Peri-urban food insecurity and coping strategies among farm households in the face of rapid urbanization in Sub-Saharan Africa: Evidence from Ethiopia

Aboye, Bahiru Haile; Gebre-Egziabher, Tegegne; Kebede, Belaynesh Article Peri-urban food insecurity and coping strategies among farm households in the face of rapid urbanization in Sub-Saharan Africa: Evidence from Ethiopia Research in Globalization Provided in Cooperation with: Elsevier Suggested Citation: Aboye, Bahiru Haile; Gebre-Egziabher, Tegegne; Kebede, Belaynesh (2024) : Peri-urban food insecurity and coping strategies among farm households in the face of rapid urbanization in Sub-Saharan Africa: Evidence from Ethiopia, Research in Globalization, ISSN 2590-051X, Elsevier, Amsterdam, Vol. 8, pp. 1-13, https://doi.org/10.1016/j.resglo.2024.100200 This Version is available at: https://hdl.handle.net/10419/331128 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/4.0/ Research in Globalization 8 (2024) 100200 Available online 15 February 2024 2590-051X/© 2024 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/bync-nd/4.0/). Peri-urban food insecurity and coping strategies among farm households in the face of rapid urbanization in Sub-Saharan Africa: Evidence from Ethiopia Bahiru Haile Aboye a , c , * , Tegegne Gebre-Egziabher b , Belaynesh Kebede c a Department of Geography and Environmental Studies, Jimma University, Ethiopia b Department of Geography and Environmental Studies, Addis Ababa University, Ethiopia c Department of Geography and Environmental Studies, University of Gondar, Ethiopia ARTICLE INFO Keywords: Food insecurity, peri-urban, farm households Coping strategies Rapid urbanization Determinant factors Intensive land-use Ordered probit model Ethiopia ABSTRACT A growing body of literature discusses food insecurity in rural and urban contexts in Sub-Saharan Africa (SSA). Peri-urban food insecurity and coping strategies remain an understudied topic. To fill this gap, household crosssectional survey data was collected from randomly selected 300 farm households in the peri-urban area of Jimma City and complemented with key informant interviews (KIIs) and focus group discussions (FGDs) to assess their food insecurity status and coping strategies in the face of rapid urbanization. A structured questionnaire was used to collect quantitative data from farm households while an unstructured questionnaire was used to gather qualitative data from key informants. Twelve indicators involving the four food insecurity dimensions were identified to develop a food insecurity index and categorize households into four food insecurity levels. The Principal Component Analysis (PCA) is employed to determine the most important indicators of household food insecurity. The ordered probit regression model was employed to ascertain significant factors affecting farming household food insecurity. The results revealed that 46% of peri-urban farm households were food-insecure at various levels. The household food insecurity status varied with their level of human capital, physical endowments, risk aversion behaviour, and institutional barriers. The expansion of built-up areas and marketable crops, particularly eucalyptus trees have greatly affected the food security status of peri-urban farm households. Income diversification, farming diversification, social ties, and farming specialization were the main food insecurity coping strategies of peri-urban farm households. The findings of this study highlighted insights into the urban expansion of medium-sized cities and agricultural land loss, marginalization of staple food crops, and market food prices increase, which resulted in increased vulnerability to food insecurity and poverty among farm households in peri-urban areas. The recommendations drawn from these findings are formulating strategies that preserve agricultural land-use and promoting food insecurity mitigation programs that capitalize on the coping strategies of households in peri-urban areas in the face of rapid urbanization. Introduction Food insecurity is one of the main threats facing the survival and development of humans, particularly in the drought-prone Horn of SSA (Bedasa and Deksisa, 2024; Gebre and Rahut, 2021; Muller, 2014)). Despite decades of research to find solutions in SSA, food insecurity remains a fundamental issue and the most urgent challenge as highlighted in the Sustainable Development Goals (SDGs) of the United Nations’ 2030 Agenda: SDG2: Zero Hunger, SDG3: Good Health and Well-being, SDG12: Sustainable Production and Consumption for which scholars, governments, and non-government organizations are striving to achieve it by 2030 (Chen et al., 2022). Food security is defined as a state where all people, at all times, have physical and economic access to sufficient, safe, and nutritious food to meet their dietary needs and food preferences for an active and healthy lifestyle (UNDP, 1996). Failing to achieve this state is regarded as food insecurity. As a multi-layer concept, it comprises four components of food: availability, accessibility, utilization, and stability (McCarthy et al., 2018). Food availability is the physical presence/quantity, accessibility is the ability to purchase the required food quality/diversity by travelling some physical distance, * Corresponding author. E-mail addresses: [email protected], [email protected] (B. Haile Aboye). Contents lists available at ScienceDirect Research in Globalization journal homepage: www.sciencedirect.com/journal/research-in-globalization https://doi.org/10.1016/j.resglo.2024.100200 Received 10 June 2023; Received in revised form 7 February 2024; Accepted 10 February 2024 Research in Globalization 8 (2024) 100200 2 utilization is the ability to acquire culturally essential and safe nutrients, and stability is the all-time/invulnerable presence, access, and safe food. Ensuring these components of food is one of the core priorities of governments since failure to do so undermines the widespread well-being, sociopolitical stability, and sustainable development of nations (Mabuza and Mamba, 2022; Dinku et al., 2023). The food insecurity threat is globally increasing. After a significant reduction from 1.2 billion in the 1990 s to 821 million in 2010, the foodinsecure population slowly increased to 828 million in 2021 and is projected to rise to 840 million by 2030 (Blekking et al., 2020; Trudell et al., 2021; Liu and Zhou, 2021; Salles-Costa et al., 2022; Awad, 2023). As the food-insecure population continues to rise in SSA, understanding the experiences of food-insecure households contending with food shortages and mechanisms to achieve food security will become increasingly important (Hadley et al., 2023). Literature to date has well-documented climate change (Hadley et al., 2023), economic downturns (Salles-Costa et al., 2022), inequalities (Dinku et al., 2023), conflicts (Trudell et al., 2021), pandemics (Ouoba and Sawadogo, 2022), droughts (Hirvonen et al., 2020), policy decisions (Liu and Zhou, 2021), and unemployment (Blekking et al., 2020) as factors causing food insecurity in Africa. The literature also differentiates the gravity of food insecurity in rural and urban contexts. Some argue that urban areas in SSA are more food insecure than rural areas because of higher urban food prices. Rural areas dilute food prices because of income from sales of farming products and livestock assets (Atara et al., 2020). For example, the global food prices increase in 2008 has more affected urban food security than rural SSA because of poverty concentration (Hadley et al., 2011). On the other hand, some studies argue that urban areas are more food secure than rural areas due to their variations in human and financial capital that are affected by geography and the institutions of people (Ejaz and Mallawaarachchi, 2023). Ethiopian-related studies also reveal that child nutritional status in rural areas is worse compared to urban areas (Sisha, 2020; Akalu and Wang, 2023). Despite the above, there is less research on urbanization and periurban food insecurity. Urbanization is the concentration of the human population and economy in towns and cities and their expansion that converts the rural cropland-use to urban land-use (Aboye et al., 2023). Rapid urbanization occurs in peri-urban areas along the rural–urban continuum, where large areas of land are still under agricultural use (Aboye et al., 2023). Agricultural land is an essential land-use type that provides livelihoods and food for millions of farm households and supplies major food and nonfood systems for the growing global human population (Aboye et al., 2023). It is also exposed to strong urban growth that affects its use for food production, leading farm households to experience food insecurity. Empirical findings revealed the opportunities and challenges of urbanization for peri-urban food security. Urbanization releases agricultural land-use and improves farmland per capita, shifting peri-urban farm households to urban areas with better employment opportunities. It mechanizes agriculture and increases food production (Wang et al., 2021). Urbanization also increases the earnings of people in nonfarming employment, which enhances food security (Ma et al., 2023). Urbanization that encroaches on the highest quality of cropland-use does not ensure food security. Improved agricultural land management, agricultural production investment, and agricultural intensification techniques increase food availability in production areas (Headey et al., 2014; Li et al., 2023). Food that is available due to mechanization might be inaccessible because of the absence of marketplaces, infrastructure, and earnings. For this reason, urbanization is observed as a challenge to peri-urban food security. Additionally, urbanization is an irreversible process and an unavoidable concern for the sustainability of cropland-use, food production, and food security, particularly in small towns and medium-sized cities (Cattaneo et al., 2022). With urbanization, cropland-use conversion is projected to increase by 1.8–2.4 % in 2035 (Aboye et al., 2023) and 50–63 % in 2100 (Chen et al., 2020). However, urbanization as a cause of peri-urban food insecurity remains an understudied topic. The urbanization-induced household food insecurity status is poorly known despite research evidence about increasing losses of cropland-use, livelihood, and food production along the rural–urban continuum. Empirical research dealing with food insecurity lacks spatially disaggregated data on peri-urban food insecurity status and household coping strategies to survive food insecurity in the urbanization process. The content and method of food insecurity research primarily focused on rural and urban areas, neglecting peri-urban areas. Theoretically, food production is perceived as a rural issue while its failure makes urban areas equally vulnerable to food insecurity (Mabuza and Mamba, 2022). Government policies and research have highlighted that an increase in rural food production inevitably results in increased urban food availability (Awad, 2023; Beyene et al., 2023; Hammond et al., 2023; Ma et al., 2023). Despite peri-urban areas’ dependence on agricultural production, policies, and research fail to consider them, mainly because they are believed to have urban institutions and functions, and also the belief that interventions in the rural arena have consequences in peri-urban areas. This stereotypical view made the rural food insecurity issue remain top of Ethiopia’s national development agenda, causing a lack of detailed household survey data on peri-urban food insecurity conditions. On the other hand, urban food insecurity was mainly discussed through the food accessibility lens, which restlessly observes urban foodsheds. The food flow network maps abandoned peri-urban areas as gray areas in urban agricultural food production due to the prevailing discourse that peri-urban agricultural land-use is an urban land bank (Rousseau et al., 2020; Mabuza and Mamba, 2022). Consequently, peri-urban food insecurity issue is subordinated to rural and urban food insecurity issues. This paper argues that there is a pressing need to investigate the issue of food insecurity in peri-urban areas, which are dynamic, interactive, and transformative. It is important to determine the level of peri-urban food insecurity status in these areas so that policymakers can design effective strategies to make food security inclusive and achieve SDG2, which aims to end hunger, achieve food security and improve nutrition in all areas, including rural, urban and periurban. Furthermore, the findings of the study support previous research efforts and government strategies to make rural and urban food security initiatives more effective. Concerning the research methods, most empirical studies largely employed subjective indicators mainly food availability (rural) and accessibility (urban) components to measure household food insecurity (Blekking et al., 2020; Atara et al., 2020; Nigussie et al., 2021). This made important issues remain undiscovered. Measuring food insecurity in terms of current consumption or income results in underestimation and misclassification of food-insecure households and conveys incomplete information for policymakers during planning, decision-making, and interventions. Using different indicators correctly and reliably identifies food-insecure households but also makes policymakers aware of what elements of food insecurity each indicator portrays (Maxwell, et al., 2014). The measurements of food insecurity and underlying mechanisms have widened, shifting from traditional self-report of current household food intake and experience-based feelings of hunger, coping strategies, and dietary diversity indicators of food insecurity status (Kolog et al., 2023) to comprehensive and long-term household vulnerability (Hadley et al., 2011) and resilience (Atara et al., 2020; Beyene et al., 2023). This paper argues that household food insecurity tools should be holistic and comprehensive to address the all-time quantity, quality, and safety of food in peri-urban areas. The purpose of this paper is therefore to (1) assess the food insecurity characteristics of peri-urban farm households, (2) examine the underlying mechanism of how urbanization affects the food insecurity of perurban farm households, (3) investigate factors that determine peri-urban food insecurity, and (4) analyze how farm households differ in coping with food insecurity challenges. Chronic and acute food insecurity is a widespread challenge in Ethiopia, particularly among smallholder farm households (Beyene B. Haile Aboye et al. Research in Globalization 8 (2024) 100200 3 et al., 2023; Woleba et al., 2023). More than 27 % and 15 % of rural and urban populations are severely food insecure, respectively (National Planning Commission, 2017). The challenge is disproportionately higher among smallholder farm households in peri-urban areas since they face a paired problem of loss of cropland-use and lack of income to access food (Sekhampu, 2013; Blekking et al., 2020). Ethiopia, which is the secondpopulous country in Africa next to Nigeria, is an interesting case in subSaharan Africa for studying peri-urban food insecurity due to its high urbanization rate on the continent (World Bank, 2021), high croplanduse loss and poor land management system (Aboye et al., 2023), and high susceptibility to drought and income shocks in the Horn of Africa (Tadesse and Tafere, 2017). Jimma City is chosen as a case study within the framework of an already-known national food insecurity challenge to quantify the effect of urban expansion on peri-urban food security and shed light on policy implications. Besides, Jimma City is one of the fastest-growing medium-sized cities in Ethiopia. It serves as a hub for health, education, tourism, and goods and services in southwest Ethiopia. It is also located at a crossroad of three highways running from three national regional states, namely Oromia, Gambella, and Southwestern Ethiopia Peoples. Its social, commercial, financial, and administrative services and functions attract large populations, leading to increased cropland demands for residential houses, and ultimately reducing food production (Aboye et al., 2023). The contributions of this study are twofold. First, in SSA, decades of research works that highlighted rural and urban food insecurity overlooked peri-urban settings. By unravelling the food security conditions of the peri-urban area, this study is essential for policymakers and researchers in SSA to rethink all-inclusive food security strategies. It also provides a systematic research framework for future related studies. The study also enriches the existing theoretical literature on food security by presenting and discussing urbanization and peri-urban food insecurity, which helps other rapidly urbanizing regions of the world achieve inclusive and sustainable food security and hunger-free urban growth of the United Nations SDGs by 2030. Research theoretical foundation and conceptual framework The basic insight into food security is about food availability, accessibility, utilization, and stability both for rural and urban populations (McCarthy et al., 2018). It is believed that an increase in food availability in rural areas may lead to increases in food accessibility in urban areas due to urban households’ non-farming income advantage to purchase food on the market (Li et al., 2023). This study assumes that the food insecurity status in peri-urban areas is exclusively exacerbated by shortages of cropland to produce sufficient food, unlike in rural areas or non-farming income to buy the required food on the market, unlike in urban areas. A commonly used theoretical approach to analyzing peri-urban farming household food insecurity is the agricultural household model, which assumes that households are utility-maximizing consumers and profit-maximizing producers of agricultural products or nonagricultural service income (Mgomezulu et al., 2023). Land, which is the base for food crop production and livestock rearing (Nigussie et al., 2021; Ayal et al., 2023), is the most important factor for peri-urban households to determine how much to maximize profits and how much to consume to maximize utility in the wake of urbanization (Taylor and Charlton, 2019). Urbanization is observed as an engine for the modernization of rural livelihoods since it expands the non-farming labour market, farming food market, social services, innovations, and infrastructure in rural areas (Cattaneo et al., 2022). Responding to the higher market demand for farming products, households intensify their cropland-use with highvalue crops to increase yields and income (Tefera et al., 2022). This suggests that farm households decide to pay higher farming wages than non-farming activities to retain labour in farming and maximize their farming yields and profits which directly translates to increased household food security and poverty reduction (Headey et al., 2014). The decisions to maximize utility and economic benefits depend on the size of urbanization and characteristics of farm households such as socioeconomic, behavioural, institutional, and agro-ecological factors (Aboye et al., 2023). Urbanization can produce unintended outcomes, including croplanduse loss, particularly in smaller towns and cities that have higher backwash effects than spread effects (Vandercasteelen et al., 2018). Moreover, peri-urban households might face income and farming input constraints to intensify their cropland-use and earn the income needed to ensure household food security since governments consider food production as a rural issue and cropland as an urban land bank (Rousseau et al., 2020; Mabuza and Mamba, 2022). Urbanization-induced infrastructure and access to social services, among other effects, improve human capital that enhances non-farming labour opportunity costs (Melketo et al., 2021). This results in rural–urban migration, which often increases farming labour shortages due to the higher urban–rural wage differential (Abebaw et al., 2020; Asefawu, 2022). Alternatively, a lack of farming inputs including labour, fertilizers, extension information, and improved seeds forces farm households to convert their cropland into low-demanding land-uses such as eucalyptus woodland (Yohannes et al., 2023). The latter requires fewer inputs once planted and needs three to four years for harvesting. Hence, the nature of urbanization (smaller towns and cities) can decrease the utility and profit maximization of agricultural food producers and consumers due to the loss of food production bases (cropland and labour). On the other hand, households may generate income from the increased non-farming labour markets to purchase food on the market as full-time cropland-use earns them less income. The effect of the nonfarming labour market is also weak to find sufficient income for households since urbanization is just driven by population growth (Aboye et al., 2023; Li et al., 2023). Consequently, farm households that move away from agricultural producers to market food consumers face income constraints to meet household food needs. The study further considers the coping strategy dimension of farm households to sustain their utility and profit maximization decisions in the face of urbanization (Mabuza and Mamba, 2022; Woleba et al., 2023). Households faced with shortages of cropland-use engage in intensive or extensive land-use to increase their agricultural area. In intensive land-use, which is explained by multiple cropping index or frequency, households produce the same few crops on existing croplands to dilute food insecurity. In extensive farming, households convert other rural land-uses to cropland to maintain crop diversification and food security (Tesfaye, 2022). Moreover, farm households exercise their existing social capital and government support to cope with food insecurity (Aboye et al., 2023). Based on the discussion so far, the following five major hypotheses were set out to test the impact of urbanization on peri-urban farm household food security. H1a: Farm households with larger farm sizes are encouraged to cultivate higher-value crops and rear substantial livestock to maximize utility and economic benefits, which increase household food security. H1b: Farm households with more labour are encouraged to cultivate higher-value crops and rear substantial livestock or earn more nonfarming income to maximize utility and economic benefits, which increase household food security. H1c: Farm households with more diversified income from farming and non-farming activities have a higher household food security status. H1d: Peri-urban food insecurity differs significantly with household socioeconomic characteristics, including age, education, dependency ratio, etc. H1e: Farm households employing multiple cropping, crop diversification, and social capital as coping strategies are less vulnerable to food insecurity. Assuming that peri-urban households as producers, are subjected to production constraints, including basic farming inputs, they as B. Haile Aboye et al. Research in Globalization 8 (2024) 100200 4 consumers, are also subjected to non-farming income constraints to ensure household food security. This implies that the food insecurity of peri-urban farm households is faced with a trade-off between the loss of urbanization-induced food production bases and the low non-farming income that comes with urbanization. In this paper, thus, we use an agricultural household model of food insecurity (FI), in which the household maximizes its utility and profits subject to the constraints of land (L), labour (La), income (I), coping strategy (C), and other farming household characteristics (O) including education, age, dependency ratio, etc. in the face of urbanization. Hence, household food insecurity is a function of these variables. Methods and materials Study area Geographically, SSA lies south of the Sahara, which comprises 49 countries with a total area of 24.3 million km 2 and an approximate population of 1.21 billion (UNDESA, 2023). More than half of SSA is a drought-prone region due to uncertain rainfall. It lacks irrigation facilities to produce food. Consequently, it has the lowest agricultural productivity and highest food insecurity in the world (Bedasa and Deksisa, 2024). This resulted in higher rural–urban migration, making it the fastest-urbanizing region of the globe populated with more than 472 million people. In addition to very rapid urbanization, food insecurity is a chronic challenge for SSA cities (UNDESA, 2023). The challenge of food insecurity in peri-urban areas is expected higher than in the rural and urban areas due to the loss of cropland-use and lack of income to purchase food. Jimma City is one of the SSA cities. It is located in southwestern Ethiopia at 740 ′ 24 ″ N and 3650 ′ 5 ″ E (Fig. 1). It is the biggest city in Southwest Ethiopia with a total area of 105.66 km 2 and a population of over 208,000 (UN, 2015). Jimma City is characterized by rapid population growth, poor urban land management systems, and squatter settlements, which cause severe cropland-use and staple food production loss (Jimma City Administration, 2021). The total monthly rainfall ranges from 37.8 mm in January to 229.1 mm in August with the total annual rainfall varying between 1400 and 2400 mm. The annual average temperature is 30.50c (Jimma Meteorological Agency, 1980–2021). Due to its warm and humid tropical monsoon climate, the area is well known for its important grain-producing cropland and diverse perennial crops, which have huge economic importance in Ethiopia (Brave, 2014). The plain land that emerged from brownish-grey alluvial soil is dominant and highly suitable for agricultural crop production and construction as well. Single cropping, double cropping, and triple cropping are practiced in peri-urban areas. Single cropping is mainly maize. Double cropping involves the rotation of maize with vegetables (sweet potatoes), while triple cropping includes the rotation of maize, wheat/tubers (potatoes), and vegetables (cabbage). Research design, sampling, data collection, and hypothesized relationship Research design and selection of samples This study applied a cross-sectional research design to understand peri-urban food insecurity status and coping strategies of farm households in the study area. This research design is anticipated to help the quantitative and qualitative data collection that was conducted between June and July of 2021 in Jimma City to grasp in-depth and breadth of the food insecurity and coping strategies during the lean season promptly. A four-stage probability sampling procedure was employed to select sample households which were designated as units of analysis. The first stage of sampling involved selecting primary sampling units, which are a sample of the study areas. A total of 4 peri-urban kebeles (the smallest administration unit), namely Bore, Gudata Bula, Kito, and Kofe were selected from 9 peri-urban kebeles (Fig. 1). All 9 peri-urban kebeles were almost homogenous in urban expansion and cropland-use loss experiences. Thus, the 4 randomly selected kebeles were assumed to be adequate to reflect the characteristics of the target kebeles. In the second stage, due to a lack of recent data, we estimated the 2021 household sample frame for the selected kebeles from their respective 2011 household registration books. To ensure the accuracy of the estimate, we manually digitized and recorded all households within the sample kebeles in a polygon shapefile using Landsat-8 imagery and highFig. 1. Map of the study area. B. Haile Aboye et al. Research in Globalization 8 (2024) 100200 5 resolution imageries such as aerial photographs and Google Earth. The results were nearly the same. The application of satellite imagery in household population study is not strange. It is used to estimate population density (Hillson et al., 2019) and assess peri-urban buildings (Schlesinger, 2013). The digitized households were assigned random numbers and used for simple random sampling. In the third stage, the sample kebeles were classified based on their estimated total households. A stratified sampling technique was employed to obtain a proportional sample size from each kebele. Finally, a total of 300 household heads (approximately 9 %) were randomly selected from 3,335 target farm households in four kebeles and used for the questionnaire survey and as a unit of analysis. The adequacy of the sample size agreed with Kothari’s (2004) formula given as follows: n=Z2*N*p*q ε 2(N−1) + Z2*p*q(1) where n =sample size, N=total population, Z=1.96, p (population attributes) =0.5, q (non-population attributes) =1– p, ε =0.05. The digitized polygons were converted to a point shapefile, acquired coordinates, and exported to GPS to locate household samples during the field visit and household survey. All recorded houses were assumed to be occupied by a single farmer. Unavailable house samples or house samples belonging to non-farming, public, or religious institutions were replaced by farming houses, keeping the proportion of four sample kebeles. We also conveniently selected experienced households, local authorities, and agricultural development experts for key informant interviews and focus group discussions based on the saturation principle of diminishing returns. Accordingly, 63 interviewees (55 farm households, 5 local authorities, and 3 farming experts) and 3 sessions (2 farm households and 1 authority and experts) comprising 8 discussants each were conducted during the qualitative data collection. Data collection Data were collected using standardized surveys of the food insecurity indicators and semi-structured in-depth interviews, focus groups, and analysis of published and unpublished records. As food security discussions went beyond making food available and accessible to meet the demand of the increasing population, this study reconsidered the environment of food production, management, and consumption as diverse pathways to effectively understand food-insecure farm households (Coronado-Apodaca et al., 2023). A structured questionnaire was developed covering the four dimensions of food security to construct the food insecurity index and determinant factors of food insecurity. This study selected staple food cropland area per capita, annual staple food production per capita, and Food Consumption Score (FCS) as proxies for household food availability estimation. Poverty and food insecurity levels would be higher in the absence of natural resource-based incomes along the rural–urban continuum implying the importance of considering cropland-use and food production as food insecurity indicators. The FCS is a composite score that mainly deals with household food availability over a 7-day recall period (Kolog et al., 2023). Household income per capita, food market access, and Dietary Diversity Score (DDS) were identified as indicators of household food accessibility. Household incomes earned from various sources, including farming (sales of crops and livestock), on-farming, off-farming, and non-farming activities were covered to measure food access. The DDS is a better tool for estimating household access to varied food groups over a 24-hour recall period (Kolog et al., 2023). Food market distance is the physical distance households travel to acquire staple food. Access to improved water sources, sanitary facilities (toilet), and food storage were used as household food utilization indicators. Peri-urban household food security situation may change over time due to cropland-use loss, food production index variability, and consumer price index variability, which were assumed to be household food stability indicators. For temporal data such as food production, cereal prices, and cropland size variabilities, 2015 was taken as a base year to capture the actual changes in food stability indicators in 2021. In Ethiopia, 2015 was categorized as the worst El Nino-induced drought in decades. It caused poor harvests and more than 10 million Ethiopians in need of food aid (Hirvonen et al., 2020). The questionnaire was also designed to capture households’ background information, such as occupation, and income, farming size, human capital, access to markets and infrastructure, and institutional barriers. The household survey was conducted by researchers through face-to-face interviews. These were used as control variables in the model analysis. The prepared questionnaire was measured on predefined sets of close-ended responses. A team of four experts in urban and agricultural fields was selected to appraise items’ content and face validities. The questionnaire was revised based on experts’ feedback and was pre-tested using conveniently selected 20 households outside the actual samples. The responses to the pilot survey were further appraised using Cronbach’s alpha coefficients. Cronbach’s alpha for reliability and validity was calculated to be 0.88 and 0.92, respectively suggesting that the items have a high degree of internal and external consistency with their theoretical bases (George and Mallery, 2003). This was followed by the actual cross-sectional survey of randomly selected 300 farm households. Qualitative data were gathered from conveniently selected KIIs and FGDs. An unstructured questionnaire was prepared to collect the opinions and thoughts of households, local authorities, and farming extension agents regarding the peri-urban cropland-use loss, food insecurity, and coping strategies among farm households. Major questions emphasized in KIIs and FGDs were food security status, coping strategies, and food security determinants of peri-urban farm households in the wake of urban expansion. During KIIs and FGDs, all responses were recorded using electronic and hard copy with the consent of participants. Operational definition of variables and hypothesized relationship The food insecurity index is the dependent variable while farming household characteristics are the core independent variable in this study. Eleven control variables that are expected to positively or negatively affect peri-urban household food insecurity status were identified based on the existing literature and our knowledge of the area. Households with a higher dependency ratio have less per capita income and food production and are more likely to face food insecurity (Beyene et al., 2023). Households with higher schooling years engage in productive work and are less likely to be food insecure (Sekhampu, 2013). Older age households are less likely to become food insecure due to their experiences in risk aversion decision-making and accumulated assets (Ouoba and Sawadogo, 2022). Households that own larger farmland have higher production of staple crops for household consumption or cash crops to generate income; hence, they are less food insecure (Kolog et al., 2023). Households with more livestock are food secure due to more income from sales of livestock and livestock products, which can be used to buy food for household consumption and farming inputs to increase food production (Melketo et al., 2021). Besides its benefit as a strategy to improve nutrition and human health and climate risk coping strategy, agricultural diversification has a strong link with child growth due to household diet diversity, diet quality, and income (Tesfaye, 2022). Findings on irrigation-based agricultural intensification are inconsistent. A study in Ethiopia suggests that agricultural intensification alleviates increasing food insecurity and farm size decline challenges, particularly in urban areas that facilitate the transformation of agricultural value chains and input supply chains, including seeds, fertilizers, labour, pesticides, herbicides, and tractors (Headey et al., 2014). However, studies in sub-Saharan Africa indicate that agricultural intensification or the substitution of lower-value products (staple food crops) with higher-value products (cash crops) is not viable to alleviate the poverty of smallholders due to small farm sizes, input supply, and B. Haile Aboye et al. Research in Globalization 8 (2024) 100200 6 income per capita (Hammond et al., 2023). Households that diversify incomes through outmigration are less food insecure since farm income failure will be compensated by the off-farm income (Abebaw et al., 2020; Asefawu, 2022). Modern extension services including fertilizers, seeds, training, pesticides, market information, and farming technologies increase household food production and food security (Aboye et al., 2023). Urban markets and infrastructure increase food access, reducing food production costs and food prices and household food insecurity (Nchanji & Nchanji, 2022). In Ethiopia, the 2015 drought increased food insecurity, particularly child undernutrition rates in areas with poor primary road networks (Hirvonen et al., 2020). Descriptions of the explanatory variables used and their hypothesized relationship in the literature with household food insecurity are given in Table 1. Data analysis method Descriptive and econometric techniques were used to analyze the quantitative data while the qualitative data were interpreted using the qualitative methods. Descriptive analysis Construction of food insecurity index. Before the food insecurity index construction, each food insecurity proxy data was refined at a household level using summations and means. Furthermore, the Fisher index, which grasps the geometric average of Laspeyres’ upward bias and Paasche’s downward bias weighted indices (Viet, 2011), was employed to calculate crop production and consumer price indices. The Shannon richness index (Shannon, 1948) was applied to index household income and crop richness. Food insecurity is a multidimensional concept and requires multivariate descriptive analysis that combines all indicators of food insecurity dimensions. To aggregate all dimensional food insecurity indicators into a one-dimensional index and ease the interpretation, Principal Component Analysis (PCA) was used. PCA generates r unrelated principal components from an original set of n related indicators (I1,I2,⋯In) as follows: PCr=ar1I1+ar2I2+ˆ A⋅ˆ A⋅ˆ A⋅ +arnIn(2) where PCr is a linearly weighted aggregation of the original indicators, arn is the weight for the rth principal component and the nth indicator. The sum of a2 r1+a2 r2+⋯+a2 rn is 1. The first principal component with the highest weight and the last principal component with the least weight respectively explain the largest and the lowest proportion of variation in the variables. Following the classification of principal components, the food insecurity index of a household was constructed as follows: FIIj=∑ k i=1 Fi[Xji −Xi Si](3) where FIIj is the food insecurity index for the household j, Fi is the weight for the ith variable, Xij is the jth household’s value for the ith variable, Xi and Si are the mean and standard deviations of the ith variable for overall sample household. The food security index ranges from negative (worst case) to positive (best case). Since there is no unanimously established food security index cut-off point, the classification thresholds employed by Beyene et al., 2023 and Guyu and Muluneh, 2015 were adapted: severe food insecure (FII ≤0.00), moderate food insecure (0.00 <FII ≤ 0.50), moderate food secure (0.50 <FII ≤1.00), and high food secure (1.00 <FII). Econometric model analysis specification Quantitative data of the four food-insecure household categories were separately screened for outliers and tested for normality, heteroscedasticity, and multicollinearity using the Statistical Package for Social Sciences. Considering the categorical and ordinal nature of the outcome variable, an ordered probit model was used to estimate the probability of peri-urban households becoming food insecure as a result of urbanization. The ordered probit model has more extensive literature in econometric studies than the ordered logit model (Beyene et al., 2023; Kolog et al., 2023). The ordered probit model formula is as follows: Yi*=Zi+ ε i,Zi=∑ k k=1 βkXk(4) where Y* is a latent and categorical variable with 0 =severe food insecure, 1 =moderate food insecure, 2 =moderate food secure, and 3 =high food secure. When the observable variable Yi is treated as an ordinal variable with four response categories, its relation with the latent variable is defined as: Yi=1ifYi*≤ μ 1;Yi=2if μ 1<Yi*≤ μ 2;Y=3if μ 2<Yi*≤ μ 3;Y=4if μ 3 <Yi* (5) where μ is a vector of unobservable threshold parameters. The probability that the observed Yi is in ordered category J is obtained as follows: Prob(Yi=J) = 1−Ф( μ J−1−Zi)(6) Prob(Y=1) = Ф( − Zi)Prob(Y=2) = Ф( μ 2−Zi) − Ф( − Zi) Prob(Y=3) = Ф( μ 3−Zi) − Ф( μ 2−Zi)Prob(Y=4) = 1−Ф( μ 3−Zi) where Ф is the standard normal cumulative distribution function and J is the four categories of responses to food insecurity. Therefore, the ordered probit regression model for the analysis of household food insecurity (FI) determinants is formulated as follows: FIIij =β0+β1X1+β2X2+⋯βiXi+ ε i(7) Table 1 Variable definitions, measurement, hypothesized relationship, and descriptive statistics. Variable Definition Measurement Hypothesized relationship Mean (SD) Dependency ratio Non-income earners to income earners Percent (%) Beyene et al. (2023) (+) 64.3 (26.9) Household size Number of persons in a household Person Beyene et al. (2023) (+) 10.1 (2.4) Crop diversity Number of crops growing on farms Shannon index Tesfaye (2022) (-) 2.14 (0.98) Access to market and infrastructure Distance households travel to purchase food Kilometre Hirvonen et al. (2020); Nchanji & Nchanji (2022) (-) 5.2 (3.4) Farm size Total cultivated land area Hectare Kolog et al. (2023) (-) 0.93 (0.6) Income diversity Number of household income sources Shannon index Abebaw et al. (2020); Asefawu, G.S. (2022 (-) 2.86 (0.7) Livestock Livestock size a household owned TLU** Melketo et al., 2021 (-) 1.24 (0.8) Age Age of household heads Year Ouoba & Sawadogo, 2022 (-) 48.1 (14.5) Education Completed household schooling Year Sekhampu (2013) (-) 1.5 (0.7) Irrigation access Households owning irrigated farms Yes =1; No =0 Headey et al., 2014; Hammond et al., 2023 (-/+) 0.4 (0.3) Extension service Households’ access to extension services Yes =1; No =0 Aboye et al. (2022) (-) 0.42 (0.2) ** Tropical Livestock Unit. B. Haile Aboye et al. Research in Globalization 8 (2024) 100200 7 where β is the coefficients of variables, X is explanatory variables, and ε i is the error term. Various tests such as likelihood ratio, maximum log-likelihood, and pseudo-R 2 were checked for how well the model fits the data. According to Hoetker (2007), if the ratio of two competing models, one generated by maximizing the whole variables and the other generated after some variable restriction, is significantly different from one, at least one of the regression coefficients in the model is significantly different from zero and the model has the goodness-of-fit. In maximum log-likelihood estimation, if the additional reduction in the residual variance is not due to chance, the model satisfied the conditions of goodness-of-fit and the value of the unknown variable explained the observed data efficiently and consistently. Pseudo-R 2 with a small ratio reflects the improvement of the full model over the intercept model. The chi-square (x2) test was used to check whether significant differences exist in the categorical data, household food insecurity was significantly influenced by the specified control variables, and the ordered probit model is robust enough for variable comparisons. For qualitative data, all the interview respondents were coded as R1, R2,R3 … R63. The three sessions and their participants were coded respectively as FGD1, FGD2, and FGD3 and P1, P2, P3 … P24. The KIIs, FGDs, and document reviews were grouped into themes, triangulated with quantitative data, and described verbally using the thematic content analysis approach following the quantitative data interpretation. Results Descriptive analysis Socioeconomic characteristics of farm households Table 1 above presents a descriptive summary of selected farming household characteristics. The results showed that households’ average age and household size were 48.1 years and 10.1 persons, respectively. The household size in the study area is much higher than the national (5.3) and zonal (6.2) averages (CSA, 2017). This could be attributed to the prevalence of polygamy, a culture of marrying many wives, the culture of households to retain members, higher late marriage, and lower outmigration. The variations could also arise from household size growth over time. Older households primarily relied on farming compared to younger households and they also faced food insecurity as urbanization increased despite their experiences in risk-averse decisionmaking and accumulated assets (Ouoba and Sawadogo, 2022). The average farmland size per household was 0.93 ha, which is similar to the 0.96 ha national average (Headey et al., 2014). The average dependency ratio was 64.3 %, implying that households have higher non-income earning members. The average school year was 1.5, showing that most households have not yet completed primary education to engage in high-earning jobs. Farm crop diversity was 2.14, which implies that households engaged in producing a few crop species due to urbanization-induced farmland shortages. The average household income diversification, which was 2.86 indicated that households engaged in more than one non-farm activity and are less likely to be food insecure in the face of urbanization. Households allocated their labour to more than two income sources since income from cropland-use and livestock sales was insufficient to meet household income needs. On average, 58 % of households have no access to extension services, indicating that they benefit less from productivity-enhancing factors. The average household distance to the urban markets and infrastructure was about 5 km, which implies that households have to travel quite a distance. On average, 40 % of households have access to rivers or groundwater to grow crops in the dry season. On average, households possess 1.24 animals that could be used in overcoming food insecurity. The mean livestock size in the study area, however, was smaller than the average livestock holding per household in Central Ethiopia (1.58) (Duguma et al., 2012) and Southern Ethiopia (1.78) (Beyene et al., 2023) because of urbanization. Principal component analysis All selected food insecurity dimension indicators were subjected to dimensional reduction. The data fitness for the PCA was examined using Bartlett’s sphericity test and the Kaiser-Meyer-Olkin (KMO) sampling adequacy measure (Table 2). The KMO test (value >0.6) shows that the sampling is adequate for running the PCA. Bartlett’s sphericity test (P<0.05) suggests that the indicators were orthogonal and the variances of their differences were not equal. The PCA was run for each food security indicator and jointly resulted in different dependable factors prioritized based on their eigenvalues greater than 1. The PCA results revealed that the first three components have eigenvalues greater than 1 and explained about 62.4 % of the total variation in the sampled household food insecurity characteristics. Therefore, the food insecurity index was constructed from these three components. The indicators with higher loadings were staple food production per capita, income per capita, consumer price index variability, food production index variability, food consumption score, market access, dietary diversity score, cropland-use loss, and sanitary facilities (toilets). These nine independent factors were the most important food security dimensions that need to be considered during the policy intervention in the peri-urban farm households’ food insecurity reduction in the study area. Household food insecurity characteristics Fig. 2 presents the household characteristics related to the food insecurity index. It was found that 46 % of households were foodinsecure at various levels, 19 % were severely food-insecure, and 27 % were moderately food-insecure. Severe food-insecure populations were higher than the national urban average. The national severe food-insecure population is around 25 % with more concentrated in rural areas (27.1 %) than in urban areas (15.2 %) (National Planning Commission, 2017). Our study revealed that foodinsecure households are higher in peri-urban areas than in urban areas. The government may not report the full scale of peri-urban foodinsecure households, which in some ways could be related to the government policy of land eviction from peri-urban areas. On the other hand, households with moderate and high levels of food security accounted for 40 % and 14 %, respectively. Econometric model analysis The fitness of the ordered probit model is validated using the Brant test, which proves the parallel lines assumption that the proportional ratios of the severe food insecure, moderate food insecure, moderate food secure, and high food secure categories are the same (Long, 2014). The goodness of fit that measures how well a statistical model fits a set of observations is high, and the values expected based on the model are close to the observed values. The results of various check tests proved that the model fits the data (Table 3). Peri-urban farm household food insecurity status varies with their level of human capital, physical endowments, and institutional barriers. The result showed that the dependency ratio was insignificant, though it negatively predicted food-secure households. Households with high dependency will have less labour to release to non-farm or off-farm activities. This will reduce their income and make them more vulnerable to food insecurity. An increase in household size decreases the proportion of the households predicted to be highly food secure, and that decrease was transferred across the other three food insecurity levels. Households with larger family sizes faced food insecurity due to farmland subdivision, despite the availability of labour that could be allocated to produce more food or income and meet household food demand. Households divided the land among household members as urbanization advanced to protect their land-use rights. This characterizes SSA situations, where decreased farmland per capita made farming economically unviable for food production and food security (Melketo et al., 2021). Age was insignificant, although it negatively predicted food-secure households. Aged households are unable to travel far to access B. Haile Aboye et al. Research in Globalization 8 (2024) 100200 8 farming inputs and work in non-farm and off-farm activities to increase food production and secure household food. This is inconsistent with the finding that aged households in SSA are less likely to become food insecure due to their experiences in risk aversion decision-making and accumulated assets, including remittances (Ouoba and Sawadogo, 2022; Ayal et al., 2023). Education negatively and significantly predicted household food insecurity. A year’s increase in household education decreases the proportion of households with severe and moderate food insecurity. An increase in household head education improves non-farm and off-farm labour opportunities. More-educated households are less likely to be food-insecure since they derive higher-income jobs in non-farm and offfarm activities compared to the less-educated households earning in farming. In SSA, climate change, population growth and land degradation frequently undermine farm yields, food security and farm income of less-educated smallholders, causing them to migrate to urban areas and earn non-farm income. The non-farm income also helps farmers overcome farm inputs and access to farm techniques and production decision-making, which raises farm and labour productivity (Yohannes et al., 2023). Livestock ownership also has a negative and significant impact on food insecurity. Households with livestock assets have the leverage to reduce food insecurity. Households relied on oxen traction power for land preparation. Incomes from livestock sales also enabled them to purchase food from the markets during food shortages or overcome shortages of farming inputs needed to improve household food production against food insecurity (Melketo et al., 2021). Besides, their manure decreases the cost of chemical fertilizers required to increase land productivity. The saved income can be invested in productive assets that protect households from falling into food insecurity. Farm size was an important factor in household food insecurity. Farm households with larger farm sizes are less food insecure than farm households with smaller farm sizes. Larger farm sizes allowed households to diversify crops, increase production, and reduce household food insecurity threats, which was reported in the previous study. Large farmland owners use farming as a basic strategy for household food selfsufficiency. They engaged in the production of diversified crop types with low risk for household consumption. Smaller farms encourage households to intensify marketable crops with high risk (vegetables) or non-food crops with longer harvesting seasons (eucalyptus). In addition, farmland is insufficient to support full-time farming for smallholders. Most SSA smallholders could earn non-farm and off-farm incomes to compensate for the farming income shortfall (Ayal et al., 2023). They intensify agricultural land-use with higher-return crops using income earned from non-farm and off-farm activities (Ejaz and Mallawaarachchi, 2023). Income diversification negatively predicted food insecurity. An increase in income sources decreased the proportion of farm households predicted to be food-insecure. It reduced the financial constraint farm households faced to produce or purchase food when income from another source such as farming is not dependable to ensure food security for the households. Food-secure households could diversify their incomes from non-farm, off-farm, and on-farm sources compared to foodinsecure households that earned income merely from farming. SSA rural outmigration can diversify income and improve household daily calorie consumption, as a failure in farm income may not lead to severe food poverty because of off-farm income (Abebaw et al., 2020; Asefawu, 2022; Getahun et al., 2023). Access to irrigation was significant in influencing household food insecurity status. Households with access to irrigation were less likely to be severe or moderate food insecure, which means they were more likely to fall within the high food security category. Access to irrigation helped households increase multiple-cropping and earn more income to purchase sufficient food for their household members. This accepts findings that agricultural intensification in SSA alleviates food insecurity of households with farm size declining, especially in urban areas because of improved agricultural value chains and input supply chains, such as seeds, fertilizers, labour, pesticides, herbicides, and tractors (Headey et al., 2014). However, this result was inconsistent with the finding in SSA that irrigation generates less cash earnings to pay off household Table 2 Component matrix (factor loadings), eigenvalues, and total variance explained by principal components. Indicators Components 1 2 3 4 5 6 7 8 9 10 11 12 Staple food production per capita 0.79 0.43 0.18 0.06 0.19 0.25 0.21 −0.19 −0.17 −0.12 −0.18 0.02 Income per capita 0.74 0.32 0.03 0.04 0.25 0.13 0.06 0.12 0.24 0.21 0.06 0.09 Consumer price index variability 0.68 0.44 0.13 0.31 0.04 0.28 0.22 0.23 0.14 0.21 0.01 0.12 Staple cropland area per capita 0.65 0.19 0.08 −0.56 0.28 0.23 −0.46 0.12 0.22 0.23 0.13 0.04 Improved water sources 0.63 0.29 0.19 0.02 0.08 0.13 0.21 0.17 0.17 0.08 0.25 0.14 Cropland-use loss 0.59 0.46 0.61 0.02 0.04 0.16 0.05 0.16 0.25 0.37 0.21 0.03 Food market access 0.51 0.55 0.43 0.61 −0.57 −0.44 0.33 0.25 0.02 0.17 −0.03 −0.02 Food production index variability 0.44 0.67 0.01 0.06 0.22 0.27 0.14 0.35 0.11 0.21 0.04 0.11 Sanitary facilities (toilet) 0.37 0.53 0.47 0.04 0.22 0.18 0.11 0.19 0.22 0.04 0.15 0.23 Food consumption score 0.36 0.62 0.25 0.15 0.27 0.14 0.12 0.13 0.26 0.11 0.27 0.01 Dietary diversity score 0.19 0.18 0.63 0.08 0.04 0.14 0.21 −0.17 −0.07 −0.06 0.13 0.02 Food storage 0.12 −0.35 0.46 0.09 0.26 0.18 −0.13 0.21 0.21 0.12 0.21 0.02 Initial Eigenvalue Total 3.57 2.38 1.53 0.83 0.75 0.62 0.57 0.48 0.42 0.40 0.32 0.11 Variance (%) 29.77 19.87 12.77 6.92 6.25 5.20 4.75 4.03 3.53 3.36 2.65 0.90 Cumulative (%) 29.77 49.64 62.41 69.33 75.58 80.78 85.53 89.56 93.09 96.45 99.10 100.00 Extraction method: Principal Component Analysis; Kaiser-Meyer-Olkin (KMO) test of sampling adequacy: 0.702; Bartlett’s test of sphericity: Approx. Chi-square: 348.294, df 32, and Sig: 0.021. Fig. 2. Food insecurity characteristics of the sampled households. B. Haile Aboye et al.