The easy explanation

(1)

Doctoral Programme in Interdisciplinary Environmental Sciences

University of Helsinki

EXPLORING THE LINK BETWEEN DROUGHT AND FOOD SECURITY IN THE EAST AND HORN OF AFRICA

Sofie Sandström

ACADEMIC DISSERTATION

To be presented, with the permission of the Faculty of Biological and Environmental Sciences the University of Helsinki,

for public examination

New York 2021

(2)

ISBN 978-951-51-7806-0 (pbk.)

ISBN 978-951-51-7807-7 (PDF)

Helsinki University Printing House

Helsinki 2021

(3)

3

ABSTRACT

In this thesis, I explore the food security analytical basis of humanitarian responses and preparedness systems in the East and Horn of Africa, focusing on Ethiopia, Kenya, Tanzania and Somalia. The overall research question my work is centred on is whether the humanitarian response and preparedness system is based on a food availability approach or whether it has evolved to reflect a more complex food capability approach. To develop an understanding of the conceptualization of food security in the humanitarian sector in the region I focused on the main documents produced by the humanitarian assessment and appeal process, analyzing what causes were most prominently cited as driving food insecurity. The main finding of this research was that drought was by far the most common cause identified, and that there had been remarkably little change in this focus over the thirty-year period I reviewed.

The next part of the research was largely driven by this finding. I conducted a document analysis of the main products of the drought early warning system in Kenya to identify the drivers of food security used as indicators in the system and found that the system is heavily reliant on indicators related to food production, specifically climate-related indicators. My in-depth analysis of the use of rainfall in the products of the drought Early Warning System in Kenya, focusing on the county of Turkana, which is one of the most food insecure counties in Kenya, revealed an inconsistency in the methodology followed to establish a “drought year”. For the final part, I used statistical methods on secondary data to explore the barriers to uptake of climate information in food production decisions at the household level, including the gender differences in uptake and use of climate information, focusing on the districts of Longido and Kiteto in Tanzania. The research revealed very low levels of use of climate information for decisions related to food production, with some correlation between use of such information and access to productive assets. In addition, the findings indicated that the design of climate services were based on assumptions of pastoralist livelihoods not based on assessment data, and that reinforce gender inequality in focusing mainly on provision of climate information for the productive sectors that men have most influence over.

My assumption at the start of the research was that the food availability narrative would be found to be dominant at the regional policy and practice level, and that variables related to food production, specifically climate-related variables, would be central in the early warning system in Kenya. My research largely confirmed this assumption. The food availability approach is widely dominant and persistent at the regional level in the humanitarian community, and this narrative is informing practice at national and community levels. The system is still heavily reliant on indicators related to food production, particularly climate-related indicators, indicating that the preparedness and response system has not evolved over the past decade to reflect a more

(4)

complex understanding of food security, as reflected by the food capability approach. This has clear implications for policy and practice, as a more nuanced understanding of the complexity of food security that would allow for better targeted interventions.

(5)

5

ACKNOWLEDGEMENTS

I would like to dedicate this thesis to the memory of two former colleagues, who both inspired me and were crucial in making this work happen. Firstly, Josie Buxton, who for years lead the Oxfam programme in Turkana with a fierce commitment to alleviating human suffering and meeting humanitarian needs, a deep respect of the people of Turkana, their livelihoods and cultures, and a pragmatic and no-nonsense approach to the humanitarian and donor community. Josie made me start asking the “why” questions that all research originates from about the way aid was provided on a more or less permanent basis in Turkana. Secondly, I would like to honor the memory of Dr Elijah Mukhala, my manager at the World Meteorological Organization in Nairobi, who encouraged me to start pursuing a PhD, leading by example in his work ethic and pursue of further learning and education. Both Elijah and Josie passed away far too soon, and the lessons they taught me have guided me through this work.

Secondly, I would like to thank Professor Sirkku Juhola, whose guidance, patience, and good humor with me throughout this long-term project has been exemplary and allowed me to continue the work. I would also like to thank Dr Aleksi Räsänen, who has worked closely with me on this project and provided expert advice and inputs, which was particularly valuable in the editing stages of the articles and the synthesis.

Lastly, I would like to thank the person for whom no words could ever be enough to describe my love and who is also the person who reminds me to keep asking “why” – my daughter Elsa. She gives more than she ever takes and brings joy to every day I have been lucky enough to get to spend with her.

(6)

I Original idea is by SS and SJ who also developed the research design, framed the paper and supervised the research process. SS conducted the data collection. Data analysis was conducted by both authors and both authors participated in preparing the manuscript parts on methods, data and results with SJ leading the preparation.

II. Original idea is by SS. SS together with SJ developed the research design, framed the paper and supervised the research process. SS conducted the data collection. Data analysis was conducted by both authors and both authors participated in preparing the manuscript parts on methods, data and results.

III Original idea is by SS. SS together with SJ and AR developed the study design with SS conducting the data collection. Analysis was conducted by all three authors and all three participated in preparing the parts on methods, data and results.

The publications are referred to in the text by their roman numerals.

(8)

ABBREVIATIONS

EARLY WARNING SYSTEM – Early warning system GFCS – Global Framework for Climate Services IPC – Integrated phase classification

MAM – March, April, May

NDMA – National drought management authority OND – October, November, December

UNISDR – United Nations International Strategy for Disaster Reduction

UNOCHA – UN Office for the Coordination of Humanitarian Affairs

WFP – World Food Programme

WMO – World Meteorological Organization

(9)

Preface

In 2011, I was working for a humanitarian organization in Kenya. At the time, the humanitarian community was scaling up food crisis responses in the Horn of Africa.

However, in the Kenyan humanitarian coordination meetings, I was struck by the divergent views on whether the country was in the midst of a food crisis caused by drought. The pressure to initiative a humanitarian response in Kenya seemed to come from outside. A famine was about to be declared in neighboring Somalia. As weather knows no borders, surely there was a similar “drought emergency” in Kenya as well? Agencies were scrambling to be first in line with launching their fundraising appeals. The big question was – should Kenya be in or out? Our programs in the most marginalized arid lands of Kenya were chronically underfunded, and the humanitarian needs were relentless. Rates of malnutrition were constantly above emergency thresholds, and trekking distances to water far above the national average. Few people consumed three meals a day. At any given time the data could be used as a basis to justify a humanitarian emergency. As such – did it really matter if it rained a lot or a little that particular year?

The experience of 2011 was an accelerator for an internal process of questioning the approach to drought in the Horn of Africa that had begun when I started working in the region in 2006. The more years I spent in the drought prone arid lands of the Greater Horn of Africa, the more I started feeling that something was not right. The questions I was struggling with on the meaning and effectiveness of what in fact was a perpetual humanitarian emergency response were the main motivator for me to undertake my research. I simply wanted to find out what was actually going on, beyond the day-to-day cycle of fundraising and implementation.

Over the years the research, has provided welcome moments for reflection and analysis. I strongly believe that the humanitarian sector could benefit from much more similar research, and that we need to acknowledge that a difference exists between criticizing the “raison d’etre” of humanitarian aid and critical research that aims to make it more effective

(10)

1. Introduction

The Greater Horn of Africa is home to some of the most food insecure populations in the world (FAO, 2020). Humanitarian emergencies occur regularly, leaving millions of people in need of emergency food assistance, with responses to these disasters costing billions of dollars. For example, in October 2019, there were 23.6 million severely food insecure people in Ethiopia, Kenya, Somalia, South Sudan and Sudan (OCHA, 2019). The humanitarian appeals in 2019 totaled 844 million USD for Ethiopia and 1bn USD for Somalia, with over half a billion dollars requested for food security interventions (OCHA, 2019). 2011 saw the largest food crisis in recent years in the region, and a famine was declared in Somalia following a severe drought and protracted conflict (Oxfam 2012). Five years later, a famine was declared in South Sudan (Pauwels, 2019), and in 2021 Ethiopia is on the cusp of a famine (FAO and WFP, 2021). The frequency of these humanitarian crises is increasing, as the average number of food emergencies per year in Africa tripled between 1990–1992 and 2000–2002 (Green, 2008). The majority of the people needing food assistance are the populations of the dryland regions, who depend on pastoralism for their livelihood (FAO, 2020, DeWaal 1997, DEC 2012, Mc Kune et al 2018).

Over the past decades, many interventions have been implemented in the region with the aim to reduce food insecurity, ranging from emergency food aid to social protection, livelihoods support and community drought preparedness (Helland, 2001). There is broad evidence on the failure of these interventions to eradicate, or even significantly reduce, food insecurity (Kaguyny et al 2017, Lesukat 2012, Gilligan et al 2007). The fact that food insecurity persists throughout the region provides the justification and rationale for more in-depth research into this complex situation, including analysis of the factors causing food insecurity (Burchi et De Muro 2015, Rubin, 2009a). This is a challenging area of research, as the causes for food insecurity and famines are complex (Sen, 1981, de Waal, 1997, Helland 2001, Simelane et al 2020).

(11)

Given the large role that rainfed agriculture plays in Sub-Saharan African food production, it is self-evident any analysis of drivers of food insecurity needs to include research on the role of climate change and natural climate variability, and recent years has seen an increase in such work (See for instance Coulibaly et al 2015, Tall et al 2014, Gumucio et al 2019). Currently available climate scenarios indicate that the scale of climate change in the region is not certain. It is it not possible to determine with any certitude which parts of the region will be most affected, nor to project what the changes will be under various emission scenarios (Barros et al, 2014, Sivakumar et al 2005). The Intergovernmental Panel on Climate Change (IPCC) emphasizes that different climate models, or different simulations with the same model, yield different patterns (Barros et al, 2014). A particular challenge lies in determining the character of the climate change signal on rainfall against a background of large natural variability, compounded by imperfect climate models. What researchers engaged in interdisciplinary work in the region tend to converge on is that the adaptive capacity and resilience of the communities in the dryland regions has been eroded, and that adaptation and resilience building is essential under any emission scenario, given the high natural climate variability in the region and the fact that resilient communities can better adapt to any shocks (Sivakumar et al 2005, Maxwell et al 2014, Apgar et al 2017). However, not all research on the impact of climate change on food security include attempts at unpacking the complexities in the livelihoods systems in rural Sub-Saharan Africa, especially among the populations in the arid and semi arid regions (see for instance Tall et al 2014, Gumucio et al 2019).

In this thesis, I explore to what extent humanitarian early warning systems and responses in the East and Horn of Africa have evolved in recent years to reflect the more nuanced understanding of drivers of food insecurity developed in academic food security research, focusing on Ethiopia, Kenya, Tanzania and Somalia. My work is framed within the theory of food entitlements developed by Sen (1981; see also Rubin 2009a) and the food capability approach that has been

(12)

developed based on the entitlement theory (Burchi et De Muro, 2015, Daoud 2017).

The overall research question my work is centered on is the extent to which the humanitarian response and preparedness system in the East and Horn of Africa is centered on a food availability approach and the extent to which the analysis in the system has evolved to reflect a food capability approach, including all elements in the food system. The underlying assumption of my work is that there is a tendency to explain complex failures of food systems as “droughts“, which in essence are crisis of food availability, with less attention paid to other drivers of food insecurity. I expect to find that that the preparedness and response system has not in practice evolved over the past decade to include a more nuanced food security analysis. If the assumption is proved to be correct, my research would reveal a response and early warning system that is still heavily based on what Ribot (2010) refers to as a risk- hazard approach and reliant on indicators related to food production, particularly climate-related indicators. Furthermore, one would expect to see responses based on increasing food production at the local level, including through weather and climate information that would make producers less vulnerable to climate related shocks.

My work is focusing on the pastoralist populations that live in the dryland regions of the countries that are the focus of this research, as the populations most affected by food insecurity. (Egeru, 2016, FAO, 2013, De Waal 1997). These pastoralist populations are some of the most marginalized groups in the region (Egeru 2016), and forty-one percent of the region’s pastoralist population live in poverty (FAO, 2013). While pastoralism is a livelihood that depends on rainfall and other climate variables, available research point to very low levels of uptake of formal weather and climate information among pastoralist populations in Sub-Saharan Africa (Hampson, 2015). The level of use is particularly low among women (Gumucio et al, 2020, Coulibaly et al, 2005).

The research strategy is a mixed methods strategy drawing upon both quantitative and qualitative methods. As outlined by Bryman et al. (2012), the mixed

(13)

methods approach can bring completeness to a set of research questions, using qualitative data to fill gaps left by the quantitative data or vice versa. This method is suitable for this research as the result tends to be more usable for policymakers (Bryman et al 2012). My aim is to provide recommendations usable for the humanitarian and development practice in the region, as outlined in the conclusions.

The research is centered around three questions that are designed to focus specifically on analyzing the dominance of the food availability approach in policy and practice:

1. What are the main drivers of food insecurity identified by the humanitarian community in the Greater Horn of Africa and how has the analysis of drivers evolved over the past decades? (I)

2. What emphasis is placed on food production, specifically climate related indicators in the early warning systems in the region? (II)

3. What barriers exist for households to use climate information in their decision-making related to food production and what are the gender differences in access and use of this information? (III)

The three questions are each responding to the issue from different angles. Question 1 focuses on the conceptual level looking at the region as a whole and attempting to conceptualize the question, whereas question 2 aims to narrow down the scope of the research to a specific issue in a specific country before moving to the household level. The intention is to identify the dominance of the specific narrative at various levels and as such look at how pervasive the food availability approach is, especially as evidenced by explanations of climate factors driving food insecurity. The third research question is designed to elaborate on the findings from the two other questions, by exploring the extent to which climate-related variables affects food production decisions at the household level, by looking specifically at how climate information is used by pastoralists and agro-pastoralists.

(14)

2. Drought and food insecurity

This section introduces the scientific literature on drought and food insecurity. This includes a summary of current available research on the topic (2.1) and an overview of the literature on drought and food security (2.2.), as well as of that of pastoralism in the Horn of Africa (2.3), and on use and uptake of climate services (2.4) as well as an outline of the analytical framework (2.5.) that was developed following the review of the literature summarized in 2.1.

2.1. Conceptualization of food insecurity

FAO defines food security as “when all people, at all times, have physical and economic access to sufficient, safe and nutritious food that meets their dietary needs and food preferences for an active and healthy life” (FAO, 2006). While there is a commonly accepted definition of food security, several approaches exist related to the dimensions of food security and to the dynamics of how food security changes in various temporal and spatial contexts. In addition, analysis of food security is often tied to a broader analysis of vulnerability and poverty (see for instance Cannon T, 2003).Food security has four main dimensions; availability, access, utilization and stability, which are functions of the physical, social and policy environment (Simelane et al 2020). The food system determines the outcome of food security, and consists of three components: food utilization, access and availability (Ericksen, 2005). First, food utilization consists of both nutritional values present in food as well as those associated with it through social norms, and issues of food safety.

Second, access to food is determined by the affordability and allocation of food and preference to foods that are culturally and socially acceptable. Third, availability is considered the result of production, distribution and exchange of food (Thonrton et al 2011, Chase and Grubinger, 2014). While food insecurity occurs in several countries every year, famines are catastrophic events that are rare. The humanitarian community has developed a system for classification of food security called the

(15)

Integrated Food Security Phase Classification. According to this system, a famine can be declared when certain indicators of malnutrition, mortality and hunger have been exceeded (FAO, 2008).

There is a long line of research that unpack the political nature of famine and food insecurity that started to gain ground in the 1980s (see for instance Torry, 1986 and Hay 1986). Much of the research on the political nature of famine that emerged in the 1980s fall within the discipline of political economy and emerged at the same time as research on the multidimensional aspects of poverty, noting the interconnectedness between poverty and hunger. These political economy approaches to hunger are based on Marxian thought and represent a deterministic approach that outlines a clear causal link between capitalism, poverty and hunger (see for instance Torry, 1986). What these studies have in common is the notion that while environmental factors contribute to food insecurity, famines are man-made. A nominal work in this line of famine research is the groundbreaking work by Sen in 1981 who criticized the “food availability decline” theory, instead focusing on the terms of exchange between labor, assets and food in an approach that focused on entitlements (Sen, 1981).

In the 1980s several academics followed on Sen’s theory and developed it further. Hay (1986) argued that the analysis on causes of hunger that dominated in the 1970s equated deficits in production with hunger and portrayed famines as catastrophic events, and in this regard fails to consider the complexities of access to food which depends less on the quantity and quality of food available but more on a

“hierarchy of claims” that need to be analyzed from a political economy perspective.

As Scrimshaw (1987) highlights, famines are caused by a complexity of factors that affect the distribution of food, including poverty, civil disturbances and political interference. What is worth noting though is that a number of the seminal works, including that of Sen, is focusing on famine, rather than the broader topic of food insecurity. During the 1990s it became increasingly clear that emergency relief was obscuring the real nature of chronic hunger and vulnerability in many countries, and

(16)

that “hunger is the norm, not an emergency” (Green, 2008, pp 210). The temporal nature of food insecurity and the difference between chronic and transitory food insecurity is often overlooked (Devereux, 2006). Chronic food insecurity is defined by the United Nations Food and Agricultural Organization (FAO) (2006) because of structural deficiency and requires much broader intervention than the safety net programs and food aid interventions that can address transitory food insecurity (Devereux, 2006). Green (2008) quotes the FAO estimate that chronic hunger lies at the root of 90 per cent of food insecurity worldwide, leaving just 10 per cent attributable to shocks or humanitarian crisis. This recognition of the significance of chronic hunger is in line with much of the literature that has conceptualized food systems more broadly as consisting of interactions between and within bio- geophysical and human environments (Thonrton et al 2011, Chase and Grubinger, 2014 and Brown et al., 2015).

Current research on food insecurity can be divided into two main and divergent lines of thought; one focused on food availability and the other focused on food entitlements (Daoud 2017). While the former is usually analyzed at both the national and household level, the latter is focused on the household level (Simelane et al, 2020). According to the former, availability is defined as the existence of food, either from own farm produce or purchased from off-farm (Riely et al, 1999). The food availability approach centers on factors related to food production as driving food insecurity and famine (Rubin 2009a). This has been the prevailing theory for several decades in both policy and practice, defining food security as a balance of the total amount of food available and the size of the population (Burchi and De Muro, 2015). This is based on Malthusian theory, whereby the main assumption is that population growth will exceed the ability of the environment to provide food for the population, and that populations will decrease due to disaster and conflict caused by the lack of resources (Daoud 2017). The core assumption focuses on the balance or imbalance between population and food: in order to maintain this balance, the growth rate of food availability should not be lower than the growth rate of the

(17)

population (Burchi and De Muro, 2015). The Neo-Malthusian theory expanded on this in predicting that food production will not be able to keep up with population growth, leading to severe crisis (Aliyu et al, 2021). Central to this line of thought is the focus on increasing agricultural productivity to eradicate hunger, and this argument is still prevalent in academic literature today. For instance, Conceicao et al (2016) argue that agricultural growth is the most effective way to reduce poverty because an increase in production drives down the price of food.

Due to the focus on food production in the food availability approach, the narrative also makes a clear link between drought and food security (see for instance Kagunyu et al, 2017, McKune et al 2018). This represents what Ribot (2010) refers to as the “risk-hazard” model, which identifies a causal link between impacts that affects vulnerability and a single climatic event (Ribot, 2010). This risk-hazard model forms the theoretic framework for much of the research on the impact of climate change on food security (see for instance Fischer et al 2002, Mbow et al 2019 and McKune et al 2018, Apgar et al 2017 Tall et al 2014, Coulibaly et al 2015). As McMichael et al (2017, pp 121) point out, this approach “brings together pressures on food crop-land with extreme weather patterns, rising energy costs, speculation and ecological stress into a perfect storm scenario”. In the food availability narrative, the response to food crisis includes measures to increase food production, including through promoting free trade and investment in the agricultural sector (McMichael et al, 2017). Spann (2017, pp 369) calls this the “dominant mainstream”

understanding and practice of food insecurity that “revolves around the efficacy of the market coordinating the production, distribution and consumption of food”.

The food entitlement approach, originating in the work of Sen (1981), differs from the food availability approach as it emphasizes how inequality and poverty are decisive factors compared to the availability of food in causing famines.

According to the work of Sen, and the academics that have followed in his footsteps, not all famines are the result of extreme changes in environmental or farming conditions. Rather Sen focused on the terms of exchange between labor, assets and

(18)

food in an approach that focused on entitlements (DeWaal 2018). “The entitlement approach concentrates on each person’s entitlements to commodity bundles including food, and views starvation as resulting from a failure to be entitled to any bundle with enough food” (Sen 1981: 434).While Sen’s work focused on famine, researchers such as Burchi and De Muro (2012) have used the approach to analyze food insecurity. Despite Sen being awarded a Nobel Prize for his work, the entitlement approach remains marginal both in research and practice (Rubin, 2009a Kula 1988).

One area where the entitlement approach has been instrumental is in informing the development of the vulnerability approach in both theory and practice, which focuses on a broader analysis of risks and vulnerability to food insecurity, taking into account factors that affect households’ food access, availability and nutritional status (Simelane et al, 2020). Rather than focusing on a single cause and its impact, the vulnerability approach aims to identify the multiple causes of a single impact and the capacity of a household to act, what is referred to by for instance Ibok et al (2019) as its adaptive capacity. However, incorporating vulnerability analysis in practice in food security assessments has proven challenging as researchers agree that the vulnerability concept in itself does not have an objective definition (Ibok et al, 2019).

What is lacking from both the food availability and food entitlement approach is a focus on policy aspects that affect food security, both at the national, regional and global level. Rubin (2009a) cites this lack of a political dimension as one of the reasons why the entitlement approach has not gained ground, noting that political crisis, institutional failure and conflict decisively contributed to the most recent famines – the famine of Somalia in 2011 and in South Sudan in 2017. Partially to address this, the food entitlement approach was further developed by Sen himself and for example by Burchi and De Muro (2015) into the food capability approach, which includes components not only related to access but also take into account the institutional and policy framework that enables a person to be food secure (Burchi

(19)

and De Muro, 2015). As such this is a broader approach than the food entitlement approach and includes for instance access to basic services. The need for the entitlement approach to be supplemented by political analysis on the macro or meso level when applied to analyzing food security at a broader level than the household has been highlighted for instance by Rubin (2009a) and Ibok et al (2019). However, the capability approach has not been widely used in practice due to a lack of guidance on how to operationalize it (Burchi and De Muro, 2015). Instead of a shift towards stronger policy analysis, recent years have seen an increased focus on environmental causes to food security both in research and practice (see for instance Tall et al, 2014). As I will outline in the next chapter, the link between climate change and food insecurity is far from linear or straightforward.

2.2 Drought and food security in East and Horn of Africa

The World Meteorological Organization (WMO, 2017) defines drought to be “a prolonged dry period in the natural climate cycle that can occur anywhere in the world. It is a slow on-set phenomenon caused by a lack of rainfall”. The United Nations International Strategy for Disaster Reduction (UNISDR) identifies four different types of drought: meteorological, agricultural, hydrological, and socioeconomic (UNISDR, 2009). According to the Intergovernmental Panel on Climate Change (IPCC), more intense and longer droughts have been observed over wider areas during the twentieth century dominantly linked to higher temperatures and decreased precipitation (Barros et al 2014). Haile et al (2020) also found that droughts will increase in both severity and frequency in East Africa by the end of the 21^st century. Available studies on the global and regional impact of climate change on agriculture do also not provide a conclusive picture. Models suggest that up to 2–

3 ^oC warming in the tropics may be tolerated by crops, especially if precipitation increases, but for livestock there is a lack of models relating climate to animal physiology limits (Mbow et al, 2019). As Dorward et al (2019) note climatologists

(20)

consider that the impacts of climate change on precipitation at the regional and country scale will remain uncertain for years to come, which makes predicting the impact on food production very challenging. In the case study description in Chapter 3.1, the current academic discussion on projected climate change in the region is outlined in more detail.

There is a wide body of research that argues that a clear causal link exists between drought and food security (see for instance Coulibaly et al, 2015, Kagunyu et al 2017, Tall et al, 2014, Sivakumar et al, 2007). Such research includes statements such as “frequent droughts and floods have led not only to chronic famines and hunger, but also dwindling pastoral economy” (Kagunyu et al, 2017, pp7). However, a food systems analysis of the possible impacts of climate change would need to take into account the social disruptions caused by climate change, such as migration, and its impact on food utilization rather than merely focusing on disruptions to food production (DeWaal, 2018). The IPCC report on the impact of climate change on food security from 2014 recognizes this complexity, stating that “food security is dependent on access and consumption patterns, food utilization and nutrition, and overall stability of the system as much as food production and availability. The overall impact of climate change on food security is considerably more complex and potentially greater than projected impacts on agricultural productivity alone” (Porter et al, 2014, p502). For instance, extreme weather events can drive up prices globally or in parts of the world (Mbow et al, 2019). Globally, the positive and negative effects of climate change on food production are expected to balance each other out for the coming decades, but that in many parts of the world the negative effects will be very significant (Mbow et al. 2019).

There are several papers focusing on exploring the complexity in the dynamics between climate change, drought and food security (see for instance Sendalo 2009, Lewis 2017, Drèze 1991, Giannini et al 2016). There are several conceptual approaches and empirical studies that have shown that a drought does not automatically result in a famine (see for instance deWaal, 1997, Lewis 2017,

(21)

Kratli 2014., Drèze 1991), nor can it alone be considered the main contributing factor (Maxwell et al 2014, Hammond and Maxwell, 2002). The Special Report on Drought published by the UN Disaster Risk Reduction office in 2021 emphasizes that drought risk result from “complex and nonlinear interactions of drought events with exposure of humans, infrastructure and ecosystems, to systems’ vulnerabilities across multiple scales, sectors and systems” (UNDRR, 2021, pp59). Dréze and Sen (1991) in their work on famine prevention in Africa turns the question on its head – rather than analyzing why drought causes famine, he looks at situations of prolonged drought, such as Cabo Verde in the 1980s, and poses the question as to why these droughts did not cause a famine. Giannini et al (2016) in their study on climate risk and food security in Mali emphasize that there is a more direct causal link between climate and food availability, but that the relationships between climate and access to food is much more complex. One notable study in this regard is the research on climate as a driver of food insecurity that was conducted by Lewis (2017). In her research, Lewis sets out to question the dominant narrative of a correlation existing between drought economic growth and food insecurity in Ethiopia, noting that the number of food insecure in Ethiopia has remained stagnant while both total and per capita food production has increased. Lewis also noted that her document analysis found that a drought had been declared almost annually, despite this not being backed up by rainfall data. Lewis bases her research on the concept of

“socioeconomic drought” that is not linked to meteorological drought. This is linked to previous work by researchers such as Meinke et al (2006), who have gone so far as to argue that drought is a social construct.

As another example, Rubin (2009b) uses the 2005 Niger famine to highlight the failure of the early warning system which relies on drought and weather-related factors rather than socioeconomic ones to accurately predict famine and uses it as an argument for adopting an entitlement approach to early warning.

Rubin (2009b) calls for an increased focus on four specific themes: conflict, the collapse of legal structures and famine dynamics at the political level. Similarly,

(22)

Drèze focuses on the political aspect of famine prevention, noting that early response

“has been much more a matter of political incentives and motivation than one of information or predictive wizardry” (Drèze 1991, pp41). Establishing an early warning system that would capture the four themes identified by Rubin would entail what Lewis (2017) refers to as a systemic change. This would be based on a detailed assessment of vulnerability in line with the capability/entitlement approach referred to in chapter 2.1 and would require “a socioecological systems perspective that can consider the susceptibility of ecosystems and deficits in coping capacities of the communities depending on them” (UNDRR, 2021).

2.3 Pastoralist populations in East and Horn of Africa

Pastoralism is a form of livestock production in which livestock keepers move their cattle, sheep and goats from place to place to take advantage of pasture and water which are available at different times during the year (Sendalo, 2009). The economic, political and social marginalization of pastoralist populations is a pervasive feature in the East and Horn of Africa. There is an evolving but persistent narrative of the pastoralist populations being unable or unwilling to manage their natural resources in a way that would allow for sustainable livelihoods (Bersaglio, 2015). The “overgrazing narrative” emerged during the last decades of colonialism, in the 1950s and 1960s (Nelson et al 2005, Bersaglio 2015). This narrative is based on the “carrying capacity” concept, where the basic assumption is that there is a set number of livestock that can be supported by the available rangeland (Nelson et al 2005). In the 1970s and 1980s, the concept of “tragedy of the commons” emerged, launched in the seminal work published by Hardin in 1968 (Hardin, 1968), building on the “carrying capacity” concept and attributing the degradation of the rangelands to the fact that these were communal lands (McCarthy, 2015). This narrative can still be observed in some research, with for instance Fafchams (1998) finding evidence that the root cause of the livestock cycles in Southern Africa is overgrazing. Both the

(23)

“overgrazing” and “carrying capacity” narratives represent Malthusian theory and are such closely related to the food availability approach (Daoud, 2017). The dominant theme is that of pastoralism in crisis, representing it as a livelihood that is no longer sustainable and presenting pastoralists as in need of outside assistance to survive and better manage their resources (Helland, 2001).

In academic research these concepts have gradually given way to an understanding about the variability of rangelands and the adaptability that pastoralists livelihoods provide, suiting this context particularly well (see for instance McCarthy, 2015, Wangui and Smucker 2017). As outlined in Nelson (2012), a considerable body of research has emerged during the last 30 years which is demonstrating that pastoralist management strategies such as mobility, rotational pasture used through different wet season and dry season grazing areas, and building up cattle numbers so as to be able to endure catastrophic but relatively common periods of drought are ecologically rational measures for coping with the variable semi-arid and arid environments that these people live in. This understanding can be described as “blessings of pastoral mobility hypothesis”, emphasizing the unique suitability of pastoralism to the conditions of the rangelands (Nkedianye et al 2011) or as the “non equilibrium concept” emphasizing that grazing availability is affected by high climate variability and that climate shocks are a driving factor for livestock populations. This concept focuses on pastoralists adapting to the unequilibrial environment they live in and emphasizes the unique ability of pastoralists to adapt their livelihood to their environment. As outlined by Grist (2008) such adaptation to climate variability has been practiced by farmers and pastoralists all over the world for millennia.

The concept builds on the theoretical framework developed by Boserup who challenged Malthusian theory and showed how farmers intensified their agricultural practices in the face of environmental constraints, and that there is no such thing as “natural carrying capacity” for any particular environment. (Liu et al, 2007). This approach reflects an increased understanding of arid and semi-arid

(24)

environments as nonequilibrial systems with significant and somewhat unpredictable variability (Liu et al, 2007).

Green (2008) argues that pastoralism evolved in response to long term climate change when the Sahara entered a period of prolonged desiccation some 7,000 years ago and as such it represents a well-developed climate risk management strategy. As Kratli et al (2014, pp48) summarize, “through strategic mobility, pastoralism finds an asset in the existence of dynamic variability in rangeland environments where sedentary agriculture finds a problem in its lack of uniformity and stability”. Pastoralists employ a number of highly specialized risk spreading strategies to safeguard their herds against drought, floods, disease and social unrest.

These strategies—including building up herd sizes as insurance against times of hardship, splitting herds across different locations to spread risk, keeping different species and breeds and loaning surplus animals to family and friends—ensure the rational use of the natural resource base and develop and strengthen social relations as a form of social capital (ODI, 2009). Rather than droughts being “emergencies”

they form an integral part of the pastoralist livelihood system. The influence of various social, economic and political factors on the risk management strategies used by pastoralists has been well researched and documented. For instance, the UNISDR Drought Risk Reduction Framework (2009) identified the restriction of pastoralist mobility, conflict and insecurity and inadequate livestock market infrastructure and access as key eroding factors for pastoralist coping capacity. Kratli et al (2014) talk about the long and widespread undermining of the pastoralist culture and its values and institutions and how this has led to increased vulnerability of the pastoralist livelihood system. Green (2008) argues that “the gravity of the current situation for pastoralist communities stems more from years of neglect and misunderstanding by the central government than from the unpredictability of rainfall” (Green, 2008, pp269).

It is therefore possible to see a shift from the early narratives in which pastoralists and their livestock were seen as driving environmental degradation and

(25)

desertification to the current narrative of climate variability, and increasingly climate change as driving causes of pastoral destitution. However, even though recent work emphasizes the ability of pastoralists to adapt to their environment, it is common to find references both to climate change causing increased variability, which coupled with other factors such as political and economic marginalization (highlighted by for instance Helland 2001, Green 2008 and Sendalo 2009 as a driving factor for pastoralist vulnerability) is constructed into a narrative that continues the “pastoralism in crisis” theme from the overgrazing narrative. So even though pastoralists are no longer described as causing the environmental degradation that is destroying their livelihood, that livelihood is still being described as not viable, with pastoralists now being the “victims” of environmental degradation ascribed to climate change.

2.4 Use of climate information among pastoralist households

Several studies have been conducted on the use of weather and climate services among rural communities in Sub Saharan Africa (see for instance Tall et al., 2014 a;

Vogel & O Brien 2006; Hansen et al 2012, Dorward et al 2020). As outlined by Vogel and O’Brien (2001) there are two general approaches used to assess the use of the forecasts: prescriptive approaches which assume that users of the information behave in a manner that is optimal according to some normative theory of decision making; and descriptive approaches, which focus on the actual behavior of users and their actual information processing and decision-making procedures. The prescriptive approach is evident in for instance Tall et al. (2014 a), Wood et al. (2014) and de Jalon et al. (2015). A commonly occurring theme in these prescriptive studies is the identification of multiple bottlenecks that prevent forecasts from supporting farmers in managing climate risk (O’Brien et al., 2000; Tall, 2014 b, Hansen, 2012).

(26)

Researchers applying a descriptive approach to the use of climate information emphasize that households see as one risk factor among a multitude that vulnerable households face. This is aligned with the entitlement/capability approach to food insecurity in that it takes into account the broad range of entitlements of households and the various risks that affects these entitlements. Thus, climatic events form just part of the dynamic decision-making process for a small farming household and understanding the perceptions and management strategies related to risk is important in analyzing bottlenecks to uptake of climate information. Several research efforts have gone into detail on the complexity of the decision-making strategies employed by farming households and understanding the role of climate information in this context (e.g., Nelson 2012; Roncoli et al 2004). As Sivakumar and Hansen (2007, p. 4) highlight “end users usually operate in an environment of considerable uncertainty, reacting to and coping with multiple stressors and risks whose impact are not always clear or predictable”. Roncoli et al. (2004) also found through in-depth interviewing that gender, ethnicity and caste can limit access and use of climate forecasts among African farmers. The overall impact is that little of the information provided reaches users, and that majority of farmers and pastoralists across Africa rely on indigenous knowledge and personal experience to make agricultural related decisions. Vogel and O’Brien (2001, p. 18) summarize the situation “even if perfect forecasts were disseminated in an optimal manner there remain significant factors constraining their use and thus limiting their value”. Such factors include attitude to risk, and the relative payoffs from taking action versus doing nothing (Thornton et al, 2011).

In the research reviewed gender inequality increasingly emerges as a barrier to access and uptake of climate information (see for instance Gumucio et al 2020, Tall et al 2014 a, Phillips 2001, Vogel and O’Brien 2001, and Roncoli et al 2001). This line of research focuses on an analysis of how gender and differing levels of access to assets function as barriers to uptake of climate information, and also how climate information can serve to promote gender equality within households and

(27)

communities. As Hansen et al (2019) point out, climate service interventions tend to treat women as a homogenous group, which overlooks the fact that in pastoralist societies, the roles, responsibilities, rights and duties of women change over their life cycle (Hodgson, 2000), which means that both the ability and need to access climate information varies for women depending on their age and status within the community. On the issue of use of climate information, Gumucio et al (2020) highlight that there is not a sufficiently large body of research available yet to determine if gender differences in the use of climate information are widespread, with the studies reviewed by Gumucio et al revealing somewhat conflicting results.

Critical research has pointed out that the focus of both researchers and practitioners on women’s roles related to livestock reflect a view on what constitutes an asset that was introduced by the colonial administration and has been perpetuated since by various development interventions (Hodgson, 2000). Research has shown that, whereas development interventions tend to focus on the economic value of livestock as an asset, for the pastoral household economy the income from livestock products is often more important that the income from the sale of livestock itself (Bailey et al., 1999). This implies that the lack of control over key productive assets among pastoralist women that many development interventions are trying to address is in fact a problem constructed to a certain extent by the development community itself. However, research has also found that female headed pastoralist households tend to be less able to diversify their livelihoods than male headed households, implying that gender differences persist outside of the domain of livestock management (Hodgson, 2000).

2.5 Analytical framework

For the analytical framework, I apply a food systems framework, using the components of the food system as identified by Juhola and Neset (2015), food availability, food access and food utilization. Within these components, I group the

(28)

factors included in the analysis of the food availability and the food capability approaches. This analytical framework forms the basis of my overall analysis as well as the analysis of data collected for research questions 1 and 2. For the third research question the approach is slightly different, using the finding from the research conducted for questions 1 and 2 as a basic assumption, i.e., that food production, dependent on climate-related factors, is a core determinant for food security to frame the research.

The components of the food system framework used as the main analytical framework for the research are outlined in Figure 1. In organizing the components in the framework, the aim is to clearly outline the relationships and interlinkages between the various components, outlining which elements of the food system forms the core of which approach, and which components form part of the elements. Using a food systems approach allows for the inclusion in the analysis of elements related to both the food availability, food entitlement and food capability approaches.

In outlining the framework, the food capability approach includes subcomponents related to food entitlement. I use the factors identified by Burchi et De Muro (2012) in their work on refining the entitlement approach to develop a capability approach to food security. The significant difference between a

“traditional” entitlement approach and the capability approach is the added dimension of the capability to have enough food, which requires an analysis of institution and environmental factors that are beyond a person’s control (Burchi and De Muro, 2012). In the analytical framework, food entitlement is therefore included as a subset of the food capability approach. This enables an inclusion of meso and macro level policy related aspects that are not included in analysis based on the entitlement approach.

(29)

The framework supports the mapping of variables cited in the various materials used for this research as the main drivers of food insecurity, allowing for a categorization of the variables within the elements of the food system. As outlined in Figure 1, the framework for my research is formed by the categorizing the three main elements of the food system into the two main analytical approaches and grouping the main components of those elements within each approach. It is important to emphasize that there are interlinkages between several of the elements, and that many of the components listed are linked to a number of the elements in the food system.

The analytical framework is designed to allow for an inclusion of both micro level variables as well as meso and macro level variables. The framework places drought as a driver of food insecurity within the production component of the availability element of the food system.

Food security

Food Availability

Availability

Production Distribution

Access

Food

entitelments Basic

capabilities

Food capability

Utilization

Capability to be food

secure

Figure 1 Overview of analytical framework

(30)

3. Materials and methods

As outlined in the introduction, the research is designed using a mixed methods approach (Denscombe, 2008), using both quantitative and qualitative methods. The primary method of data collection is qualitative, with quantitative methods such as statistical methods and rainfall analysis applied to specific components of the research to complement the qualitative data. The aim of applying a mixed methods approach to the research is therefore to add to the completeness of the analysis and enhance the findings of the quantitative analysis (Denscombe, 2008). As Schoonenboom (2007) outlines, one of the five purposes of mixed methods research is complementarity, seeking “elaboration, enhancement, illustration, clarification of the results from one method with the results from the other method”. However, using quantitative analysis at certain points of the research process also allow for a certain degree of corroboration of the findings from the qualitative research (Denscombe, 2008).

As outlined in Table 1 below, the research is qualitative driven (Schoonenboom, 2007), sequential exploratory research. Exploratory sequential mixed methods is an approach to combine qualitative and quantitative data collection and analysis in a sequence of phases (Creswell and Plano Clark, 2017). The rationale for this approach lies in first exploring the conceptual framework using qualitative methods (I) before deciding what variables need to be measured for the empirical research conducted for pieces (II and III). Hence, as outlined by Schoonenboom (2007), the data collection and data analysis for the first part of the research (I) took place before the data collection and data analysis of the other components (II and III) and depended on the outcomes of component I.

Components II and III were completed concurrently.

(31)

Table 1 Methods used

Materials Case study area

Methods

Humanitarian appeal and evaluation documents (I)

Kenya, Ethiopia and Somalia

Content analysis: coding analytical and grounded themes/sub-/categories that were organized according to the food system elements (Bryman, 2012) Humanitarian funding data

(I)

Kenya, Ethiopia and Somalia

Quantitative trend analysis

NDMA/KFSSG biannual assessment documents (II)

Kenya Content analysis: coding analytical and grounded themes/sub-/categories that were organized according to the food system elements (Bryman, 2012) Turkana county drought

bulletins (II)

Turkana county, Kenya

Content analysis: coding analytical and grounded themes/sub-/categories that were organized according to the food system elements (Bryman, 2012) Lodwar precipitation

observational data (II)

Turkana county, Kenya

Quantitative trend analysis

Secondary household survey data (III)

Longido and Kiteto county, Tanzania

Regression analysis

Focus group discussions Longido and Kiteto county, Tanzania

Thematic analysis (Kvale &

Brinkmann, 2009) deductively identifying the analytical themes.

As outlined in the table above the method used in the document analysis for I and II was content analysis through thematic coding (Bryman, 2012). Both thematic coding and content analysis allow a researcher to organize and structure qualitative data for further analysis. The analytical framework used to categorize the thematic coding is outlined in Section 2.4. An inductive coding approach was also used to identify subcategories and look for patterns and contradictions in the material (I and II). This element was an important part of the analysis as (I) included documents from a long-time frame over 30 years and (II) included two different sets of materials. For the quantitative analysis of sets of precipitation data used on materials in II, I used descriptive statistical analysis, whereas regression analysis was applied to show correlation for the quantitative analysis of the material in III.

(32)

3.1 Case studies

The East and Horn of Africa is used as a case study because of the persistence of food insecurity associated with drought. The 2011 crisis affected over 13 million people in the Greater Horn of Africa and was most severe in Somalia, where a famine was declared in July 2011. According to estimates, the famine resulted in excess mortality of around 250,000 people in Somalia (Maxwell et al., 2014). FAO estimates that over 60% of cattle in Ethiopia and Kenya died during the drought (cited in Coghlan et al., 2014).

Drought is a naturally occurring phenomenon in the drylands of East and Southern Africa. The climate of the drylands is characterized by scarce and unreliable rainfall. High temperatures ensure that much of what does fall is lost in evaporation. There are also substantial and unpredictable differences in total rainfall between years, within years and between areas in one year, so that neighboring villages can experience very different crop yields and harvest success (FAO, 2013).

Extreme precipitation changes over Eastern and Southern Africa such as droughts and heavy rainfall have been experienced more frequently during the last 30-60 years (Barros et al, 2014). An assessment of 12 Global Climate Models over eastern Africa suggest that by the end of the 21st Century there will be a wetter climate with more intense wet seasons and less severe droughts during October- November-December (OND) and March-April-May (MAM) (Barros et al, 2014).

These results indicate a reversal of historical trend in these months (Williams and Funk, 2011). However, small precipitation gains will largely be offset by high temperatures resulting into un-productive high soil evaporation and shortening of growing period. A continued warming in the Indian–Pacific warm pool has been shown to contribute to more frequent East African droughts over the past 30 years during the spring and summer seasons (Williams and Funk, 2011).

The Governments of the region are increasingly recognizing that drought related disasters are largely avoidable (Obunde, 2009) and as Glantz (2001)

(33)

outlines, there is now a perceptible shift away from reliance on drought and disaster management towards better preparedness and prevention. Drought early warning systems have been developed that integrate information from various sources (UNISRD, 2009). Existing early warning systems analyze a range of factors, including weather, agriculture, livestock, markets and nutrition, and are becoming more sophisticated and predictions more reliable (Oxfam, 2012).

For the analysis of the use of rainfall data in the Kenyan early warning system (II), I narrowed down on one county as a case study approach was the most appropriate, providing the opportunity to focus on the rainfall data for one county in depth, rather than taking a wider approach, which might lead to the local variations being obscured Kenya has seen a devolution of drought early warning with development of county level early warning systems in recent years (Government of Kenya, 2013), and as such for part of my research I focused on the county of Turkana.

Turkana is located in northwestern Kenya bordering Uganda, South Sudan, and Ethiopia. The county has an arid to semi-arid climate, with average daytime temperatures from 24 °C to 38 °C and average annual rainfall fluctuating within the range of a minimum of 120 mm to a maximum of 500 mm. The rainfall is largely distributed in the March to May long rain season and the October to November short rain season. Approximately 94.3% of the nearly one million population of the district live in extreme poverty and malnutrition is the most prevalent condition (Government of Kenya, 2013). Large-scale distribution of food aid in Turkana by the government and humanitarian organizations began in response to the drought of 1960–1961, and relief in the county was first distributed in 1934 (Bersaglio et al, 2015, Lind 2005). With the severe famines of the early 1980s (1979–1980 and 1984–

1985), relief assistance was scaled up under the coordination of the Turkana Rehabilitation Project to reach around 85,000 people or about half the population of Turkana District at that time (Bersaglio et al, 2015).

For the research on household use of climate information (III), the geographical areas of focus were Kiteto and Longido districts, located in Manyara

(34)

and Arusha regions in Northern Tanzania. These semi-arid districts with most pastoralist populations were the target districts for a multi-agency climate service programme, the Global Framework for Climate Services (GFCS) Adaptation Programme in Africa. The GFCS Adaptation Programme in Africa was a three-year programme implemented in Malawi and Tanzania from 2014 to 2016, aiming at improving climate services for agriculture, health and disaster risk reduction (Coulibaly et al, 2015). Key activities implemented included the downscaling of climate products such as the seasonal forecast to the district level and disseminating this to farmers through trained intermediaries, radio and mobile phones. As part of this programme, a quantitative baseline study was conducted in 2014 by the World Agroforestry Centre (ICRAF) under the leadership of the Consultative Group for International Agricultural Research (CGIAR) research programme on Climate Change, Agriculture and Food Security (CCAFS) (Coulibaly et al., 2015). This data provided the basis for the analysis using statistical methods.

3.2 Qualitative materials and methods

Two main methods were used to gather and analyze data for the research conducted to explore the main drivers of food insecurity identified by the humanitarian community in the Greater Horn of Africa (I). First, to identify how a change in conceptualizing the problem translates into changes in practice the causes of the food crisis identified in humanitarian appeal documents (Common Appeals) generated by the UN Office for the Office for the Coordination of Humanitarian Affairs (OCHA) for Kenya, Ethiopia and Somalia were analyzed over a 10-year period (2004–2013).

The causes were categorized according to the food systems framework. To put these appeals into context, they are compared with the main climate product, the seasonal rainfall forecasts, which are used by the analysts managing the early warning system and decision makers for humanitarian responses. The September forecasts were chosen as they reflect the information the humanitarian community had at the time

(35)

of drafting the appeal document. They do not reflect the actual precipitation observed in the region during the period.

Second, to take a more in-depth look at the rationale and justification for interventions used by humanitarian agencies, a document analysis was undertaken using humanitarian programme evaluations for one case study, the 2011 food crisis, in order to identify what agencies described to be the cause of the crisis, and the significance placed on drought and rainfall in relation to other factors. I expected to see a more in-depth analysis of the background and causes to the crisis in the evaluations than in the appeals, given the reflective nature of these documents as well as the timing of their writing, after the crisis had subsided. The content of these documents was analyzed using a thematic analysis approach as described in Bryman (2012). The thematic categories were identified following the literature review, with data gathered for six categories: label of crisis, description of main cause of crisis, reference to rainfall variability, other causes identified, reference to the use of early warning system and information concerning the balance between relief interventions and long-term development programming. The documents used were the agency evaluations conducted by the largest actors in the response, including the World Food Programme (WFP), International Federation of Red Cross and Red Crescent Societies, Care, Save the Children, Oxfam, Action Contre le Faim and World Vision. The year 2011 was chosen because this was the largest humanitarian response in the Horn of Africa to date, both in terms of beneficiary numbers and financial resources.

The document analysis of the NDMA/KFSSG biannual assessment covered a five-year period of 2013–2017. The documents used for this research were the section “Short and Long Rains Assessments for Turkana County”, published in August and February each year based on the outcomes of the short rains (October, November, December) and long rains (March, April, and May) as well as the county drought bulletins. Once identified, the indicators used in the Long and Short Rains Assessments were categorized using the food systems categories of food availability,

(36)

access, and utilization, as per the analytical framework outlined in section 3. Finally, the monthly county drought bulletins were analyzed using the codes developed in the categorization of the indicators for the long and short rains assessment, with a view of conducting a comparative analysis between the two sets of documents.

The methods used for research into the use of climate information at the household level (III) comprised of contextual analysis, semi-structured key informant interviews and statistical analysis of secondary data. Key informant interviews were held with a total of eight respondents to complement the literature review. The focus of these interviews was to identify steps taken to address dynamics in the provision of climate information, acknowledging the anecdotal nature of this information. Key informants were selected by the authors based on posts held in organizations implementing climate service projects, including Care International, University of Reading, the Food and Agricultural Organization (FAO) and the World Food Programme (WFP). The data collected through the key informant interviews was analyzed using a thematic approach, with the aim to identify key themes within the data with regards to the use of climate information. (Bryman et al, 2012)

3.3. Quantitative materials and methods

To establish whether a perceptible shift in the conceptualization of food crisis could be observed over time the causes to the food crisis outlined in the OCHA situational reports for 2011 were compared with those mentioned in the situational reports for the 1984/1985 Ethiopia famine. Finally, funding data obtained from OCHA situational reports for the 1984 crisis and from OCHA, which managed the financial tracking service for the 2011 crisis, was analyzed to compare the allocation of funding for the humanitarian responses to these two major famines. This analysis looked specifically at the share of funds from the total budget that was dedicated to food aid, based on an assumption that if the understanding of these food crisis has evolved over the years more resources would be dedicated to activities that aim to support

(37)

livelihoods and resilience building and a smaller share of the funds would go to food aid

The rainfall data analysis was examined for the county drought bulletins over a five-year period 2013–2017. The Kenya Meteorological Department has one observational station in Turkana, in Lodwar. The station collects daily observations of rainfall and temperature. Rainfall data from the station from the Kenya Meteorological Department between 1980 and 2017 was used as data for the analysis. We calculated 5- and 30-year moving average of annual rainfall and rainfall during the main rainy seasons of October–December (OND) and March–May (MAM).

Secondary survey data collected in 2014 (Coulibaly et al., 2015) was used in the research on the barriers to use of climate information by households in Tanzania (III) both to provide the context as well as to conduct a regression analysis to answer the research question on whether there is a relationship between women and men utilizing climate information and key productive assets. The secondary data used was collected by ICRAF through structured household questionnaires in sampled villages. A stratified random sampling design including village experiments (to receive the program) and controls (to serve as comparison) was used for the individual household data collection process. Data were collected in 17 villages (9 experimental and 8 control) and on 340 individual households (25% female headed) using trained teams of enumerators (Coulibaly et al., 2015). Statistical tests were run on selected variables to identify whether a relationship was to be observed from the data between women’s and men’s use of climate information and their use of key productive assets and inputs.

3.4 Limitations of methodology

One of the core challenges with a mixed methods approach as outlined by Bryman et al (2012) is how to adequately integrate the quantitative and qualitative findings.

The easy explanation - exploring the link between drought and food security in the East and Horn of Africa

Doctoral Programme in Interdisciplinary Environmental Sciences

University of Helsinki