrived at one distribution center to find a crowd of 3,000 people waiting for assistance. He saw two children on the verge of death and commented: ‘‘There were many others approaching the same condition, and I was really horrified at the state of these people” (as quoted in Vaughan, 1987, p. 44). In another location, 2,000 people gathered to buy food. By the end of January 1950, free food was provided to those near starva- tion. Officially, 71 people were reported to have died in these newly established camps, but ‘‘countless numbers of people died on their long journeys to find food” or ‘‘succumbed to the normal round of diseases,” ‘‘old people were left to die,” and ‘‘young chil- dren were abandoned by their mothers” (Vaughan, 1987, p. 47).

The mortality was stratified, with the old and young most likely to starve. But as Vaughan (1987) argues: ‘‘We need more than one classification, because the structure of famine and the definition of who suffered were the result of a number of factors interacting” (p. 115). In part, it was concentrated in the areas affected by the drought, but more particularly among those who did not have access to employment. The elderly and young were not only bio- logically vulnerable, but also socially at risk, with many in both age groups being abandoned and unable to find their way to assis- tance. The deaths thus resulted from a combination of the pressure and the ramifying implications of the social breakdown that accompanied it.

Rebalancing. Eventually, both the natural and policy holds were broken. The rains in 1950 were good, and, although the population was weak and had limited agricultural inputs, the new harvest of maize arrived, breaking the natural hold on the population. At the same time, the introduction of free assistance through camps and an intensive effort to support maize and cassava production constituted an uneven, but much more engaged effort at the policy level to address the crisis, when it was finally understood to be out of control.

These changes enabled the self-reinforcing dynamics to re- balance and dissipate. With food from the harvest and assistance from the government, the price spirals diminished, ‘‘many of the ‘lost’ husbands” returned to the area, crop thefts went down, and life began to return to normal (Vaughan, 1987, p. 48). However, even in this process of rebalancing, there were small ‘eddies’ of negative dynamics. Hungry, but recovering people died when they ate the green cobs from the new harvest, and their bodies could not cope with the food. But finally, more than a year after the rains failed, the famine system collapsed, and the famine was over.

4.2. Greek famine, 1941–42

During the early years of World War II, many parts of Greece fell under occupation and experienced a severe famine. Drawing prin-

cipally on the work of Violetta Hionidou (2002, 2006), it is possible to see how the famine evolved through the stages on the Cycladic island of Syros.

Pressure. When German forces and their Italian and Bulgarian allies quickly occupied Greece in April 1941, they placed pressure on the food security and nutrition of the populations through a ser- ies of disruptive measures including: confiscating food stocks, pro- hibiting fishing, and preventing the transport of goods and the movement of people between regions. Soon afterwards, and in retaliation for the occupation, the Allies imposed a naval blockade on the country. As a result of this combination of disruptive factors, many people on the island of Syros began to struggle to meet their food and nutrition needs.

But the pressure had differential impacts. The island consisted of three main population areas: Hermoupolis, the capital; Ano Syros, the nearby old town; and the rural communities. While the rural communities could conceal some of their food supplies from confiscation, and were able to grow more food, the urban population of Hermoupolis was dependent upon supplies from rural areas of Syros and other parts of the country. Within that urban center, the unemployed, which included many demobilized soldiers, had particular difficulty meeting their nutritional needs. With decreasing supplies available because of the confiscation, restrictions on movement, and blockades, and limited sources of livelihood income, this ‘band’ of the population felt particular pressure.

Hold. The hold in Syros consisted of three, almost concentric layers. The outermost layer of the hold was the blockade. Although there were appeals and warnings about the deterioration of the sit- uation, the British decided to keep the blockade in force for mili- tary reasons, restricting the movement of goods, including food assistance, for most of the period of the famine process (April 1941 to November 1942). Between October 1941 and February 1942, the British did permit the transfer of food from Turkey, but the supplies were small (10,000 metric tonnes) compared to the need and were only distributed in Athens. Although these distribu- tions reflected a certain ‘permeability’ in the hold, for the most part the blockade contributed to the pressure being maintained and intensified.

The second, middle layer of the hold was comprised of the mea- sures of the occupying powers. The Germans also debated inter- nally whether to maintain these measures in the face of the deteriorating food security and nutrition and financial conditions, but decided for military reasons to keep them in place. Finally, as the innermost layer, the town of Hermoupolis experienced a social hold: the predominantly Greek Orthodox community in the town was unable to access the Catholic rural areas or benefit from their network of support because of differences in religion and conse- quent social separation of communities. Anos Syros, by contrast, was Catholic and, because of some peri-urban villages, had more direct access to farming areas, where food supplies were greater.

Self-reinforcing dynamics. While there is not extensive informa- tion on how the populations coped in the face of these holds, it seems clear that self-reinforcing dynamics began to materialize rapidly. Not only the shortages of food, but also the printing of cur- rency for the payment of occupation ‘expenses’ to the Germans, led to an astonishing rise in the price of food. The price of bread ‘‘un- derwent a 23-fold increase between April 1941 and January 1942 and an 89-fold increase between April 1941 and July 1942” (Hionidou, 2002, p. 183). As a result, a bartering system replaced the cash economy.

Famine system. With the pressure and holds in place, and clear signs of self-reinforcing dynamics, there had been warnings about the deteriorating situation for months, with the US ambassador to Greece saying in July 1941 that it was a ‘‘‘progression towards fam- ine’” (as quoted in Hionidou, 2002, p. 183). But it appears that the

152 P. Howe / World Development 105 (2018) 144–155

situation in Syros reached a tipping point for a rise in malnutrition and deaths in September 1941, when the baseline rate of 50 deaths per month was exceeded and rose rapidly. By January 1942, the number of deaths per month had increased to 484 in comparison to 35 in January 1939, representing the peak of the famine. In this case, and unusually for a famine, disease did not play a significant role in the deaths and was not a part of the reinforcing dynamics in the famine system. Also, unlike many other famines, there was not significant outmigration in this ‘‘closed” (Hionidou, 2006, p. 6) sys- tem, because of the restrictions placed on movement.

However, there were important variations in the impact, which reflected the differences in pressure experienced by the communi- ties. In non-Catholic Hermoupolis, with its urban dependence on food supplies, the death rates were the highest, with the level 8 times the pre-famine rate. By contrast, the countryside, with its better access to food and social networks, had a death rate three times the baseline. In Ano Syros, the level was in between the other two locations with deaths 6 times greater than the pre-famine rates. Since it had a much larger population, and higher crude death rate, Hermoupolis also experienced the greatest levels of absolute deaths. Within this geographic pattern, boys and young men, who were unemployed and poor, from the urban centre of Hermoupolis, appeared to be most vulnerable to death, especially in the early stages of the crisis, as a result of undernutrition as sol- diers and then insufficient means to access food once they were demobilized.

Rebalancing. Just after the peak of famine deaths in January 1942, and following pressure from the Greek government in exile and the United States, the British lifted the blockade in February 1942, representing a break in the hold. As early as June 1941, the Germans and other occupying forces had already pledged not to requisition food assistance. With the agreement that the Interna- tional Red Cross would deliver the assistance, food supplies began entering Greece irregularly and by November 1942 were provided on a more regular basis. Although there was a ban by the Allies on the transport of food in small vessels that temporarily reversed the gains, the assistance broke the holds, including the social ones, and undermined the self-reinforcing dynamics. The famine system col- lapsed, with the crude death rates dropping below the baseline of 50 per month in November 1942.

4.3. Somalia famine, 2011–12

Although the famine in Somalia in 2011–12 occurred much later than the other two crises and had distinctively contemporary characteristics, these same stages appear to have been replicated. The description of the events draws principally on a series of arti- cles on the Somalia famine published a year after the crisis (cf. Maxwell, Haan, Gelsdorf, & Dawe, 2012).

Pressure. A combination of three main disruptive factors came together in 2011 to constrict the ability of pastoral and agro- pastoral communities in southern Somalia to obtain sufficient food and nutrition: drought, food price rises, and conflict (Maxwell & Fitzpatrick, 2012). The rainfall in the deyr season in 2010 was the lowest in some places in 50 years and had a dramatic impact on agricultural production, livestock pasture and watering holes. While the rains improved somewhat for the gu season, the drought caused livestock deaths and severely damaged crops, leaving pro- duction of maize and sorghum at their lowest levels in the previous 15 years (Maxwell & Fitzpatrick, 2012). While these factors, when combined with the ongoing conflict, undermined livelihoods and incomes, the global food price crisis, which had begun in 2008, led to extremely high prices for cereals and other essentials and rapidly deteriorating terms of trade.

But these disruptive shocks affected a population that was already highly vulnerable to food insecurity and undernutrition

after many years of immiseration. Maxwell and Fitzpatrick (2012) argued that underlying factors, including prolonged insecu- rity and a history of drought, had ‘‘gradually eroded livelihoods and coping strategies over the past 20 years, making populations vul- nerable to many hazards” (p. 8). Their dependence on rain-fed pas- toralism and agriculture also increased their vulnerability to the effects of drought and prices.

Hold. In this case, the political hold took two interrelated forms that effectively prevented major food assistance agencies from pro- viding an adequate response to these events. On the one hand, al Shabaab refused to permit humanitarian assistance from major agencies into the areas it controlled. CARE International was expelled from southern Somalia in March 2008; WFP withdrew in January 2010; and even the ICRC was forced to close down its operations in 2012. On the other hand, counter-terrorism laws, the naming of al Shabaab as a foreign terrorist organization, and the U.S. Supreme Court ruling that it was illegal to provide ‘mate- rial’ assistance to these groups, created a climate of fear, in which agencies, worried about the possibility of criminal prosecution, were hesitant to deliver assistance in al Shabaab-controlled areas (cf. Menkhaus, 2012). There were also a natural hold related to the seasons and an economic hold based on global price movements.

Self-reinforcing dynamic. In this context of severe pressure, and a strong hold preventing assistance, the self-reinforcing dynamics were generated quickly. Maxwell and Fitzpatrick (2012) indicated that: ‘‘The terms of trade between livestock sale prices and labor wages vis-à-vis food declined rapidly in late 2010 and stayed at historically low levels throughout the first half of 2011. This led almost instantly to a dramatic drop in entitlements for several identifiable groups in southern Somalia” (p. 6). There were also peaks in migration in January 2011 following the failure of the deyr rains in late 2010, and again to a greater extent in July 2011 in response the failure of the gu rains (Maxwell & Fitzpatrick, 2012).

These dynamics were linked. The price rises and the drought combined to greatly undercut the terms of trade, making it extre- mely difficult for populations to meet their food and nutrition needs. As Maxwell and Fitzpatrick (2012) explain: ‘‘In terms of cau- sation, it is difficult to separate out the impact of the drought from the impact of the price rise. The former partially caused the latter; the latter dramatically exacerbated the effects of the former; and both contributed substantially to the collapse of entitlements” (p. 6). This deterioration, in turn, led to both rises in outmigration and an increase in malnutrition levels. These conditions were actu- ally seasonally ‘endogenous’ to parts of Somalia. However, in some cases, it was in areas where the pre-existing levels were somewhat less severe, but the hold was stronger that the situation tipped over into a famine system.

Famine system. While mortality rose above the baseline as early as October 2010, after the failure of the deyr rains, a clear tipping point and steep rise, signalling the formation of a famine system, occurred in April 2011, peaking around July 2011, and finally returning to the baseline level by January 2012. ‘‘These rapid increases in levels of acute malnutrition and mortality between April and July 2011 in Mogadishu are testimony to the speed at which the crisis unfolded” with GAM rates reaching an astonishing 55 per cent and SAM rates rising to 29 per cent (McCloskey Rebelo, Prost, Renk, Guduri, & Hailey, 2012, p. 67). Between May and Octo- ber 2011, at the peak of the famine system, there were on average more than 20,000 excess deaths a month (Checchi & Courtland Robinson, 2013). Overall, the entire process, including the famine itself, had an enormous ‘output’ in terms of mortality, with the total figure estimated, from 2010 to 2012, at 258,000 deaths (Checchi & Courtland Robinson, 2013).

The famine, which was a closed system with some limited openness, involved ‘emptying’ in both senses of the term at least

P. Howe / World Development 105 (2018) 144–155 153

to a certain extent. Where possible, people fled the controlled areas in search of assistance in Mogadishu and other major towns. There was also an emptying in the sense of a high mortality rate, which reduced the most vulnerable populations of children under 5. Of the total deaths, approximately 52% (or 133,000) were children under five years of age (Checchi & Courtland Robinson, 2013). While this group represented a vulnerable band, these cases mainly occurred within other bands of the population, including the Reewin and Bantu minority groups, that for political and socio-economic reasons were particularly vulnerable and dispro- portionately affected (Majid & McDowell, 2012).

Rebalancing. In some ways, the declaration of ‘famine’ in July 2011 helped to trigger the necessary efforts to break some of the holds. Funding doubled within a few months, and the US made a conscious effort to relax the counter-terrorism clauses (Maxwell & Fitzpatrick, 2012). As a result, it was possible to expand support significantly. Although areas were still facing the ‘hold’ of al Shabaab policies, other locations that received people fleeing from the famine were able to provide more comprehensive coverage. At the same time, the next harvest arrived, addressing the natural hold, and by January 2012, food prices returned to normal levels, ending the economic hold (Maxwell & Fitzpatrick, 2012). These changes are reflected in the collapse of the famine system and, by April 2012, a declaration of the end of the famine.

4.4. Bosnia crisis, 1992–95

It may also be helpful to examine a situation in which a famine was widely predicted, but did not materialize, to better understand the development of these crises. Drawing principally on Fiona Wat- son’s study (2007) of the crisis in Bosnia in the early 1990s, it is possible to see how the crisis exhibited several elements of the model, but also why it did not tip over into a famine system.

Pressure. In 1992, the Bosnian Serb army besieged parts of the newly independent state of Bosnia and Herzegovina in an attempt to claim territory for the competing Republic of Sprsk. The army encircled the city of Sarajevo, shelling the center and attempting to cut off supplies to the inhabitants. These disruptive factors had several impacts on the trapped populations: they interfered with economic activities, contributed to the breakdown of the health system, and led to ‘‘an abrupt and severe reduction in food availability” (Watson, 2007, p. 209, italics in the original). At the same time, naturally-induced pressure in the form of a cold winter was superimposed on the human-induced siege leading to predic- tions of a rapid and large-scale famine. As Watson (2007) notes, the alarming forecasts of imminent famine were in part based on pre- vious European experiences, including the blockade of Germany during the First World War, and the Warsaw Ghetto, the Siege of Leningrad, and the Dutch ‘hunger winter’ during the Second World War.

Although not using the same terminology, Watson (2007) argues that a famine did not occur in the winters of 1992 or 1993, in part, because the pressure, which depends on both disrup- tive factors and vulnerability, was not sufficiently strong to initiate the formation of a famine system. Though the siege disrupted the availability of food, access was still possible through a variety of strategies, including gardens, food parcels from abroad, use of sav- ings, paid employment, and remittances. At the same time, the pre- crisis vulnerability of the Bosnian population was low. The protec- tive factors included: ‘‘sound socio-economic conditions, good health and nutrition status, and a demographic profile with a majority of adults who were able to look after the vulnerable – children and the elderly (i.e. a low dependency ratio)” (Watson, 2007, p. 280). This combination of strong but not comprehensive

disruptive factors and low vulnerability helped reduce the inten- sity of the pressure.

Hold. A famine may also have been averted because of the nat- ure of the holds. The most important hold was the siege itself, which aimed to limit the flow of supplies into the cut-off areas. In practice, however, it appears to have been permeable. During a trip to a besieged eastern enclave, a donor noted that inhabitants were growing tobacco and making cigarettes: ‘‘Though the roads into Gorazde were all carefully guarded, the ‘‘lines” surrounding the city were fairly porous. Residents would sneak across the lines and trade their home-made cigarettes for other needed items” (Knight, 1998, p. 11).

More significantly, the hold in key locations was effectively bro- ken by the provision of humanitarian assistance. Watson (2007) notes that, although the levels were rarely sufficient to meet the assessed needs, significant amounts of food assistance reached the besieged populations, representing 84 percent of the food requirements in Sarajevo, and 38 percent in Central Bosnia, in the five months before the ceasefire in 1995. Another study (Stites, Lautze, Mazurana, & Anic, 2005) suggested that the food assistance was well-targeted and, on average, in six Bosnian areas, 47 percent of the population was dependent on the assistance as their primary source of food, with the proportion rising to above 70 percent in some locations. The natural hold of winter had a lim- ited duration and was broken after the change of seasons.

Self-reinforcing dynamics. Although the pressure and hold were not sufficient to initiate a famine system, they did generate self- reinforcing dynamics. The war and the unavailability of food led to extreme inflation and the breakdown of currency-based mar- kets. This was combined with a steep drop (from 75 to 16 percent) in the population having a household member receiving pay as a professional or worker (Stites et al., 2005). In its place, there was increasing reliance on a barter system (Stites et al., 2005; Knight, 1998), with commodities being exchanged and labor being offered in return for commodities. The desperation also led to an increase in both private giving, on the one hand, as people responded gen- erously to those in need and thefts, on the other, by those unable to meet their food requirements (Stites et al., 2005). Yet, because of the uneven pressure and the permeability of the hold, the situation never tipped over into a famine system.

5. Conclusion

Several of the potential crises today have worrying resonances with these historical and contemporary examples, underlining the importance of better understanding the formation and evolu- tion of famines. This paper has proposed an analytical model con- sisting of five elements – pressure, hold, self-reinforcing dynamics, famine system, and rebalancing – that provides insight into the development of these crises and that could have several significant implications for future policy:

� Identifying situations that have the potential to develop into fam- ines. Using the famine systems model, it should be possible to identify situations across the world in which severe pressure combined with a hold and self-reinforcing dynamics signal that a famine might be developing. As illustrated by the diverse crises in Malawi, Greece, and Somalia, the elements will express themselves in different ways based on the specific circum- stances, but the combination of pressure, hold, and self- reinforcing dynamics will point to locations that require careful attention.

� Analysing the likelihood of situations developing into famines. The concepts can then help sharpen our analysis of whether crises exhibiting these characteristics will actually evolve into

154 P. Howe / World Development 105 (2018) 144–155

famines. The concern about Bosnia arose because of the combi- nation of pressure, a hold, and self-reinforcing dynamics, but the permeability of the hold and the resilience of the popula- tion, which lessened the pressure, partly explained why the situation did not deteriorate into a famine. Similarly, self- reinforcing dynamics were a common seasonal occurrence in Somalia, but it was in locations that experienced a strong, impermeable political hold that a famine system developed.

� Responding to the potential occurrence of famines. In places where famines are likely to develop, the key to an effective response is to remove pressure, break the hold, and rebalance the self- reinforcing dynamics. But different situations will involve dif- ferent degrees of difficulty and may require different approaches. In Greece and Somalia, the strong, impermeable, and fundamentally political holds made them difficult to break and required complex advocacy and actions, which were not effective until a famine system formed. In Bosnia, by contrast, efforts were undertaken earlier and more successfully. In Malawi, where natural and policy holds were in place, there were fewer challenges, and with appropriate awareness, the cri- sis could have been averted.

� Understanding the differential impacts of famines. Once a famine system is formed (and even earlier), there will be ‘bands’ of pop- ulations that are particularly affected, because of economic, social, political, environmental, demographic, and biological factors. In some cases, the most vulnerable bands may seem counterintuitive (e.g. young men and city dwellers in Greece) or may have been overlooked in the past (e.g. Reewin and Bantu minority groups in Somalia). However, identifying these ‘bands’ is critical to preventing undernutrition and mortality in a crisis.

� Preventing famine in the longer-term. The model points to two broad ways to prevent famine. The first is to focus on addressing the specific elements of the model. For example, reducing a population’s vulnerability by strengthening resilience is one important way to limit the pressure. At the same time, a range of possible actions – from political advocacy to policy shifts – can help ensure that disruptive factors and holds are not per- mitted to remain in place for extended periods. In Bosnia it was a combination of resilience and breaking the hold that pre- vented the development of a famine. The second is to use the model as a whole to enhance accountability for the creation of famine systems. By bringing greater clarity to the formation and evolution of these crises, the model may support these broader efforts and thereby help shift priorities to famine pre- vention in the long-term.

Famines have haunted humanity’s past and now our present.6

Yet, with a better understanding of the development of these crises and a vocabulary to discuss their evolution, there may be more pos- sibilities to identify and respond to these systems before they become part of our future.

Conflict of interest

None.

Acknowledgements

I would like to thank the anonymous reviewers for their very helpful comments and suggestions on an earlier draft of the paper and the staff of the World Food Program Library and Research Cen- ter for their patient support of my research. The interpretations

6 This line echoes the opening sentence of the Preface of Hunger in history (Newman, 1990).

and conclusions are solely my own and do not necessarily reflect the views of the World Food Program or the United Nations’ Mem- ber States. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

References

Aykroyd, W. (1974). The conquest of famine. London: Chatto and Windus. Becker, J. (1996). Hungry ghosts: Mao’s secret famine. New York: The Free Press. Checchi, F., & Courtland Robinson, W. (2013). Mortality among populations of

southern and central Somalia affected by severe food insecurity and famine during 2010–2012. Rome and Washington, D.C.: Mimeo.

Corbett, J. (1988). Famine and household coping strategies. World Development, 16 (9), 1009–1112.

Curtis, P. (1995). Urban household coping strategies during war: Bosnia- Hercegovina. Disasters, 19(1), 68–73.

Dando, W. (1980). The geography of famine. New York, NY: V. H. Winston & Sons. de Waal, A. (1989). Famine that kills: Darfur, Sudan, 1984-1985. Oxford: Clarendon. de Waal, A. (2017). The end of famine? Prospects for the elimination of mass

starvation by political action. Political Geography, 2017, 1–12. https://doi.org/ 10.1016/j.polgeo.2017.09.004.

Devereux, S. (1993). Goats before ploughs: Dilemmas of household response sequencing during food shortages. IDS Bulletin, 24(4), 52–59.

Devereux, S. (Ed.). (2007). The new famines: Why famines persist in an era of globalization. Oxon: Routledge.

Dolot, M. (1985). Execution by hunger: The hidden holocaust. New York, NY: W.W. Norton & Company.

Duffield, M. (1994). The political economy of internal war: Asset transfer, complex emergencies and international aid. In J. Macrae & A. Zwi (Eds.), War and hunger. London: Zed Books.

Dyson, T. (2002). Famine in Berar, 1896–7 and 1899–1900: Echoes and chain reactions. In T. Dyson & C. O. Grada (Eds.), Famine demography: Perspectives from the past and present (pp. 93–112). Oxford: Oxford University Press.

Garenne, M., Waltisperger, D., Cantrelle, P., & Ralijaona, O. (2002). The demographic impact of a mild famine in an African city: The case of Antananarivo, 1985–7. In T. Dyson & C. O. Grada (Eds.), Famine demography: Perspectives from the past and present (pp. 204–217). Oxford: Oxford University Press.

Garnsey, P. (1988). Famine and food supply in the Graeco-Roman world: Responses to risk and crisis. Cambridge: Cambridge University Press.

Glantz, M. (Ed.). (1987). Drought and hunger in Africa: Denying famine a future. Cambridge: Cambridge University Press.

Hendrickson, D., Armon, J., & Mearns, R. (1998). The changing nature of conflict and famine vulnerability: The case of livestock raiding in Turkana district, Kenya. Disasters, 22(3), 185–199.

Hionidou, V. (2006). Famine and death in occupied Greece, 1941–1944. Cambridge: Cambridge University Press.

Hionidou, V. (2002). ‘Send Us Either Food or Coffins’: The 1941–2 Famine on the Aegean Island of Syros. In T. Dyson & C. O’Grada (Eds.), Famine demography: Perspectives from the past and present (pp. 181–203). Oxford: Oxford University Press.

Howe, P. (2003). Contesting ‘Famine’: A study of conceptual ambiguities and their implications for response and accountability in Southern Sudan, 1998 (Unpublished doctoral dissertation). Brighton: Institute of Development Studies, University of Sussex.

Howe, P. (2010). Archetypes of famine and response. Disasters, 34(1), 30–54. Howe, P., & Devereux, S. (2004). Famine intensity and magnitude scales: A proposal

for an instrumental definition of famine. Disasters, 28(4), 353–372. IPC (2017). IPC: Integrated Food Security Phase Classification. Retrieved from http://

www.ipcinfo.org/home/en/. Keen, D. (1994). The benefits of famine: A political economy of famine and relief in

southwestern Sudan, 1983–1989. Princeton: Princeton University Press. Knight, T. (1998). Bosnia-Herzegovina: Basic themes from a complex emergency.

Retrieved from https://www1.essex.ac.uk/armedcon/story_id/ Bosniacomplexemergency.pdf

Macrae, J., & Zwi, A. (1994). War and hunger: Rethinking international responses to complex emergencies. London: Zed Books.

Majid, N., & McDowell, S. (2012). Hidden dimensions of the Somalia famine. Global Food Insecurity, 1(2012). Retrieved from http://www.fao.org/fileadmin/user_ upload/drought/docs/Majid%20McDowell%20Hidden%20Dimensions%20Somali %20Famine.pdf.

Marcus, D. (2003). Famine crimes in international law. The American Journal of International Law, 97(2), 245–281. Retrieved from http://www.uio.no/studier/ emner/hf/iakh/HIS4319/h16/famine-crimes-in-international-law.pdf.

Maxwell, D., & Fitzpatrick, M. (2012). The 2011 Somalia famine: Contexts, consequences, and complications. Global Food Security, 1(2012), 5–12.

Maxwell, D., Haan, N., Gelsdorf, K., & Dawe, D. (2012). The 2011-12 Famine in Somalia: Introduction to the Special Edition. Global Food Security, 1(2012), 1–4.

Maxwell, D., Majid, N., Adan, G., Abdirahman, K., & Kim, J. (2016). Facing famine: Somali experiences in the famine of 2011. Food Policy, 65(2016), 63–73.

McCloskey Rebelo, E., Prost, M., Renk, S., Guduri, S., & Hailey, P. (2012). Nutritional response to the 2011 famine in Somalia. Global Food Security, 1(2012), 64–73.

Menkhaus, K. (2012). No access: Critical bottlenecks in the 2011 Somali famine. Global Food Security, 1(2012), 29–35.

P. Howe / World Development 105 (2018) 144–155 155

Natsios, A. (2001). The great North Korean famine: Famine, politics, and foreign policy. Washington, DC: United States Institute of Peace Press.

Newman, L. (Ed.). (1990). Hunger in history: Food shortage, poverty, and deprivation. Oxford: Basil Blackwell.

Oxfam (2001). Report of an inter-agency workshop to discuss minimum standards for food security in disaster response. Oxford: Oxfam.

Dyson, T., & O’Grada, C. (2002). Famine demography: Perspectives from the past and present. Oxford: Oxford University Press.

Plumper, T., & Neumayer, E. (2009). Famine mortality, rational political inactivity, and international food aid. World Development, 37(1), 50–61.

Rangasami, A. (1985). Failure of exchange entitlement’s theory of famine: A response. Economic and Political Weekly, 20(41), 1747–1752. 1797–1801.

Ravallion, M. (1997). Famine and economics. Journal of Economic Literature, XXXV (September 1997), 1205–1242.

Rubin, O. (2009). The merits of democracy in famine protection – fact or fallacy? European Journal of Development Research, 21, 699–717.

Rubin, O. (2016). Understanding famine. In Contemporary famine analysis. Retrieved from file:///C:/Users/paul.howe/Downloads/9783319273044-c2%20 (2).pdf.

Scoones, I. (1998). Sustainable rural livelihoods: A framework for analysis. IDS Working Paper 72. Brighton: Institute of Development Studies.

Sen, A. (1981). Poverty and famines: An essay on entitlement and deprivation. Oxford: Clarendon Press.

Stites, E., Lautze, S., Mazurana, D., & Anic, A. (2005). Coping with war, coping with peace: Livelihood adaptation in Bosnia-Herzegovina, 1989–2004. Retrieved from http://fic.tufts.edu/assets/Coping-With-War.pdf.

United Nations Administrative Committee on Coordination/Sub-Committee on Nutrition [UN ACC/SCN]. (1997). 3rd Report on the World Nutrition Situation. Retrieved from https://www.unscn.org/web/archives_resources/files/rwns3. pdf.

Vaughan, M. (1987). The story of an African famine: Gender and famine in twentieth century Malawi. Cambridge: Cambridge University Press.

Walker, P. (1989). Famine early warning systems: Victims and destitution. London: Earthscan Publications.

Watson, F. (2007). Why are there no longer ‘war famines’ in contemporary Europe?: Bosnia besieged, 1992–1995. In S. Devereux (Ed.), The new famines: Why famines persist in an era of globalization (pp. 269–289). Oxon: Routledge.

Watts, M. (1983). Silent violence: Food, famine and peasantry in northern Nigeria. Berkeley: University of California Press.

Wisner, B., Blaikie, P., Cannon, T., & Davis, I. (2003). At risk: Natural hazards, people’s vulnerability and disasters (2nd ed.). Oxon: Routledge. Retrieved from http:// www.preventionweb.net/files/670_72351.pdf.

Young, H., & Jaspars, S. (1995). Nutrition, disease and death in times of famine. Disasters, 19(2), 94–109.