The 2015–2016 famine threat in Ethiopia: a study of the relevance of famine archetypes

In recent years, Ethiopia has experienced fast economic growth and has been a major recipient of development and humanitarian aid. However, these developments were unsuccessful in eliminating food insecurity problems, and Ethiopia continues to be a con siderably famine-prone country. The aim of this paper is to examine the applicability of Howe’s framework of the six archetypal situations symptomatic to famines (watch, price spirals, aid magnet, media frenzy, overshoot, and peaks) to the 2015–2016 food crisis, which left an estimated 15 million Ethiopians in need of acute food assistance. This paper observes that the food crisis proved to have some of Howe’s archetypes including watch, price spirals, and, to a lesser extent, media frenzy and peaks. Even though the aid magnet and overshoot were not recorded, the dynamics of the 2015–2016 food crisis confirmed Howe’s argument that the current system of humanitarian assistance does not lead to timely and effective responses. In this paper, I also argue that the Ethiopian political context further exacerbates the food insecurity situation of the country.


Introduction
In 2015, East and South Africa were hit by one of the worst droughts in more than 50 years, caused by the El Niño phenomenon. In Ethiopia, two rain seasons brought below-average rainfalls and over 15 million people were in need of acute food assistance in 2016 (FEWSNET 2015a). Some Ethiopian regions were classified as 'emergency' by the Famine Early Warning Systems Network (FEWSNET), which is the phase preceding a full-scale famine according to the FEWSNET classification (FEWSNET 2015b). It implies that the country was facing high risk of mortality and in acute need of humanitarian assistance. The World Health Organization predicted that the situation would worsen until August 2016 and the country would need at least another year to recover from the crisis (WHO 2015). This was already the fifth extensive food crisis since 1984, of which two were declared famines. In December 2016, the number of Ethiopians suffering from undernourishment had reached 28.8% in the 3-year average from 2014 to 2016 (FAO 2017). Paradoxically, Ethiopia experienced strong economic growth with 10.8% between 2003 and 2014 and a slight drop to 8% in 2015 to 2016 (WB 2017). At the same time, Ethiopia is one of the major recipients of development aid (OECD 2016).
This paper examines the applicability of Howe's (2010) six archetypal situations recurring in famines to a 'famine threat'. These archetypes are known as watch, price spirals, aid magnet, media frenzy, overshoot, and peaks. Scrutinizing the presence or absence of factors that underlie a transition from food crisis to famine should help to understand whether they would be relevant in a situation of a severe food emergency not translated into famine. Thus, the contribution of this paper is to appraise the relevancy of these archetypes in non-famine situations which might help humanitarian actors to better assess and react to upcoming food insecurity emergencies. The paper collates various secondary data and information to discuss whether and how these situations took place during the 2015-2016 crisis. The article has the following structure: Section 2 discusses conceptual issues. Section 3 addresses data and methods. Section 4 briefly outlines the history of Ethiopian famines and their patterns. Section 5 provides the main analysis of the 2015-2016 famine threat then examines the crisis origins, food price evolution, movement of population, world media attention, and the humanitarian response. Section 6 discusses the findings and summarizes the main conclusions.

Conceptualizing famines
Famine, as an ultimate demonstration of food insecurity, is characterised as high mortality linked to starvation. According to the UN Food and Agricultural Organization (FAO) definition, a famine is declared when at least 20% of households within an affected area face extreme food shortages; acute malnutrition reaches 30% and mortality exceeds two persons per day per 10,000 people (FAO 2008). However, a common definition has not been approved by the international community due to the temporal, scale and sectoral ambiguities and uniqueness that every famine comprises (Howe 2002). For example, there was no expert agreement on the extent of the crises in Sudan in 1998, Ethiopia in 1999to 2000to 2003 or Malawi in 2002 where famines were not officially declared by the UN even though the countries were hit by severe food emergencies. Whether a famine is officially recognized or not it brings political and emotional connotations. A delayed declaration of famine may lead to a delayed response which may have fatal consequences. A famine definition and conceptualization is thus needed for both practical and political reasons (Howe, Devereux 2004).
Early theories considered famine a consequence of food production failures due to natural disasters. This view was challenged by Amartya Sen who highlighted the critical importance of access to food related to various forms of socioeconomic inequalities. He explained famines by referring to the absence of 'food entitlements' in terms of means and resources to secure food rather than food unavailability as such (Sen 1981a). In subsequent literature, two major famine conceptualizations can be distinguished with one referring to the food availability decline and another referring to the food entitlement decline (Devereux 2009). These two aspects also represent two main pillars of the food security definitions adopted at the World Food Summit (1996). Congruently, a major shift in the focus of famine debates has been the consideration of response failure theories. It accentuates the role of governments and political representation to prevent large-scale food emergencies in the modern era. As Devereux (2007) puts it, new famines are always preventable, so their nature is necessarily political. The persistence of famine in a globalized world signifies political failures of states and the international relief system (Devereux, Howe, Biong Deng 2002).
Various factors can lead to a famine outbreak. These include natural disasters, economic shocks, international and internal conflicts or other political failures. In practice, however, a single cause is often emphasized in the official depiction of famines such as, typically, drought or other natural disasters. A narrow focus on natural causes or immediate triggers and insufficient attention paid to other drivers hinders development actors from defining complex policies and responses (Sandstrom, Juhola 2017). In order to complement early warning systems and reach effective and appropriate responses, Howe and Devereux (2004) suggested a new approach to defining famine based on intensity and magnitude scales. 'Intensity' refers to the severity of the crisis, and 'magnitude' refers to the scale of the crisis. Howe and Devereux (2004) designed a graduated concept of famine to avoid 'a binary conception of famine / no famine' and allow a 'minor famine' to be declared even without high mortality. The authors also stress that stakeholders should concentrate more on prevention policies to reduce large-scale food emergencies.
Equally, famine represents deep social disruption (Rivers et al. 1976). Humanitarian and non-governmental organizations tend to attract international attention in early stages of a food crisis, but a needed reaction usually comes only after shocking images appear in popular media (Hammond, Maxwell 2002). This demonstrates the tremendous power international media can have in international relations. The way a state responds to famine depends mainly on two factors: the ruling regime's political priorities and their relationship with potential donors (Cutler 1991). Donor countries also consider the political situation of the affected country and their geopolitical interests. There is a high probability that aid may not get into the hands of those who really need it if there is ongoing civil conflict or an undemocratic rule. Conversely, the likelihood of famine or starvation on a massive scale is relatively low in a well-functioning state thanks to different protection mechanisms such as early warning systems, social protection programs, regional food reserves or crop insurance.
As Edkins (2002) points out, there are tendencies to comprehend famine as a failure which can be solved solely with technological solutions. Whether deliberate or intentional, Edkins writes that the responsible actors should be brought to justice and political actions should come first to prevent famine. It is no coincidence that the severest food crises occur in war-torn countries and authoritarian or dictatorship regimes. Marcus (2003) also calls for famine criminalization within international law. He defines four types of government behaviour. First, actions of incompetent and corrupt politicians do not respond to a population's acute needs. Second, government activities can be labelled as indifference to the current emergency. Third, policies are dominated by recklessness despite direct impacts on starving people. Forth, hunger is used deliberately as a political weapon to exterminate regime enemies. This last argument underlines intentional behaviour and can be applied to the case of Somalia in 2011 when famine was declared in several regions controlled by the radical group Al-Shabaab which forbade most international aid agencies to provide humanitarian aid (Menkhaus 2012).
In practice, there are several safeguarding mechanisms to alert food insecurity crises such as the Global Information and Early Warning System (GIEWS) managed by FAO and FEWSNET developed by the US Agency for International Development (USAID). GIEWS provides an information analysis of crop prospects and monitoring, markets and trade, price and policy, supply and demand balance, vulnerability and risk. FEWSNET focuses on weather and climate, markets and trade, agricultural production, livelihoods, nutrition, and food assistance. It identifies the following five food insecurity phases as distinguished in the Integrated Food Security Phase Classification (IPC) of Global Partners (FAO 2012): 1) minimal, 2) stressed, 3) crisis, 4) emergency and 5) famine. FEWSNET predicted the food crisis in East Africa in 2011 six months in advance and has thus been labelled an 'excellent and timely tool' and better than GIEWS which has a lower predictive ability (Ververs 2012).
Arguing against the linear view of the famine process, Howe (2010) analyses larger patterns within famines through the lenses of systems thinking analysis. He seeks analogies between the basic elements of the systems approach, such as reinforcing feedback loops, balancing loops, delays, and leverage, to understand specific situations or archetypes of famines. The 'watch' situation is when it is difficult to assess whether a food crisis will develop into an absolute food emergency. Such a situation is usually linked to poor food production and inadequate coping strategies of the affected people. The watch situation is followed by 'prices spirals' which is when food prices increase due to lack of food in markets. The third situation is called 'aid magnet' which refers to the fact that aid centres attract large inflows of hungry people resulting in their capacity being exceeded. Overcrowding then leads to poor sanitation conditions and the needs of the hungry population may not be met due to limited food resources. Subsequent higher mortality provokes the fourth situation 'media frenzy' where media outlets compete for the best stories and images. However, media attention comes too late to prevent the emergency. The fifth situation 'overshoot' relates to ramping up of assistance after the most severe period of the crisis has passed. The sixth situation 'peaks' combines all five situations and refers to successive spikes characterising famines and responses. However, Howe concedes that 'many situations will not conform neatly to the archetypes'.
Howe argues that the current setting of humanitarian aid does not provide timely and effective responses because aid does not come until after all peaks take place. Howe puts emphasis on system thinking. Even though he does not precisely quantify the specific archetypes (such as the increase of food prices, number of fleeing people to relief sites or number of media input to be considered as media frenzy), he leaves the space for specific conditions, specific situations, and specific considerations. However, this approach has not been applied to analyse a concrete case (Rubin 2018). Later, Howe has expanded this approach by introducing five basic elements characterising formation and evolution of a famine: pressure (on food security), hold (worsening situation), self-reinforcing dynamics (pattern on accelerated changes), famine system, and rebalancing (Howe 2018). Both Howe's works create a coherent framework and a convenient tool for analysis of famines in practice and for further research. On the other hand, it is not clear if these archetypes would be relevant to a situation of famine threat or a severe food emergency.

Data and methods
In this article, I collated information from various sources to examine the applicability of Howe's archetypes to the case of the 2015-2016 food crisis in Ethiopia. To assess the presence of the watch situation, reports and alerts of early warning systems and organizations working on spot can be examined. For the purposes of this article, FEWSNET monthly reports were used to create a time sequence of the issued emergencies and a framework of available information about worsening food insecurity in specific regions. For price spirals, I looked at three different indicators: 1) food inflation rates between 2014 and 2016 as provided by the World Food Programme (WFP); 2) staple food prices in the two most insecure regions in terms of food from January 2014 to December 2016 (data provided by FAO GIEWS) and 3) price spikes provided by the WFP Alert for Price Spikes (ALPS). As an indicative assessment of the aid magnet condition, I focused on the number of internally displaced Ethiopians in the period between 2014 to 2016 from the International Organization for Migration (IOM) and the Internal Displacement Monitoring Centre (IDMC). Distinguishing of the cause of home abandonment provides another insight into the peoples' motivations and another possible aspect of this archetype. The media frenzy was analysed through the outputs of the most world known English and French-speaking media and non-governmental organizations working in Ethiopia. The overshoot situation was examined through the humanitarian funding situation with a focus on humanitarian needs and funding gaps. For this purpose, data from the UN Office for the Coordination of Humanitarian Affairs (OCHA) was used. The assessment of the last archetype combined all of the abovementioned information.

Previous famines and food crises in Ethiopia
For a long time, Ethiopia has been a symbol of famine due to the affliction of recurrent famines since the 9th century. Their brief characteristic is described in Table 1. About one third of the population most likely died during the 1888-1892 famine (Sen 1981a). Since that, Ethiopia has experienced several severe food crises labelled as famines and there were similar patterns occurring during those famines. Table 1 suggests that poor rainfalls and subsequent droughts triggered the famines and food crises in Ethiopia. Lack of rain increases hunger incidences and the probability of crisis. Farmers' food stocks usually run out during lean seasons (periods of regular food insecurity and rises in food prices) before new harvests. Lean season usually lasts from April to June in Belg-receiving areas (listed famine-prone regions) and from February to April in pastoralist areas. In the lowlands, pasture and land availability decline thus making livestock products (such as milk) generally unavailable. Moreover, the Ethiopian government and often also international organizations did not respond adequately to recurrent food crises despite their apparent predictability. The Table 1 also indicates that similar Ethiopian regions in the Northeast, East and South of the country were hit the most. These are arid and semi-arid ecological zones with the prevalence of low rainfall, high temperatures and ongoing desertification (EPA 1998).
The 2015-2016 crisis recorded similar pattern like previous famines and food crises: trigger (subsequent waves of drought), geographical distribution and extent, reinforcing factors (high food prices) and late responses. Interestingly, no famine has materialised since the implementation of the Productive Safety Net Programme in 2005 implying the mitigation effects of the programme. On the other hand, Ethiopia still faces periodic food emergencies and decades of food aid have not had any substantial impact on hunger and poverty reduction (Sabates-Wheeler, Devereux 2010).

The 2015-2016 food emergency
Borrowing the structure of famine dynamics elaborated by Howe (2010), this section analyses how the food crisis cycle evolved during 2015 and 2016 due to the threat of famine outbreak.

Watch
During this period, early warning systems usually provide information about production and food deficit, but it is not yet clear whether the afflicted population will be able to cope with the upcoming situation.
Coping strategies include, for instance, a reduction in food consumption, sale of livestock and assets, seeking of off-farm jobs or even migration. However, this period plays a key role in famine prevention. Yet, there are two dangerous moments. First, possible underestimation of the severity level of the crisis may result in non-response of donors. Second, coping strategies may bring some short-term relief but usually prevent people from reacting to the future crisis while selling land, livestock or tools (Howe 2002).
In order to analyse the beginning and evolution of the food crisis in 2015 to 2016, monthly Famine Early Tab. 2 FEWSNET reports from January to June 2015.

Region
Jan 2015 (2015) Notes: Integrated Food Security Phase Classification (IPC): 1) minimal, 2) stressed, 3) crisis, 4) emergency and 5) famine (FAO 2012) Warning System reports are scrutinised as FEWSNET is considered to be a precise tool for averting emergencies (Ververs 2012). The article takes into account the whole period of 2015 which includes the uncertainty of crisis creation and time for coping strategies.
In reality, the watch period lasted for a shorter time. Table 2 shows that the FEWSNET expected food security to decline from stressed (IPC Phase 2) to crisis (IPC Phase 3) from January 2015 in the six affected regions: Afar, Amhara, Oromia, Tigray SNNPR and Somali.
A FEWSNET alert came in June 2015 calling for food emergency assistance. It said that '20 to 50 percent less rain than average between March and May, (…) planted area was less than 60 percent of average across SNNPR' (FEWSNET 2015c). In addition, the Ethiopian government declared the Belg rain failed in June (HRD 2016). Acute malnutrition and emergency food assistance needs were also predicted in the following months. In October 2015, several regions were expected to be at risk of or to move to an emergency (IPC Phase 4). Another alert was issued in December 2015 calling for 'immediate sustained, large-scale, multi-sectoral emergency assistance to save lives and livelihoods' (FEWSNET 2015a). Table 3 shows the prediction of crisis severity change.
The FAO GIEWS system provided information about deteriorating food security conditions in June 2015 observing that major problems would be felt in central-northern SNNPR and 'the estimated number of people in need of relief food assistance is expected to increase significantly' (FAO GIEWS 2015). The Figure 1 shows the IPC phases in the food security outlook from October to December 2015 provided by FEWSNET.
It is also important to highlight that the food security situation was deteriorating in the western regions as well, especially due to violence and conflicts in neighbouring countries. By the end of 2015, Ethiopia hosted more than 730 thousand refugees and asylum-seekers from South Sudan, Sudan, Somalia and Eritrea. Most of the refugee camps were located in the Gambella and Benishangul-Gumuz regions. Deficiency in financial means had a strong negative impact on humanitarian conditions in these camps resulting in tensions between local Ethiopians and the incoming people (FAO GIEWS 2015).

Price spirals
For the price spirals archetypes, it is important to consider the fact that poor households in less developed countries spend as much as 60-80 percent of their incomes on food (WFP 2012), so their vulnerability to rising food price is very high. According to the World Food Programme, food prices showed an upward trend compared to the long-term average in 2010 to 2014 (WFP 2015). Prices of different staples differed in particular regional markets, though. To assess this trend, I focused on food inflation rates from 2014 to 2016, prices of staple food in two regions covered by GIEWS: Mekele (Tigray region) and Dire Dawa (east of Ethiopia, on the border between Oromia and Somali), and data from the Alert for Price Spikes.
Usually, the highest food prices are recorded during regular food insecurity periods in the lean season (February to June). Food inflation was growing throughout the year in 2015 with two-digit values ( Figure 2) making food more expensive and thus less available for poor households. The food inflation peak in December 2015 corresponds with a second failed harvest that year. There was a considerable decrease in inflation in 2016.
In relation to food prices, the GIEWS tool focuses on two local markets in two considerably food-insecure regions: Mekele and Dire Dawa (Figure 2). According to the FEWSNET (2015d), the most consumed cereal by the poor in rural areas of Ethiopia is maize, the cheapest is white sorghum, and teff (a local crop) is a very important cereal throughout the country. A significant part of the local diet is also made up of white wheat. Thus, Figure 2 focuses on these crops in order to ascertain price instability which generally further exacerbates food insecurity.
In Mekele, maize recorded the most stable price while the price of sorghum almost doubled in the 18 months following January 2015. The data also disproves the FEWSNET claim that sorghum is the cheapest staple food in Ethiopia. Teff was the most expensive crop, experiencing a steady rise in price since 2015. In Dire Dawa, the price of maize was a little higher than in Mekele but it recorded a stable level. The price of sorghum was about 30% higher than in Mekele in 2015, reaching the same level by April 2016. The price of wheat recorded the highest level in both markets. Information about the price of teff is not available for Dire Dawa.
For the detection of abnormally high food prices, the Alert for Price Spikes focuses on the gap between observed prices and long-term seasonal trends. It characterizes a situation as 'normal', 'stress', 'alert' or 'crisis' (WFP 2017). The following five markets in food-insecure regions (defined by the FEWSNET  Howe (2010)   Source: IDMC (2018) capacity and delivery of services prevents excess mortality in humanitarian situations. There were two causes of the new internal displacement in Ethiopia. In 2015, 104 thousand people left their home due to disasters (drought or sudden flooding) and 56 thousand were on the move due to a conflict. Sixty seven percent of the internally displaced were from the Somali region (predominantly pastoralist), 14% were from Oromia, 11% from Afar and the rest from Gambella, Amhara, SNNPR and Hareri (IOM 2016). In 2016, Ethiopia experienced a sharp rise in internal displacement. The number of people leaving their homes due to disasters rose more than three times to 347 thousand and more than five times to 296 thousand due to a conflict (Figure 4).

Media frenzy
The media frenzy archetype is characterised as competition amongst the media and substantial public interest following exposure to a horrific image from famine hotspots (Howe 2010). For the purposes of this article, the following English and French speaking best-known world media were analysed: The Guardian, BBC, New York Times, Washington Post, Reuters, CNN, Al Jazeera, Le Monde and Radio France International (RFI). The choice was based on my language skills and on testing the presence of the topic within the concrete journal, television or radio. I searched for the words Ethiopia, food crisis, famine, drought, 2015 and 2016 in news and opinion articles and only articles focusing on drought and upcoming food crisis published online were considered. A timeline was formed, and the frequency of coverage was created. Also, I looked for the international non-governmental organizations working in Ethiopia to discover their possible influence on media coverage of the food crisis after their warnings and calls for emergency aid. These were OXFAM, Save the Children, Care International, and Catholic Relief Services. World media news about the drought and its consequences for food security in Ethiopia started to appear in August 2015 (The Guardian, RFI). Clear links to the 1984 famine were made and a famine threat was pictured as real even though famine was not predicted by any early warning system. The number of news and NGO media outputs is indicated in Figure 5.
There was no media coverage on the food crisis in Ethiopia in the first half of 2015. The first media response corresponds with the FEWSNET monthly report in October when an emergency (Phase 4) was predicted and the Ethiopian government asked for international assistance from donors. In November 2015, the media focused on weather conditions and Ethiopia's capacity to respond. In December, FEWS-NET issued the alert for emergency assistance and non-governmental organizations published most of their stories, however, media attention was already decreasing. The Ethiopian government asked for foreign aid in January 2016. In March 2016, the media reacted to a United Nations report on the humanitarian situation in Ethiopia and in May 2016, world media publicized the FEWSNET severe food crisis prediction. After August 2016, the Ethiopian food crisis disappeared from international news and did not reappear until March 2017.

Overshoot
Donor countries may be hesitant to provide aid in the beginning of a crisis which may lead to an overshoot of emergency responses when too much aid arrives at the end of a crisis. It also takes several months for emergency aid to come as the process involves preparation and examination of proposals, fund releasing and aid transportation (Howe 2010). In Ethiopia, the UN Office for the Coordination of Humanitarian Affairs (OCHA 2015) uses the Hotspot woreda (administrative unit) classification to prioritize response, especially in supplementary feeding. The number of areas needing urgent nutrition support quadrupled between February and December 2015. Woredas with Priority 1 increased from 40 in February to 97 in May, and from 142 in August to 196 in December.
There were 10 million people in need of aid in November 2015. This number increased to more than 15 million in 2016 (OCHA 2016), including Ethiopians who receive food and cash transfers through the Productive Safety Net Programme. From a geographical point of view, most of the people in need of food assistance were in the Oromia region (3.7 million), followed by the Amhara (2.2 million), Somali (1.5 million) and Tigray regions (1.2 million) (HRD 2016).
According to the OCHA, there was a significant gap (Table 4) in required humanitarian funding throughout both years, especially in food assistance but also in other sectors such as health, nutrition, water and sanitation, agriculture and education. The greatest emergency was recorded after the failure of Kiremt rains and the subsequent Meher harvest in the autumn of 2015 when humanitarian needs skyrocketed by more than three times and donors released their aid with a delay of several months. In this case, there was clear evidence of the emergency.

Available (millions $)
Jan 2015  The United Nations provided humanitarian aid through the Central Emergency Response Fund CERF and specialized agencies such as the World Food Programme, the Food and Agricultural Organization and UNICEF ($ 27 million in 2015/$ 20 million in 2016). Aid was also provided, to a lesser extent, by private donors or non-governmental organizations ($ 17 million in 2015/$ 28 million in 2016) (OCHA 2016b).

Peaks
This archetype combines all successive spikes characterising famine. Howe (2010) concentrates on peaks of death, peak of food prices, peak of media attention and peak of deliveries. In the case of the 2015-2016 crisis, following peaks can be identified: food price inflation, migration (progressive in 2016) and media attention (two peaks). There was no peak of death related to aid magnets (or internal migration) and specific peak of aid deliveries.

Discussion and conclusion
The findings presented in the previous section have several implications. First, the FEWSNET was able to predict the impending emergency and as a result issued two alerts in 2015 (in June and December) calling for immediate emergency assistance. There were two subsequent periods of failed rains in 2015 which intensified the food insecurity of affected people not only after the lean season in June but also due to the limited Meher harvest which started in October. By July 2015, it was clear that the food crisis would turn into an emergency in the absence of external humanitarian aid and that the coping strategies would not be sufficient. The watch situation was therefore rather short, lasting only the first half of 2015. From June onwards, it was evident that the crisis would continue to deepen. Late responses cannot be attributed to the insecurity of the crisis evolution typical for the watch situation. Moreover, aid donors are long familiar with the situation in Ethiopia. Thus, subsequent waves of drought, warning reports and chronic food insecurity were clear signs for the Ethiopian government and the international donor community to respond on time.
Second, rises in food prices were recorded during the crisis. In 2015, food inflation reached two-digit values with the peak being in October (usual start of harvesting and time of low prices) and decreasing in 2016. The year 2015 was exceptional in this regard in comparison with 2014 and 2016. Local markets in the two most food-insecure regions recorded a significant rise in prices of sorghum (100% rise) and teff in the Mekele market. Wheat prices rose steadily from the end of 2014 to mid-2016. The Alert for Price Spikes issued two alerts in both lean seasons of 2015 and 2016 when prices are usually high. Thus, Howe's price spike archetype was recorded without a specific peak of all examined indicators.
Third, regarding the aid magnet situation, there were no excessive mortality rates in aid centers thanks to long-term operations of the UN humanitarian organisations. On the other hand, there was a sharp rise in internal displacement in Ethiopia, especially in 2016. Although Howe (2010) focuses on people's motivation to leave their homes (aid provision), this article highlights the causes of internal migration (disasters and conflicts). It cannot be unambiguously stated that internal migration was solely triggered by aid provision as displacement in Ethiopia was a present phenomenon. Displacement is caused also by competition over production resources and clashes between communities. In addition, extreme vertical hierarchies in the society reinforce geographic, economic and political marginalisation (Hammond 2011) and land confiscation or redistribution by regional authorities or federal government discredits traditional customary law (Abbink 2006). Land tenure stands for a highly sensitive political issue. During the examined food crisis, Ethiopia experienced political turmoil caused by ethnic and land conflicts. Protests in the Oromia region spread due to increasingly sensitive power divisions along ethnic lines (BBC 2014). Demonstrations against displacement of Oromo farmers across Oromia were suppressed with a disputed number of victims throughout 2015 when security forces were using 'unlawful force' and 'arbitrarily arrests' (HRW 2016). Other protests against unequal distribution of power and economic benefits spread to Amhara andSNNPR in 2016 (HRW 2017). In order to subdue the 'anti-peace forces', the government declared the state of emergency in October 2016 (BBC 2017). It is important to emphasize that these regions were hit the most by the 2015-2016 food crisis because access to land and production resources represents an indivisible part of food production and food security. Its loss and internal conflict have severe consequences on the state of food insecurity and they further deepened the discussed food crisis.
Fourth, media frenzy was proved. World media reacted to the prediction of FEWSNET in October when the food crisis was already ongoing. There were two media output peaks (October/November 2015 and March 2016) but otherwise media attention went down gradually. Even though a severe emergency was foreseen, there was no official prognosis of famine outbreak by any warning system. We can say that the famine threat was rather a media bubble to catch attention but, in any case, this claim should not underestimate the situation perceived by some Ethiopians themselves whose stories, when presented in the media, were reminiscent of the year 1984. Also, no other links to the crisis were made and drought caused by the El Niño confirmed the generally narrow point of view in relation to food crises. The most active media outlets were the Guardian, Reuters and the BBC followed by RFI and Le Monde. RFI also informed about the Ethiopian government's efforts to make press and non-governmental organizations remained silent to not spoil the picture of a booming economy or associate the country with misery (RFI 2015a). RFI also claimed that some drought-affected areas were almost impossible to reach by foreign journalists (RFI 2015b). The political influence on media and the non-governmental sector over providing accurate information represents a major hindrance and unacceptable intervention in crisis management. In this case, the media were able to report on the crisis despite Ethiopian efforts to conceal the reality. This fact supports the argument of Devereux (2007) about the political nature of famines which applies to severe food crises as well.
Fifth, humanitarian needs in Ethiopia skyrocketed in autumn 2015 but most of the aid came with several month-delays in the first quarter of 2016. There were also significant gaps in funding. The greatest gap was recorded in December 2015 when the total requirements reached $ 1.4 billion and the gap was $ 1.042 billion. This fact suggests that Howe's situation of overshoot emergency response was recorded during this food crisis but on the other hand, it does not confirm the condition that there was 'too much aid too late'. It is not clear whether ramping up aid without fulfilling all humanitarian needs suffices to realisation of this archetype. The response rose in connection with growing requirements in 2016 but funding gaps persisted. Personally, I am inclined to think that the overshoot archetype is less applicable in this case. The late response might be attributed to donors' stretched budgets due other urgent crises (Syria, Yemen, Iraq or South Sudan) or their fatigue. It is also important to emphasize, however, that the funding was adequate to prevent a full-scale famine. What seems to be problematic is the aid structure. As Sandstrom and Juhola (2017) point out, most of the humanitarian budget is usually comprised by food aid as it was in the case of the Ethiopian food crisis in 2011 (58%) and during the emergency in Kenya andSomalia in 2004 to 2005 (over 80%). Numbers from the 2015-2016 food crisis also confirm this trend with food aid ranging from 68% to 86%. A stronger focus should be put on other sectors linked to food insecurity such as health, nutrition, education, agriculture or water, sanitation and housing.
Sixth, the dynamics of the food crisis of 2015 to 2016 does correspond to Howe's peaks archetype, to a lesser extent. The most urgent need of humanitarian assistance can be identified in the period of autumn and winter 2015. A rise in food prices as well as internal displacement was rather progressive processes. Media attention was experienced in two peaks and there was no peak in deliveries. Yet, the progressive characteristic of the crisis suggests a long-term food crisis in Ethiopia. It is also necessary to highlight that Ethiopia suffered from another wave of drought exacerbated by disease outbreaks and displacement in 2017. Eight and a half million Ethiopians faced crisis-level food insecurity (OCHA 2017). Parts of Amhara, Tigray, Oromia and SNNPR remained in IPC Phase 3 in September and an emergency (IPC Phase 4) situation was recorded in parts of the Somali region with delayed humanitarian assistance (FEWSNET 2017). The humanitarian needs of the country were $ 1.471 billion ($ 893 for food) with a $ 267 million gap by the end of 2017 (OCHA 2018).
There was no famine outbreak but on the other hand, famine was never officially predicted by any warning system. Application of the Howe's (2010) archetypal framework suggests its relevance in famine threat situations, although the magnitude of particular archetypes and their peaks may be more significant in full-scale famines. Even though not all archetypes took place, the dynamics of the 2015-2016 crisis proved Howe's argument of belated responses to crises. Furthermore, the tendency to see the failure of food systems only due to negative weather conditions with no acknowledgment of the complexity of food insecurity leads to limited impacts of humanitarian activities. There is an acute need for changes in humanitarian thinking towards more systemic and complex solutions. As the political context deepens the food crisis in Ethiopia, chronic food insecurity and the probability of future food emergencies in the country remain very high.