1 Introduction

Humans must have access to sufficient and nourishing food [1]. Achieving food security is critical, as it directly impacts not only individual survival, but also societal progresses. It encompasses basic dimensions namely, availability, accessibility, utilization, and stability, which are constrained by multiple determinants. When communities or individuals lack food security, it hampers their ability to thrive and fully develop [2] [3]. Food insecurity arises whenever economic and physical access to sufficient, safe, and nutritious food becomes unreliable [4] [5]. Projections show that approximately 9.9% of the global population experienced food insecurity in 2020, and this figure is expected to rise in the future [6] and [7]. The world's urban population grew by 1.84% annually between 2015 and 2020, and is expected to expand by 1.63% annually between 2020 and 2025. It was estimated that 54% of the world's population lived in urban regions in 2016; by 2030, this figure is expected to rise to 60% [8]. The United Nations Department of Economic and Social Affairs projects that from 55% in 2018, 68% of people on earth will live in urban areas by 2050. According to UN DESA [9], this suggests that nearly two out of every three people reside in urban settings. By the same year, an additional 2.5 billion people are predicted to live in urban areas due to increased urbanization, with about 90% of this growth taking place in Asia and Africa. This means that it will have a big impact on urban food insecurity because of the rapid increase in urban population, leading to food insecurity in cities [1], [10].

Rapid growth of urban populations often leads to stressed food systems. This stress arises from various interactions between people and the environment which affect food security throughout all stages of food production and consumption [1]. Urban populations typically rely heavily on purchasing food from markets rather than producing it themselves. While some urban households engage in small-scale food production through activities like backyard gardening or rooftop farming, these contributions are often modest compared to the overall food needs of the household [11]. Urban households that rely heavily on purchasing food are indeed vulnerable to food price shocks, particularly those that impacting low-income urban populations. For those living in poverty, who often spend a large percentage of their income on food, even small fluctuations in prices can have significant consequences, potentially leading to food insecurity and malnutrition. Despite ongoing urbanization across the African continent, investigations into the vulnerability of economically disadvantaged urban households to food insecurity have not received the same level of attention as those in rural areas. [12]. Yet, the urgency of addressing food security in urban areas is growing as the population of these places’ increases.

Similarly, the food security issue is indeed a significant challenge for economically poor and low-income households in the urban and rural areas of Ethiopia. However, advocacy and policies on food security have traditionally focused more on rural areas [13] and [14]. Today, several pieces of evidence show that urban food insecurity is a growing problem that requires dedicated attention and approach [15]. Ethiopia’s government has implemented various programs related to food security, such as the Urban Productive Safety Net Program, considering it as one of the major development problems of the country. The challenge in small towns up to the big cities of the country, and in the study area in particular, is immense. This is because of the complex nature of food security and the need for clear identification of prevailing impediments [16]. On the other hand, there is a longstanding conceptualization of urban centers as more developed and homogeneous areas [17].

Ethiopia has the fastest annual urbanization rate (4.5%), exceeding the average sub-Saharan Africa of 4% [18]. The population projections for Ethiopia show that the urban population will more than double in the next 20 years, reaching 42.4 million by 2037. This is because rural villages are upgraded to towns, and increasing migration to new urban centers can lead to unexpected food shortage issues [19]. Inquiries show that in Ethiopia, two out of five households are food-deficient, with little difference between urban and rural areas [20]. In addition, three of the four households experienced food insecurity among the urban populations [21]. Despite the sociocultural, economic, and ecological diversity of Ethiopian cities, along with the variability in day-to-day food costs and conditional differences among urban communities, there is little attention from the researchers on the issue of food security in urban settings. In line with this, studies such as those by [22,23,24,25,26,27,28], have focused on rural areas with little understanding of urban settings. Such rural-oriented studies have not examined the underlying causes of household susceptibility to food insecurity in an urban context. On the other hand, studies have been conducted on vulnerability to food insecurity in major Ethiopian cities such as Addis Ababa, Gondar, and Bahir Dar [14, 15, 29] among others. However, the inquiries were limited in scope to particular geographic areas within the country and did not comprehensively cover towns in southern Ethiopia.

In the urban oriented studies of the Wolaita zone, the studies by Tadesse et al. [30] and Samuel et al. [31], conducted in Sodo and Areka towns in southern Ethiopia, have aimed to analyze urban food security issues. However, they were conducted approximately ten years ago. Therefore, it does not reflect current urban development, the economic landscape, or prevailing levels of food insecurity, and do not consider the vulnerable groups of urban households. To solve the problem, which has been increasing from time to time, a scientific study considering the current situation was found to be necessary. Thus, this study aimed to answer the questions: (1) Currently what empirical evidence can be provided on the vulnerability of urban households to food insecurity? (2) What is the level of vulnerability to food insecurity? (3) What are the factors influencing the vulnerability of households in the study areas to food insecurity? Finally, this research is structured into introduction, review of relevant literature, detailing the materials and methods, presenting results, discussions, limitation, concluding with policy recommendations, and areas for further studies.

2 Review literature

2.1 Approaches and conceptualizing vulnerability to food insecurity

Vulnerability is defined as the diminished capacity of individuals or communities to anticipate and withstand the impact of both natural and man-made hazards [32]. Food vulnerability arises when shocks related to food put pressure on a household's capacity to obtain food that is both safe and nutritious [33]. This pressure originates from the interactions of biophysical and socioeconomic systems operating at various levels, impacting the household. Two approaches- outcome vulnerability and contextual vulnerability, can be explored to understand these systems. The approach of outcome vulnerability has taken into consideration, especially in recent times, the growing interest in analyzing outcomes related to climate change. It primarily concerns biophysical changes in a closed system with a boundary between nature and society [34]. As a result, the role of socioeconomic components in ameliorating the effect of risk was not explicitly included. The approach is dependent on reliable panel data. However, according to [35] and [36], panel data is frequently unavailable in many developing countries, and even when accessible, it often lacks representativeness of cross-sectional components, reducing its usefulness for policy analyses. In contrast, a contextual vulnerability perspective involves an exploration of various factors influencing the socio-economic and ecological systems. It necessitates that researchers embrace a multidisciplinary approach. Contextual vulnerability utilizes cross-sectional data to evaluate socioeconomic factors that contribute to vulnerability. This encompasses aspects such as social marginalization, economic inequality, accessibility of household food and resources, efficacy of local support institutions, and the prevailing economic and political systems. This approach identifies social groups that are more prone to food insecurity, rendering it highly pertinent to conceptually underpin the development of a comprehensive vulnerability index for food insecurity.

2.2 Household vulnerability to food insecurity

Household vulnerability to food insecurity can be understood as its exposure, sensitivity, and adaptive capacity (Fig. 1). Exposure pertains to shocks related to food that affect a household's ability to access safe and nutritious food. It is broadly defined as the extent to which a system is exposed to the risk or impact of such shocks [37]. These shocks, when they occur, have varying effects on different levels of the system. Food-related shocks threaten household food availability. For instance, health problems can threaten a reduced ability to work and earn income, further exacerbating food insecurity [38] [39]. Households facing food-related shocks utilize their resources and implement various measures to mitigate losses and enhance their overall food security. Viewing this from a vulnerability standpoint, a household's ability to respond is referred to as its adaptive capacity, which includes the social and technical skills and tactics used by individuals and communities to cope with environmental and socioeconomic shifts. In the sphere of food systems, adaptive capacity is often utilized to maintain livelihoods, sustain food production, or ensure access to food [40]. The concept of adaptive capacity is generally recognized as a favorable characteristic in reducing the vulnerability of a system. A household with high adaptive capacity stands a stronger chance of effectively coping with food vulnerability [41]. Households that face ongoing shocks often rely on their assets to deal with the resulting stress. If their resources are insufficient, they may resort to extreme coping strategies, leading to fluctuating levels of food insecurity over time. In terms of food security, these recurring episodes can be seen as a measure of food resilience, and in discussions about vulnerability, this is known as susceptibility [42]. In this setting, sensitivity is characterized as the fundamental vulnerability of a household to address food-related shocks. The reaction of a household to such an event result in various impacts along the range of food vulnerability. These outcomes can be utilized to classify households into different categories based on their level of vulnerability to food insecurity: those that are not vulnerable, moderately vulnerable households, and highly vulnerable households [43]. Therefore, we concluded that applying a contextual perspective on vulnerability and utilizing the framework of vulnerability to food insecurity as determined by exposure indicators, adaptive capacity, and sensitivity (see Fig. 1).

Fig. 1
figure 1

Source: Created by the authors, 2023

Conceptual framework of vulnerability to food insecurity.

Full size image

2.3 Urban food insecurity in Ethiopian context

Our country Ethiopia, currently ranks as the second most populous in Africa [44], with a population exceeding 117.9 million. While it is surpassed by Nigeria with over 211 million inhabitants, Ethiopia’s urbanization rate of 25% [45] lags behind that of other Sub-Saharan countries at 37% [46]. Over the years, there has been an increase in the proportion of urban residents in Ethiopia-from just 6% in 1960 to an estimated 23% in 2022 [47] [48]. This rapid pace of urbanization is expected to continue, leading to approximately 42% of the total population residing in urban areas by the year 2050 [46]. Consequently, this surge poses new challenges for food security within Ethiopian urban settings.

Urban food insecurity and vulnerability among households in Ethiopia have become complex issues of recent concern. Although previous assessments have concentrated on rural areas, the situation in urban centers is now being recognized as critically important due to various factors encompassing social, economic, cultural, and institutional aspects. A study conducted in urban settings of Ethiopia revealed that 33.1% of participants were experiencing food insecurity [49]. The study also revealed that 9 out of every 100 individuals had experienced instances of ‘sleeping hungry,’ while another 6 individuals per hundred reported 'not eating throughout the day or night.’ Furthermore, household food insecurity showed significant associations with the gender of the household head, engagement in casual labor activities, and homeownership status. Additionally highlighted was a crucial need for comprehensive interventions like integrated urban housing development initiatives, job security measures, urban agricultural programs alongside inclusive safety net projects aimed at addressing urban food insecurities. An examination undertaken by [50] assessed households' vulnerability to food insecurity within East Hararghe areas of Ethiopia and found high levels indicating nearly one-third (32%) are undernourished. As well as nearly two-fifths (40%) consume less than daily recommended calorie intakes respectively.

In a recent study conducted in Southern Ethiopia by Mamo and colleagues [51], it was found that 91.4% of the two hundred thirty-eight interviewed households experienced food insecurity. The research indicated that household food insecurity exceeded the national average, with factors such as the current occupational status of the household head and dependency ratio significantly contributing to this issue. Other studies have also suggested that aspects like homeownership [32], health status [50], asset possession [15], among other factors play a crucial role in ensuring food security across different regions of Ethiopia.

3 Materials and methods

3.1 Overview of the study area

This research encompasses three towns within the southern Ethiopia: Sodo, Areka, and Boditi (Fig. 2). Sodo is situated approximately 329 km along the Addis Ababa-Hosanna route. It is located at 6° 53′ 59" North latitude and 37° 44′ 59" East longitude. Sodo, currently comprising seven Kebeles, functions as the administrative center of the Wolaita Zone and holds significance as an administrative and political hub within the Southern Ethiopian Region. Areka, situated 350 km away from Addis Ababa, lies at an elevation of 1600 m above sea level and is located at 7º 04′24″ North latitude and 37º 41′30″ East longitude. This town experiences an annual temperature of 22 °C and an average rainfall of 1123.15 mm. Boditi, the third town to study, was founded in 1945. It is among the reformed towns in the southern Ethiopia, with a structured plan and town administration. According to the report of Wolaita Zone Finance and Economic Development, the urban population within Wolaita zone totals 821,886 (405,570 males and 416,316 females), and the three study towns, Sodo, Areka, and Boditi, have a total household population of 283,760 (140,433 males and 143,327 females), 60,208 (44,446 males and 45,762 females), and 78,981 (38,913 males and 40,068 females), respectively. The population of urban areas is experiencing an annual increment of 4.8%.

Fig. 2
figure 2

Source: Designed by Author

Location of the Study Area.

Full size image

3.2 Study design and approach

This study employed a mixed-methods approach, considering concurrent embedded mixed-method procedures. Castro and his colleagues [52] and Schooneboom & Johnson [53] suggested that, in concurrent data collection, both qualitative and quantitative data are collected simultaneously. Moreover, simultaneous data collection in mixed method research is important, as it allows for a more comprehensive understanding of the research [54]. Considering this perspective, we employed a concurrent nested research design to collect data simultaneously throughout the data collection.

3.3 Sampling techniques

This research is likely to involve collecting data from sample households or beneficiaries of the UPSNP in the three sample towns (see Table 1). The research employed a multi-stage sampling approach; study areas were purposefully considered primarily. After consulting with higher officials, four study Kebeles were purposefully nominated according to certain criteria, such as the number of beneficiaries and spatial distribution. Third, the sample frame was determined from the beneficiary lists of the UPSNP for each Kebele administration. Subsequently, 310 respondents were identified through a systematic random sampling method. This involved identifying the first household randomly and then selecting every Kth interval household from the lists of beneficiaries in each sample Kebele until the desired number of sample households was reached. The total sample size was derived from the Yamane [55] formula, which is commonly used when the population is known [56]. This formula was widely applied in recent studies by Masese et al. [57], Tora et al. [58] and [59]. The sample size of 310 was determined using Yamane’s formula. This formula is expressed as follows:

$$n = {\text{ N}}/{1} + {\text{N}}(e^{2} ) \, = { 1363}/{1} + {1363}\left( {0.0{5}} \right)^{2} \, = { 1363}/{1} + {1363}\left( {0.00{25}} \right) \, = { 31}0$$
(1)

where, n = sample size being determined, N = refers total population (beneficiaries of UPSNP in selected Kebeles) in the areas, 1 = is constant and e = refers the error term perceived to be 0.05 level of precision.

Table 1 Summary of the selected sample towns, Kebeles, and total sample size
Full size table

3.4 Sources of data and instruments

The data collection involved gathering both qualitative and quantitative pieces of information from primary and secondary sources. Primary data were acquired through surveys conducted with households, interviews with key informants, and focus group discussions. Moreover, published and unpublished materials such as journals, books, articles, and reports were used as secondary data sources.

3.5 Analysis

Data analysis encompassed qualitative and quantitative methods. Qualitative data were analyzed using narration and summarization techniques, while quantitative data were subjected to statistical methods for analysis. Descriptive statistics were first generated and employed to provide a summary of household attributes (UPSNP beneficiaries). To classify the vulnerability levels to food insecurity among the study participants, we utilized the Vulnerability to Food Insecurity Indicator (VFII). To identify critical factors that affect households prone to food insecurity in study areas, an ordered logit model (OLM) was used. IBM SPSS Statistics version 27 and Microsoft Excel were used for analyses.

3.5.1 Vulnerability to food insecurity index (VFII)

The Vulnerability to Food Insecurity Index (VFII) is an inclusive food security indicator that assesses households to food insecurity. This index operates under the premise that households are prone to food insecurity as a result of existing socioeconomic difficulties and incapacity to cope with unexpected food-related challenges. After selecting each variable that affects households’ vulnerability to food insecurity, we normalize these variables using the min–max normalization method shown in Eq. 2 [36].

$$VFII^{\prime}{ } = { }\frac{{\varvec{X^{\prime}} - {\varvec{Xmin}}}}{{{\varvec{Xmax}} - {\varvec{Xmin}}}}$$
(2)

where x is original value, X’ is normalized value, Xmax and Xmin are the minimum and maximum values of the normalized vulnerability to food insecurity index (VFII) and having the values between 0 or 1. Furthermore, we generated weights for these variables employing equal weights for the VFII variables. Finally, to calculate the VFII score, the aggregation method shown in Eq. 2 was used. Wherein E is the exposure, S is the sensitivity and AC is the adaptive capacity.

$$\text{VFII }=\sum \text{AC}-\left(\sum \text{E}+\sum \text{S}\right)$$
(3)

Households with lower a VFII composite values are more vulnerable to food insecurity compared to households with higher VFII composite scores. However, the higher the composite value of VFII, the lower the influence of vulnerability on household food security [42, 50].

3.5.2 Classification of VFII

A household can be highly exposed to or sensitive to the insufficiency of food; nevertheless, this is not a sufficient condition to say that this household is vulnerable to food insecurity. Hence, a vulnerable household is characterized by limited adaptive capacity, hindering its ability to cope effectively with stress resulting from exposure and sensitivity. A household is considered vulnerable when its VFI score is above, below, or within a certain level. According to Sileshi et al. [50], households were categorized into several food insecurity and vulnerability categories based on their VFI scores. We established our VFII threshold as the point at which the household's adaptive capacity surpasses the collective impact of exposure and sensitivity. With the VFII comprising three equally weighted components, each component holds a weight of 0.33 in the threshold definition. In other words, each component represents 1/3 of the vulnerability dimensions. We employed the aggregation method in Eq. 3 to compute our VFII. Then we used this score to categorize households into three different food vulnerability groups according to Ojo [43] and Ibok et al. [42] classification procedures. The first group of households classified as not vulnerable to food insecurity are those with a composite score exceeding 0.14. The next category comprises households moderately vulnerable to food insecurity, with composite scores greater than 0.065 but falling below 0.14. The final group consists of households highly vulnerable to food insecurity, representing the most severe levels of food insecurity and vulnerability, characterized by composite scores below 0.066 (see Table 2).

Table 2 Threshold for VFII among urban households
Full size table

3.5.3 Statistical analysis

We used an ordered logistic model (OLM) to investigate the factors influencing food insecurity among households enrolled in the UPSNP program. This method is particularly useful for analyzing ordinal data such as vulnerability to food insecurity categories. Previous studies have utilized ordered logit and probit models to investigate similar multiclass problems [60, 62]. The ordered logit model (OLM) is a suitable approach for examining ordinally dependent variables. It facilitates the estimation of the multinomial logistic regression model parameters of Swindale and Bilinsky [61, 63, 64]. Unlike standard regression analysis, which assumes continuous dependent variables, OLM operates on a latent regression of ordinal scales. The categorical dependent variable (level of vulnerability to food insecurity) takes the values (1, 2…, j) [44]. Following [45], the specification of the ordered logit model can be expressed as:

$$Yi^* = Xi\beta + \varepsilon i$$
(4)

where Yi* is the latent unobservable variable with more than two ordered or classified categories and denotes the level of vulnerability observed in the household i, X is a vector of explanatory variables i, β is the vector of parameters to be estimated, and εi is a random error term distributed independently. In the OLM, the ordinal variable Yi here is the observed discrete, ordinal rating on a three-point scale for, vulnerability level of households, i.e., Yi = 1, 2, or 3 for not vulnerable, moderately vulnerable, and highly vulnerable categories, respectively. Yi is thus demonstrated in the following formulas [64, 65].

$$Yi = \left\{\begin{array}{ll}1 & if\; yi* \le \mu 1\;(Highly\; vulnerable\; to\; food\; insecure)\\ 2 & if\; \mu 1<yi* \ge \mu 2\;(moderately\; vulnerable\; to\; food\; insecure\\ 3 & if\; yi* \ge \mu 3\;(Not\; vulnerable\; to\; food\; insecurity)\end{array}\right.)$$
(5)

According to [46], µ1 to µ3 are cut-off points for food insecurity. The model estimation is performed by the maximum likelihood method. The maximum likelihood estimator can be represented as [65]:

$$\begin{aligned}& logL={\sum }_{i=1}^{N}{\sum }_{j=1}^{J}Yijlog\left[F\left({u}_{j}-{x}_{i}^{t}\beta \right)-F({u}_{j-1}-{x}_{i}^{t}\beta )\right], \\ & \text{Where},Yij =\left\{\begin{array}{ll}1 & if\; y1=j\\ 0 & otherwise\end{array}\right. \end{aligned}$$
(6)

Based on the literature and observations in the study area, 16 explanatory variables were selected for to estimate the model. The Variance Inflation Factor (VIF) is an important tool for identifying and addressing multicollinearity, which can affect the statistical significance of independent variables in regression analysis. This is typically addressed before executing the model. A VIF value below 5 is considered acceptable [66,67,68]. While VIFs above 5 and below 10 indicate that they may be problematic, those exceeding 10 generally indicates serious multicollinearity that requires corrections such as eliminating variables or employing other techniques to resolve the issue [69]. Thus, we evaluated multicollinearity among the predictor variables by ensuring that the VIF was less than 5. This examination confirmed that there was no collinearity between the chosen variables (see Table 7 in Appendix-A).

3.5.4 Variables

Vulnerability to food insecurity as a dependent variable for this study indicates (1) those highly vulnerable to food insecurity, (2) moderately vulnerable, and (3) not vulnerable. This was measured on an ordinal scale. For example, this ordered sequence reflects an increasing level of vulnerability, where households categorized as highly vulnerable are expected to face a higher degree of food insecurity compared to those labeled as moderately vulnerable, and so on. Based on previous related literature, the selection of explanatory variables in the model was hypothesized to affect urban households’ vulnerability to food insecurity in (Table 8 in Appendix-B).

4 Results

4.1 Respondents’ characteristics

As presented in Table 3, among the sampled population, there was nearly an equal distribution in terms of gender. Approximately 50.3% of the individuals were male, and 49.7% were female. Among the 310 households whose heads were part of the UPSNP beneficiary program, different family structures were observed. About 21.6% of these households were categorized as single, indicating that they were composed of one adult, either living alone or with dependent family members or typically refers a household led by a single parent. The majority, accounting for 65.2% of these households, were identified as having a nucleated family structure. This finding suggests that these families typically consist of parents and their children living together in one household unit. 13.2% of the households were classified as having an extended family structure, which means they might include relatives beyond the nuclear family, such as grandparents, aunts, uncles, or cousins living together.

Table 3 Statistical Summary of categorical dependent variables (N = 310)
Full size table

The Table 3 also reveals that approximately 2.3% were categorized as singles, indicating individuals who were unmarried. The majority, accounting for 52.9% of the household heads, were reported as married, suggesting that they were in a marital union. About 17.4% of the household heads were classified as divorced, signifying that they had previously been married but were no longer. Moreover, 27.4% of the individuals were classified as widowed, signifying the loss of their spouses. Similarly, 27.4% of these household heads had no formal education, indicating a lack of schooling. The largest group, comprising 40.0% of household heads, had completed primary education. About 18.1% attained a secondary education level, signifying the completion of higher schooling beyond primary education. A smaller percentage (14.5%) had achieved college-level education or above, suggesting that they had received higher education beyond secondary school. Among surveyed household heads, various occupational statuses were observed in (Table 3). A small proportion (7.4%) was engaged in informal employment activities. The majority, comprising 59.4% of the household heads, were involved in casual or daily labor work. 31.0% were engaged in petty or small-scale trade activities. A smaller segment, approximately 3.2%, reported relying on remittances as a source of income. A majority, accounting for 81.9% of the household heads, were employed as public workers, indicating they participated within the public worker category of UPSNP. In contrast, a smaller proportion, approximately 18.1%, received direct support from the program. A majority, constituting 55.2% of the surveyed heads, lived in a rented house, acquiring housing through private rentals. About 25.5% of these heads owned their homes, signifying outright ownership.

The finding reveals that more than 19% of the household heads reside in kebele houses. The respondents were asked about their health status, and the results indicated a fairly balanced distribution. 48.1% of the respondents reported falling ill at some point, while approximately 52% stated that they had not experienced illness during the specified period. Regarding the urban agricultural practices of households, 77.1% of the respondents reported not participating in urban agriculture practices. In contrast, a smaller portion, about 22.9%, mentioned that they are engaged in urban agricultural activities. Table 8 shows 70.6% of the respondents having access to credit services. In contrast, about 29.4% mentioned not having access to these financial services. Regarding socio-cultural activities such as funerals, festivals, dowries, graduation ceremonies, etc., 85.5% of respondents affirmed the existence of socio-cultural practices within their community and were affected by such practices. During the survey, households were asked about the impact of food prices on their lives. The result showed a notable pattern, with roughly 91.9% of the households mentioning that they felt the impact of food prices. In addition, around 81.9% of the surveyed households reported not having any assets.

The result in Table 4 shows 25 8.1% that the age of respondents indicates a mean age of nearly 47 years, with a standard deviation of 11.24. The minimum reported age among the respondents is 22 years, while the maximum reported age is 80 years. When considering family size among a sample of 310 respondents, the mean family size is calculated at 5.41 individuals per household. The mean of 2.08 dependent family members among the surveyed households provides insight into the average number of individuals who rely on others within these families. Dependent family members typically include children, elderly individuals, or those with disabilities or specific care needs. On the other hand, the number of household members benefiting from the Urban Productive Safety Net Program (UPSNP) ranges from a minimum of 1 to a maximum of 4 individuals per household. The (Table 4) reported information suggests that among Urban Productive Safety Net Program (UPSNP) households in the surveyed sample, the average monthly income ranges from a minimum of 800 to a maximum of 3650 Ethiopian Birr.

Table 4 Summary of statistics of continuous dependent variables (N = 310)
Full size table

4.2 The level of households’ vulnerability to food insecurity measured by VFII

The categorization of households by the Vulnerability to Food Insecurity Index (VFII) we used in this paper (see Table 5). The study found that a substantial proportion of households were vulnerable to food insecurity, with 41.0% being highly vulnerable, 48.4% moderately vulnerable, and only 10.6% not vulnerable. This suggests a notable prevalence of vulnerability, particularly in the high and moderate categories, among the sampled households.

Table 5 The household’s level of vulnerability to food insecurity among the study towns
Full size table

4.3 Determinants of urban household vulnerability to food insecurity

Understanding the factors contributing to food insecurity among urban households is crucial for comprehending the underlying reasons for their vulnerability. The dependent variable in this analysis is vulnerability to food insecurity, taking a value of 1 for highly vulnerable households, 2 for moderately vulnerable ones, and 3 for those that are not vulnerable. Given the sequential nature of the dependent variable, we opted for the ordered logit model (OLM) as the preferred approach to analyze the factors influencing vulnerability to food insecurity. The results of the estimation are detailed in Table 6. Before conducting the analysis, we assessed variables thought to impact household vulnerability to food insecurity for multicollinearity using the variance inflation factor (VIF). The test results confirmed that there is no strong correlation among independent variables (see Appendix -1). Among the 16 variables, households’ ownership of houses, household health status, urban agriculture, socio-cultural practices, asset ownership, age, and average monthly income were found to be significant in determining the vulnerability of households to food insecurity.

Table 6 Determinant factors associated with urban household vulnerability to food security (OLM)
Full size table

The information about how well the OLM fits and its effectiveness in explaining vulnerability to food insecurity is summarized in Table 8. The Likelihood Ratio Test (LRT) was employed to compare the intercept-only model with the complete model. The results revealed a notable disparity in the −2 log-likelihood, with a chi-square value of 246.193 and a highly significant p-value (P < 0.001, df = 16). This indicates that the full model is a good fit for the data, signifying its adequacy in explaining the observed patterns. Other statistical tests, such as Pearson chi-square (433.866, P > 0.05, df = 602) and deviance chi-square (346.098, P > 0.05), further support the overall fitness of the model. The pseudo-R-square estimates (Nagelkerke = 0.643) indicate that around 64.3% of the variability in predicting the dependent variable is explained by the selected independent variables. The −2 log-likelihood was highly significant (−2LL = 346.098, chi-square at df = 16, P < 0.001), and the likelihood ratio test also confirmed the model's validity. In Table 6, we presented the estimated coefficients and odds ratios for each explanatory variable, along with their P-values testing their significance. A negative coefficient implies a decreasing impact on household vulnerability to food insecurity, indicating a potential protective effect. Conversely, a positive coefficient suggests an increasing in vulnerability compared to the reference category, signifying an increased vulnerability to food insecurity.

5 Discussion

The issue of food insecurity extends to urban areas as well, affecting households beyond rural regions. While urban areas may have an abundance of food, not everyone has equal access to nutritious and affordable diets. Now, the issue is a growing concern, as many urban households, particularly the poor, are highly susceptible to income and food related shocks [2]. As a result, they are vulnerable to food insecurity at various levels. There are critical factors that contribute to the vulnerability of urban households to food insecurity, resulting in different levels of vulnerability, ranging from relatively low to highly vulnerable. The study examines the concerns of urban households regarding the factors that contribute to their vulnerability to food insecurity. The results of Table 5 showed that nearly (89.4%) of the sample households in the study towns experienced varying levels of vulnerability to food insecurity. Studies such as [70] found that 80.9% of urban slum households in Ibadan, Nigeria, had food insecurity. Studies from Ethiopia such as [21] and [51] reported that among respondents, 77.1% and 91.4% of urban households in Addis Ababa and Arba Minch respectively experienced food insecurity, while Mabuza and Mamba [70] found that 88% of the surveyed households in one of the densely populated informal settlements in Mbabane, had experienced food insecurity. Furthermore, [71] reported a 90.6% prevalence of food insecurity in southwestern Nigeria. Several factors contribute to the vulnerability of urban households to food insecurity in the study areas, particularly those low-income urban households [32]. Among the factors, this study highlights (Table 6) the significant ones that affect households’ vulnerability, discussed as follows:

5.1 Age

In an ordered logistic regression model, the coefficient for the independent variable age was positive and significant at a significance level of p < 0.083* or 10%. This suggests that as the age of the household increases, there is an increased probability of it being moderately or highly vulnerable to food insecurity. This association is primarily attributed to the rise in the dependency ratio, indicating an increased ratio of dependents to earners within the household. This can lead to a decrease in the household's income and purchasing power, making it more difficult for the household to access sufficient food resources. The finding aligned with a study conducted in East Hararghe, Ethiopia found that vulnerability to food insecurity increased with the age of the household head [50]. The result is also associated with [72]. In addition, the study by Miller et al. [73] also suggests a potential association between age and vulnerability to food insecurity. The odd ratio (OR = 1.023) refers to the fact that as the age of households increases by one unit, the odds of experiencing moderate or high vulnerability to food insecurity are estimated to increase by approximately 2.3%.

5.2 Average monthly income

Table 8 OLM results show income was found to affect households’ vulnerability to food insecurity negatively and significantly at a p value of < 1%. The estimated coefficient of −0.002 signifies that for every one-unit increase in monthly income, there is an associated decrease in the odds of transitioning to a moderate or highly vulnerable food insecurity category. This inference aligns with the odds ratio (OR) of 0.998 and shows that with each unit rise in monthly income, the odds of experiencing moderate or highly vulnerable levels of food insecurity decrease by approximately 0.2%. The statistical significance reflected by a (p = 0.000***) underscores the robustness of this relationship. These findings suggest that higher monthly income is associated with a reduced likelihood of moving towards moderate or more vulnerable categories of food insecurity. A study conducted by Mekonen and colleagues [15] found that household income was a significant driver of urban household food insecurity in Bahir Dar and Gondar towns, Ethiopia. Other similar studies, such as [32] and [74], support our finding. Therefore, while the impact of each incremental rise in income may seem slight in our findings, this is due to the fact that nearly 81% of UPSNP beneficiaries rely on public work programs. They invest significant time in the program and lack the opportunity to seek additional work. Despite this, a collective and gradual improvement in income can significantly boost households' capacity to obtain sufficient and healthy food.

5.3 Health status of households

Health status was found to affect households’ vulnerability to food insecurity negatively and significantly at p < 1%. Its estimated coefficient is −1.061, with an odds ratio (OR) of 0.346 and a significance level of 0.000. This means that for each unit increase in health status, the odds of households failing to be in the moderate or highly vulnerable food insecurity category decrease by a factor of 0.346, or equivalently, the odds decrease by 65.4% for each one-unit increase in health status. The significance level of 0.000*** is very strong evidence, indicating that health status is a significant determinant of vulnerability to food insecurity. Existing studies, for example, [50] and [74], have also used similar models and found that health status was a significant determinant of food insecurity, with households with poor health status being more vulnerable to food insecurity than households with good health status. In other words, our finding indicates that households facing health-related issues, such as occasions of illness, increase the risk of vulnerability to food insecurity. This is due to the fact that falling sick can lead to a loss of income and increased medical expenses, impacting the ability to purchase an adequate amount of food, thereby contributing to food insecurity. A similar influence was also reported by [75], who showed that nearly 87.4% of vulnerable people, including those living with HIV, in Ethiopia are intensely affected by food insecurity, facing a shortage of sufficient and nutritious food. Moreover, this finding is consistent with other studies that have shown a strong relationship between health status and food insecurity [76].

5.4 Participating in urban agriculture

Table OLM analysis investigated the impact of participating in urban agriculture on the vulnerability to food insecurity, considering categories. The estimated coefficient for participating in urban agriculture is −1.211, accompanied by a significant odds ratio (OR) of 0.298 and a statistically significant p-value of 0.006**. This indicates a substantial inverse relationship between urban agriculture participation and the odds of being in the moderate or highly vulnerable category compared to the reference group (not vulnerable). Notably, studies by [77] reported similar negative associations between participating in urban agriculture and varying degrees of vulnerability to food insecurity. Our results, therefore, contribute to this growing body of evidence, emphasizing the potential of urban agriculture to mitigate vulnerability across different strata of food insecurity. The negative coefficient and OR suggest that households engaged in urban agriculture are less likely to fall into the moderate or highly vulnerable category in terms of food insecurity. During our data collection period one of Key Informant, regarding the participation of households in urban agriculture, suggested that “most urban households faced limitations due to they lacked access to adequate yards or space. On the other hand, individuals who owned their own houses and had large spaces were able to actively participate in urban agriculture and benefit from growing cabbage, carrots, and fruits. It helps to fulfill food needs and sometimes as an income source.” These findings highlight the importance of urban agriculture initiatives as a strategy to enhance food security resilience, offering valuable insights for policymakers and practitioners seeking effective interventions in the context of diverse vulnerability levels [48].

5.5 Socio cultural practices

In both urban and rural areas of Wolaita, people follow and support several customs. Some of them are harmful and indirectly affect households’ food security status. Table 5, the coefficient of −1.663 refers to the socio-cultural practices (like excessive expenses during funerals, marriage or dowry processes, celebrations, etc.) inverse association between vulnerability and food insecurity significant at 0.000***. The OR of 0.190 (95% CI −2.58 to −0.745) suggests that households that do not participate in socio-cultural practices are associated with a much lower likelihood of being vulnerable to food insecurity compared to their counterparts. The study's findings are consistent with existing research on the social determinants of food insecurity, which highlights the influence of cultural and social factors on individuals' food security status [78]. Additionally, [79] conducted research that revealed a negative relationship: households that face significant influences from sociocultural practices have an 83.2% lower likelihood of achieving food security compared to households that do not face significant influences from sociocultural practices. During the data collection period, one of the participants in FGDs said, “In our community, we have special customs and traditions that hold great significance, including practices like giving kilograms of grain and thousands of moneys as condolences upon a person’s passing, observing dowry customs, and commemorating anniversaries with grand celebrations. However, it’s an inherent part of our culture that those who might lack the necessary funds often resort to borrowing or selling their possessions to uphold these traditions.”

5.6 Asset ownership

The results of the logit model in Table 8 indicate that households’ asset possession (ownership) has a significant impact on their vulnerability to food insecurity in study areas. Specifically, individuals with assets demonstrated notably lower odds of experiencing vulnerability to food insecurity (coefficient = −0.630, OR = 0.196, p < 0.000***), indicating a substantial decrease of approximately 80.4% in the likelihood of being vulnerable to food insecurity compared to those without assets. Owning or having access to certain assets or equipment can sometimes provide households with opportunities to generate income. For instance, individuals or families might rent out their tangible assets or items to others in exchange for money that could then be used to purchase food or other necessities. The finding also corroborated previous literature showing that asset possession was a significant factor affecting food security in urban households [13, 72, 80]. Another study by [81] has consistently shown that greater asset possession, encompassing property and valuable goods, is often correlated with improved food security outcomes.

5.7 House owner

The OLM Table 5 indicates a significant association between asset ownership and levels of household food insecurity. Those who reported being the owner of a house showed a distinct difference in their vulnerability status compared to those who did not own (who live in a rental house). The result shows that households that own their house have a negative coefficient of −0.741 and an odds ratio of 0.477, with a significance level of 0.017**, indicating that they are less vulnerable to food insecurity. This finding is supported by other studies that have found a positive association between housing ownership and food security [82] [74]. Homeownership provides households with a sense of stability and control over their living situation, which can positively impact their food security status. Moreover, according to [53], owning a home can provide households with a financial asset that can be used to access credit or generate income, which can also contribute to their food security.

6 Conclusion and recommendation

Based on the study's objective, urban centers in the Wolaita zone among Urban Productive Safety Net Program beneficiary participants face high levels of food insecurities, with 41.0% being highly vulnerable to food insecurity and nearly 48.4% being in the moderately vulnerable to food insecurity category. The increased occurrence of household vulnerability to food insecurity in our study may be attributed to the focus on poor urban households enrolled in the UPSNP support program (since being poor is a criterion for USNP), unlike prior studies from various areas that did not involve safety net programs and did not focus on vulnerable households. To tackle the factors mentioned earlier, the government, non-governmental organizations (NGOs), and individuals engaged in urban development, social welfare, poverty alleviation, or food security initiatives in the study areas are implementing specific support plans. These efforts are directed towards urban households facing vulnerabilities, with a special emphasis on those benefiting from the UPSNP in selected towns. Initiatives should focus on income-generating activities or boosting participant numbers in PSNP, payment levels, and microfinance opportunities. Moreover, even though the current update is explained (Appendix C), it is comparatively lower than other zones of the region and it should consider the current living conditions of urban areas and the price of food. Based on the findings from the study, policies should also promote and facilitate urban agriculture initiatives, as the study found it to be a significant factor in reducing household vulnerability, moderating cultural expenses, and promoting asset-building initiatives are also important. These strategies can effectively mitigate urban household vulnerability and enhance food security among vulnerable groups in the study areas.

7 Limitation and future research direction

The research relied on a sample comprised of households actively enrolled in the UPSNP at the time of the study. It is important to note that while there may be waiting lists for households to potentially join the program in the future, the sample for this study did not extend to those households on the waiting lists. Instead, the researchers focused their analysis on the experiences and perspectives of households who were currently benefiting from the UPSNP interventions. As such, it is recommended that future studies aim to capture the perspectives of the entire target population, including those who have not yet enrolled in the program.