Skip to main content

Profiles of child labour: prevalence, activities, contexts, and children’s harm extent in Raya-Kobo and Angot Districts, Ethiopia


Using a mixed research design, the overall objective of this study is to investigate the profiles of child labour in Ethiopia’s districts of Raya-Kobo and Angot. The study’s specific objectives are to examine types of activities and prevalence of child labour, to identify the children’s working contexts, and to evaluate the extent of harm children face in workplaces. The study hypothesizes the existence of significant variations in the profiles of child labour (activity types, prevalence rates, settings, and level of harm to children) as a result of several circumstances. While survey methodology was used to collect primary data from 431 randomly selected respondents, desk review and document analysis were also used to gather secondary data. The sampling procedures used include multi-stage purposive and systematic random sampling. Descriptive and inferential (Ordered Probit Regression model) methods of data analysis are utilized to analyze the data. The results of the descriptive analysis demonstrate a high incidence and magnitude of child labour, as well as children’s early entry to work. The regression analysis reveals children’s exposure to high levels of harm at work, which is directly exacerbated by household age, income diversification, land fragmentation, livestock number, shocks, and the predicted value of remittances. Contrarily, access to rural transportation, household heads’ education level, cart ownership, and family size reduce children’s harm extent while working. The study highlights the need to design strategies for age-appropriate work to protect children from hazardous and high-intensity activities, undertake community awareness campaigns about the impact of child labour on children, and strengthen local stakeholders’ capacity. Additionally, connecting rural areas and farmers’ croplands to towns or major routes via repaired or new roads would be beneficial.


Despite scientific and technological progress, rural communities continue to face multiple threats to their way of life, including the COVID pandemic and climate change. These challenges exacerbate poverty, social exclusion, and food insecurity, leading rural parents to resort to sending their children to work to make ends meet (IFAD, 2022). Consequently, child labour rates in rural areas are alarming. The agricultural sector alone accounts for about 70% of all child labour globally, with 73% of children working in agriculture aged between 5 and 11 years. Additionally, 72% of child labourers in agriculture are involved in small-scale family farming, and nearly 50% of these children face hazardous conditions that jeopardize their safety, health, and moral development (ILO & UNICEF, 2022). The onset of the twenty-first century witnessed the warmest period in climate history (World Meteorological Organization, 2012), with numerous natural disasters occurring in Ethiopia, particularly in North Wollo, including droughts, diseases, and floods (Hyder Baloch & Behrman, 2016). Thus, natural disasters have tapered rural people in the least-developed countries to spend, consume, and invest (Morton, 2007). As a result, rural communities in the least-developed countries have been compelled to expend limited resources on their immediate needs, leaving little room for investment (Mengstu, 2017).

Ethiopia, one of the oldest countries in Africa, has a long history of poverty. Due to the unpredictability of their sources of subsistence, many rural households in Ethiopia rely on their children for labour, significantly reducing their chances of attending school (UNICEF & UNESCO, 2015). Furthermore, rural girls in Ethiopia face additional burdens as they are often tasked with household chores, which limit their opportunities for economic and educational progression, compounded by the country’s low ranking (117th) in the 2018 global gender index (UNICEF, 2020). The aforementioned source also highlights that ethnic conflict, political violence, and instability in Ethiopia in 2019 led to evictions, disruptions to business operations, and restricted access to social services, which resulted in children transitioning from school attendance to work (UNICEF, 2020).

The involvement of rural children in household activities from a young age has been a common and integral part of their childhood and socialization processes, as many household heads consider children as a form of old-age security (Emerson & Knabb, 2006). While school and work often coexist for these children, many struggle to balance multiple demands on their time (Crivello & Pankhurst, 2015). Economists argue that rural parents often prioritize their children working to supplement their current income, despite the negative impacts on their children’s prospects (CSA et al., 2020). To substantiate this proposition, information generated on the status of early childhood education in rural parts of Ethiopia reveals the existence of poor educational coverage. The data regarding early childhood education indicate that millions of Ethiopian young children are not reaching their full potential due, among other things, to a lack of early stimulation, learning, nurturing care, and exposure to stress, all of which adversely affect their development (AACAMO, 2022). On top of this, there are no specific guidelines in place to ensure that the early childhood educational curriculum and teaching materials facilitate the inclusion of children with disabilities and other vulnerable groups. For example, schools lack the teaching–learning resources necessary to fulfil the needs of all pupils, and the curriculum is inflexible to permit local modifications (Mergia, 2020). One of the reasons for this could be the low level of the government’s expenditure on education. Like other governments in sub-Saharan Africa, budget allocation to early childhood development (ECD) services is as little as 0.01% of Ethiopia’s gross national product (WHO, 2020).

Despite all the information provided above regarding the difficulties, children face daily as result of child labour in rural areas, the literature lacks theoretical knowledge about child labour profiles with particular socio-economic, cultural, and settlement characteristics. What is also lacking is knowledge about the full extent and severity of injury that children experience in workplaces. This global lack of knowledge is more prevalent in the agricultural sector, where a larger share of child labourers is found (ILO, 2017a, 2017b). For example, there exist limited data on profiles of child labour in rural areas, as most national surveys overlook children’s domestic labour and fail to account for their burden (FAO, 2020). When examining the empirical data on child labour, we find that earlier research frequently lacked thorough information on the profiles of child labour activities of child labour, how common it is, the types of work environments in which it occurs, and the degree of harm that children endure during labour. This is mainly because the majority of prior research on child labour has focused on urban areas (Afriyie, et al., 2019; Ahad, et al., 2021; De Carvalho, 2012; Thévenon & Edmonds, 2019), non-financial factors that cause or contribute to child labour (Mbebi, 2018; Thévenon & Edmonds, 2019; Wondimu, 2022), the nexus between household preferences and child labour (Basu & Dimova, 2024), and the effects of child labour on children’s education and health (Kaur & Byard, 2021; Roggero, et al., 2007). To the best of the researchers’ knowledge, there is also a scarcity of previous studies to show the extent of children’s harm in rural areas of developing nations, particularly in Ethiopia, using the ordered probit model, a regression model exclusively applicable to show the different levels of injury children suffer in workplaces. As far as we know, very few prior studies on children labour have been done using ordered probit model, a useful regression model for illustrating the various degrees of children’s workplace injuries in rural areas of developing countries including the study areas. This further suggests a methodological deficiency in the world of research examining extent of child labour.

This foregoing paragraph highlights existing theoretical, empirical, and methodological gaps in knowledge on the profiles of child labour activities and prevalence of child labour, and children’s work contexts and extent of harms in their workplaces. Therefore, the main purpose of this locally disaggregated analysis is to fill such theoretical and empirical gaps. Besides, this study aims to contribute methodologically, going beyond analyzing the consequences of child labour to investigate the degree of harm experienced by children engaged in child labour using Ordered Probit Regression.

The general objective of this research was to examine the profiles of child labour in the research areas. The specific objectives of the study were;

  • To identify key activities, and describe the prevalence of child labour,

  • To analyze the working contexts in which children undertake child labour, and

  • To evaluate the extent of harm children endure due to child labour in the study areas.

Against the backdrop of the problems mentioned above, the following questions were set for investigation in the study areas:

  • How prevalent is child labour in rural areas, particularly in the agricultural sector?

  • What are the specific activities in which children are engaged while working in rural areas?

  • What are the working contexts in which children are involved in child labour in rural areas?

  • What are the extent and types of harm experienced by children engaged in child labour in rural areas?

This study on the profiles of child labour in Ethiopia’s rural areas of Raya-Kobo and Angot areas is essential for several reasons. First of all, it challenges a big social issue that has an impact on children’s rights and well-being, especially in areas with low economic status. Second, designing suitable interventions and policy approaches to fight child labour requires a complete understanding of the prevalence, activities, circumstances, and level of harm connected with child labour. The study can also help community-based programs and advocacy efforts that support child welfare and socio-economic development by revealing the particular situations and causes that contribute to child labour in the study areas. As a whole, the study would have a potential contribution to the pursuit of more ambitious sustainable development objectives, such as the eradication of child labour.

The conceptual framework for analyzing profiles of child labour in the Ethiopian districts of Raya-Kobo and Angot encompasses several key components. First, it incorporates prevalence rates of child labour, examining its incidence and distribution across different demographic and socio-economic groups. Additionally, the study investigates the types of work performed by children, distinguishing hazardous tasks from non-hazardous ones, and identifying the sectors employing children the most. The study also integrates contextual factors, such as land size, economic shocks, household demographics, access to transportation, income sources, and remittances, are also considered. Furthermore, the framework assesses the extent of harm inflicted on children, encompassing physical, mental, and developmental effects. By thoroughly examining these elements, the framework aims to provide crucial insights for developing targeted interventions to effectively reduce child labour in the specified study locations.

Materials and methods

Research perspective, design, and sampling procedures

This study used a mixed-methods approach to investigate the characteristics of child labour in the districts of Raya-Kobo and Angot in Ethiopia. To find out how common child labour is and what demographic and socio-economic characteristics are linked to it, quantitative surveys were first carried out. Then, focus groups and qualitative interviews were used to look into the complex situations surrounding child labour, including family dynamics, cultural standards, and financial limitations. The study also included quantitative evaluations of the kinds of jobs performed by young workers and the severity of their injuries, including physical, psychological, and developmental elements. This research endeavors to furnish a thorough comprehension of the characteristics of child labour in the designated districts through the integration of quantitative and qualitative methodologies. This would enable the formulation of focused interventions and policy suggestions aimed at reducing the incidence of child labour and lessening its effects on children. Pragmatism was considered as the research paradigm guiding this work, as it allows the collection and analysis of data on complex issues such as child labour that require coordination among relevant actors (Creswel, 2009). Both primary and secondary data were used in this study. The data collection instruments employed include structured questionnaires, qualitative interviews, and checklists of focus group discussion. Finally, trained data collectors and researchers obtained the data from 431 targeted households. As a social science study, no ethical standard permit was required. Nevertheless, this study followed stern ethical principles to respect the privacy of all participants. To do so, participants’ informed consent was gained for this study prior to data collection.

Multistage purposive and random sampling techniques were employed to select study areas and samples. In the first stage, North Wello zone, which has four livelihood zones, was purposively selected, because it is the most drought-prone zone of the Amhara region (Ege & Adal, 2002), and it is relatively accessible to the researchers. In the second stage, two livelihood zones, North Wello East Plain Livelihood Zone (NWEPL) and North Wollo Highland Belg Livelihood Zone (NWHBL), were purposively selected based on their variations in crop production practices, cropping seasons, types of crops and livestock produced, production seasons, and agroclimatic conditions. In the third stage, Raya-Kobo and Angot Districts were randomly selected from NWEPL and NWHBL, respectively, using stratified random sampling. In the final stage, sample households were selected using systematic proportionate random sampling from all kebeles, which were used as sampling frames. To ensure a representative sample, both male- and female-headed households were included in the study. The sample size was determined using Taro Yamane’s formula (1967), which is suitable for large populations with easily recognizable characteristics


where n = the desired sample size, N = total number of population, and e = the level of precision or the quality of being accurate which is equal to 0.05 (Yamane, 1973). About 400 samples and 20% contingency for non-response (all being 480) samples were down from 80,157 rural population of both districts using the systematic proportionate sampling approach (see Fig. 1).

Fig. 1
figure 1

Geographic location of study areas. Source: North Wollo Office of Agriculture, 2021

Methods of data analysis

Descriptive and econometric analysis techniques were utilized to analyze the collected data. Additionally, a bivariate analysis method was employed through cross-tabulations to examine the association between potential explanatory variables and the dependent variable, which is the severity of child labour (Creswel, 2009). To account for the qualitative nature of the dependent variable in this study, which measures the severity of harm children face while working, we used the Ordered Probit model as the primary estimation technique. The dependent variable has five ordered categories (1 “very low”, 2 “low”, 3 “moderate”, 4 “high”, and 5 “very high”) that reflect a natural ranking of the underlying continuous values. The Ordered Probit model is suitable for such ordinal variables, as it is based on the multinomial distribution and can accommodate ranking categories (Puurbalanta & Adebanji, 2016).

The fundamental concept underlying ordered probit regression is that the researcher’s observations of ordinal responses represent latent continuous metrics. The latent continuous variable, y*, is a linear combination of some predictors, x_i, plus a disturbance term that follows a standard Normal distribution (Jackman, 2000). Thus, we estimated the regression coefficients of the observed dependent variable (the degree of harm children face while working) using the Ordered Probit model, where y* = α + ∑β_iX_i + . Here, y* represents the exact but unobserved dependent variable, X_i denotes a set of explanatory variables, β_i represents unknown parameters (regression coefficients), and _i is independently and identically distributed disturbances with a probability density function (pdf) denoted as f(, θ) and distributional parameters θ. We were only able to observe the categories of responses on working conditions, y, instead of y* (as shown in Table 1).

Table 1 Summary of the categories of responses on working conditions y instead of observing y*

To estimate the ordered probit model, the “threshold” parameters (∂’s) along with the coefficients (α and βi’s) must be estimated. We assume that the disturbances, ϵi, are normally distributed with a mean of zero and a standard deviation of one. Maximum-likelihood estimation is used to estimate the model, where the likelihood function requires that the disturbances are distributed as standard normal. The probability of observing a dependent variable value of 1, given the explanatory variables, is calculated as the probability that the latent variable y* is less than or equal to the first threshold value (∂1). This probability can be written as an integral of the standard normal density function from negative infinity to the difference between the first threshold and the linear combination of the explanatory variables (α1 + ∑X1β1). The probability of observing a dependent variable value of 2 is calculated as the probability that the latent variable y* is greater than the first threshold and less than or equal to the second threshold (∂2). This probability can be written as an integral of the standard normal density function between the first and second threshold values, with the linear combination of the explanatory variables (α2 + ∑X2β2) subtracted from the second threshold. Similar expressions can be used to calculate the probabilities of observing the other dependent variable values.

Identification is achieved by setting the first threshold (∂0) equal to zero and specifying that the disturbances have a mean of zero and a standard deviation of one. The likelihood function is the product of the probabilities for each observation. The maximum-likelihood estimates for the coefficients and thresholds are obtained by maximizing this function (Becker & Kennedy, 1992).

Specification of study variables

Dependent variable

The dependent variable for this study is children’s harm extent at workplaces (extent_harm). It is an ordinal variable with five outcomes: 1 “very low”, 2 “low”, 3 “moderate,” 4 “high”, and 5 “very high”. Both continuous and categorical variables were chosen as the study’s explanatory factors.

Independent/explanatory variables

Independent variables thought to have an impact on the dependent variable were included in the study. These explanatory variables include both continuous and categorical ones. Table 2 presents an inventory of independent and dependent factors.

Table 2 Summary of independent variables

To ensure the validity of the quantitative results, data were triangulated. To triangulate the quantitative data, we employed qualitative data collection methods like focus group discussions, field observations, and key informant interviews.

Results and discussion

Descriptive analysis of non-model variables

Child labour: prevalence, magnitude, and work starting age of children

Table 3 presents the cross-tabulation results for the involvement of children in household labour and their age of commencement. The results of the study reveal that all households in the sample reported the participation of their children in household labour activities (Table 3). In the study areas, unpaid agricultural work by children is prevalent. Among the surveyed children (n = 431), the majority (326) commence assisting their parents between ages 4 and 6. Additionally, the study found that the average number of hours children work is 49.27 per week (SD = 15.37). This suggests that child labour is a significant issue in the area under study. Compared to the findings of Pankhurst and his colleagues on the weekly working hours of children, who found that youngsters work 33 h a week on average across Ethiopia (Pankhurst et al., 2015), the result of this study implies that there are increasingly more weekly working hours in rural of Angot and Raya Kob districts.

Table 3 Children’s work participation and work starting ages.

The fact that child labour is most prevalent, high in intensity, and kids start labouring at their youngest ages illustrates that children in these research areas are not attending school in their early ages. Although there may be higher school enrollment rates among the children in the rural parts of Raya-Kobo and Angot Districts, this does not guarantee that the children would have performed better in the classroom in terms of proper grade advancements, maintaining good attendance, and other educational outcomes.

There are various reasons for this according to the empirical findings of different studies. Despite the paucity of research on inclusive early education policy in Ethiopia, findings of some studies indicate that socio-cultural, economic, and pragmatic obstacles seem to prevent policy frameworks from being put into reality (Mergia, 2020). Besides, Ethiopia’s nominal allocation to the national education budget rose from ETB 95.7 billion in 2017/18 to ETB 162.2 billion in 2021/22, but the allotted fund’s real worth showed a 20% reduction (UNICEF, 2020). UNICEF also reported that conflicts in Tigray, Amhara, and Afar caused at least 3.2 million children to drop schooling in 2020–2021. In addition, the same report indicated that the conflict destroyed 1090 and 3220 schools partially and entirely, respectively.

Working conditions of children in Angot and Raya-Kobo Districts

According to Fig. 2, the most common working conditions for children include harsh physical environments, long hours of work, exposure to domestic and wild animals, and extreme weather conditions. Participants in focus groups and key informants assert that these working conditions are detrimental to the physical and mental health of children. In our field observations, we also noticed that children working under dangerous conditions, such as long hours, heavy loads, exposure to extreme weather, climbing trees, and working in muddy terrain, all lead to poor education performances, numerous health problems, including physical injuries, infectious diseases, and emotional and psychological stresses. The findings of this study are corroborated by earlier studies that demonstrate how child labour negatively affects children’s health and welfare (Eurofound, 2022; Ibrahim et al., 2018; John & Murugan, 2021).

Fig. 2
figure 2

Source: own survey data, 2020/21

Working conditions of rural children in Raya-Kobo and Angot districts

Children’s extent of harm at workplaces

The results in Table 4 demonstrate that rural children engaging in various activities experience different levels and forms of harm, with most working children often experiencing high levels of harm while participating in these activities. The data indicate that respondents who classified their children’s activities as “high and above” comprise a majority of the scores (2406 out of 3448), while those classified as “low and below” comprise a smaller proportion of the scores (448 out of 3448). These findings suggest that children in rural areas are more likely to suffer from illness or injuries resulting from work, regardless of gender. These results are consistent with previous research by ILO on child labour in rural areas (ILO, 2017b).

Table 4 Harm extent rural children face at work in Angot and Raya-Kobo districts.

Descriptive analysis of model variables

Continuous independent variables of the model

Most of the independent variables that have a statistically significant influence on the dependent variable are measured on a scale. The social shocks index is one of the variables that (predicted to directly influence children’s harm extent at workplaces) affect children’s harm levels. Table 5 displays the lowest, mean, and maximum indices of social shocks, which are, respectively, 0, 0.075, 0, and 0.2. The educational status of household (expected to inversely affect children’s workplace injury) heads varies widely, with some having no schooling years and others having 10, but the majority have spent approximately 1.7 years in school. Livestock ownership (measured by the number and kinds of households’ livestock and assumed to directly affect children’s workplace harms) was also considered in this study, with an average of about 14 animals owned by household heads. However, some households have as many as 45 livestock, while others have none. Data on land fragmentation (the number of households’ farmland locations and anticipated to put straight influences on children’s harm extent) indicate that while some households have no crop land, others have crop lands in six locations. In terms of income sources (kinds of income earning activities and assumed to have opposite impacts on children’s harms), some households have up to six sources, while others have only one. The majority of households have approximately three different income streams, as shown in Table 5.

Table 5 Summary of continuous explanatory variables of children’s working conditions.

Dummy independent variables and the degree of children’s harm in working

We conducted correlation analysis to determine the linear relationship between the dependent and independent variables. Two independent variables (households’ access to rural transportation and cart ownership) were investigated. Both are dummy variables with a value of 1 indicating access and 0 otherwise. Results indicated a strong association between households’ possession of carts and children’s extent of harm at workplaces. From 431 observations, 26, 48, and 171 households (most of them cart owning) reported very low, low, and moderate harm extent, respectively. Conversely, 145 and 41 households (most of them non-cart owning) reported high and very high levels of harm extent in their children’s working conditions. We reject the null hypothesis (no variation exists on children’s workplace level of harms due to cart ownership), as the Chi-square value (χ2) of 52.03 and the accompanying P value of 0.000 demonstrate variation households’ ownership of carts and the children’s harm extent at workplaces.

Similarly, the Chi-square test on the association between rural transportation and children’s degree of workplace injury indicates the existence of strong correlation between the two variables. Therefore, the null hypothesis (rural transportation brings no variation on children’s workplace harm extents) is rejected, since as the Chi-square value (χ2) of 102.94 is significant, as the P value is 0.000, demonstrating a high level of variation.

Ordered probit model on the determinants of children’s harm in working

In this study, an ordered probit regression model was used to examine the predictive power of 13 independent variables on the outcome variable-children’s harm extent in working. The dependent variable was treated as an ordinal variable with five categories ranging from "very low" to "very high." The model was based on 425 observations, with nine missing. The results showed that the model as a whole was a good fit for the data, as indicated by a likelihood ratio Chi-square value of 406.939 and a P value of 0.000. Additionally, ten of the estimated independent variables were found to have a statistically significant effect on the extent of harm children experienced while working at a 1% significance level (see Table 6).

Table 6 Summary of dummy explanatory variables of children’s working conditions.

Predictive margins, and average marginal effects of explanatory variables on the children’s harm extent at workplaces

The results from the probit model indicate that household head age has a significant impact on the extent of harm experienced by children while working. Table 8 of Appendix A indicates children’s vulnerability to high/very high workplace harms tends to increase by 0.3% with each additional year of household head age, while their vulnerability to very low, low, or moderate levels of harm tends to decrease by about 0.2%, citrus paribus. The reverse happens with decreases of household heads’ age.

Table 7 Ordered probit regression on the determinants of children’s working conditions.

Numerous theories could explain these results. Social conventions are one potential explanation. Children’s involvement in household tasks may be seen by elder household heads as important for socialization and discipline. As a result, many elder parents in rural areas might urge their kids to pursue career ambitions, increasing the danger to the kids. Another factor is the migration of children from the study sites to the Middle East, which could drive the remaining kids to engage in potentially dangerous domestic chores at a young age. The third factor would be problems with capacity and health that come with getting older. A key informant from the Angot District noted that as people age, their physical capacities steadily deteriorate, which may increase the involvement of children in household chores. Additionally, focus groups in the Aradum Kebele of Raya-Kobo District highlighted how older households gradually lose their strength, which would increase the involvement of children in risky employment. The findings of this study are consistent with the previous research by Sanusi and Akkinnniram (2013).

Figure 3 displays the probability of children facing high or very high levels of harm while working as household heads’ age increases from 27 to 85. The Figure illustrates a steady increase in the probability of harm with increasing age.

Fig. 3
figure 3

Source: own survey Data, March 2020/21

Predictive margins and average marginal effects of household age.

The ordered probit model in Table 7 indicates that household income sources directly affect the extent of children’s harm while working. Other variables being constant, the average marginal effects in Table 8 of Appendix A reveal that a unit increase in kinds of income sources would result in a 1.2% decrease of children’s participation in household activities with very low or low harm risks, and a 1.8% decrease in moderately harming activities. However, the probability of children exposure to high or very high risking activities would rise by 2.4% and 1.9%, respectively, for every unit increase in kinds household revenue sources. The predictive margin of household income on children’s working conditions displayed in Fig. 4 demonstrates that as household income sources increase from one to five, the likelihood of children engagement in very low, low, or moderate harm risking activities declines, while their probability of participating in high and very high harm risking activities increases, other variables being equal. Focus group discussants from Raya-Kobo District indicate that when rural households involve in income diversification activities, they often rely on off-farm and non-farm activities (trading, charcoal/wood selling). This withdraws adults from agricultural tasks, leaving children to engage in different household activities that are hazardous to their health.

Fig. 4
figure 4

Source: Own survey Data, March 2020/21

Predictive margins and average marginal effects of income sources.

Other factors held constant, and Table 8 of Appendix A demonstrates that owning a cart has a large and reverse impact on children’s exposure propensity of job injuries. Children’s inclination to engage in harmful activities would reduce by 5.1% and 4.8%, respectively, as a result of households having carts. On the other hand, having a cart increased children’s likelihood of engaging in activities with very low, low, and moderate levels of harm risks, respectively, by 3.4 percent, 3 percent, and 3.6 percent. Key informants of both districts verified that possession of carts has the potential to have a sizable impact on children’s possibilities of working in safe or hazardous situations. Carts are essential tools for rural households to carry out various agricultural tasks-transporting crops and farm tools, fetching water, and collecting crop leftovers, fodder, and farm-fence items. Cart ownership enables households to transport agricultural products from farms to homes and markets, which can reduce the need for children to engage in hazardous work. The results of this study are corroborated by Chowa and Masa’s investigation, which concluded that asset ownership enhances children’s health, propels educational attainment, and reduces the prevalence of child labour (Chowa et al., 2010) (see Fig. 5).

Fig. 5
figure 5

Source: Own survey data, March 2020/21

Predictive margins and average marginal effects of cart ownership on children’s working conditions (WC).

The index of social shocks (constructed from divorce, illness or death of children’s parents, and illness/death of children) is calculated by averaging these dummy variables. The ordered probit regression results indicate that social shocks have a statistically significant and direct effect on the level of harm that children experience while working. The average marginal effects of social shocks on children’s working conditions reveal that an increase of one unit in the index of social shocks results in a 45.1%, 87.6%, and 70.6% decrease in the likelihood of children engaging in household activities with very low, low, or moderate levels of harm, respectively. Conversely, there is a 66.2% and 62.7% increased likelihood of children being involved in work with high or very high levels of harm, respectively (Table 8, Appendix A). Figure 6 illustrates that children whose parents frequently experience social shocks are more likely to face worse working conditions than children whose parents are not as vulnerable to social shocks. On the other hand, Table 8 of Appendix A shows that children of parents who experience little or no social shocks have a higher likelihood of carrying out domestic duties in superior, good, or moderate working conditions. Key informants and focus group participants claim that older children whose parents are unwell or pass away are obliged to take on the burden of providing for the rest of the family. Regardless of their age and holding capacity, they would be exposed to a variety of household activities. Elder children are responsible for caring for their younger siblings, managing the farm, herding the animals, taking part in social activities, and managing all other forms of tasks. All of these tasks, along with others, will surely increase the likelihood that children may be exposed to high levels of harm at work. The results of this study agree with the findings of Mendolia and his colleagues (2019).

Fig. 6
figure 6

Source: Own survey data, 2020/21

Predictive margins and average marginal effects of social shocks on children’s working conditions (WC).

The ordered probit model indicates that rural transportation has a statistically significant and reverse impact on children’s working conditions. The marginal effects of the rural transportation dummy indicate that households’ access to rural transportation would upsurge their children’s likelihood involvement in activities with very low, low, or moderate levels of harm would increase by about 6.7%, 6.8%, and 10%, respectively, and would reduce their chances of engaging in activities with high and very high levels of harm by approximately 13% and 10.5%, all else being equal.

According to focus group discussants from both districts, rural transportation improves the availability of adult labour and facilitates transporting farm inputs to and from farmlands, increases people’s awareness and allows for other non-farm revenue-earning activities for adults, eliminates the home-school distance barrier for children’s education, and thus reduces households’ dependence on child labour in comparison to households located far away from such road networks., as illustrated in Fig. 7. The results of this study are reinforced by findings of Adeoye et al.(2017).

Fig. 7
figure 7

Source: Own survey data, 2020/21

Predictive margins and average marginal effects of rural transportation on children’s harm levels.

The regression analysis indicates that kind and number of livestock have a direct impact on the level of children’s harm, other things held constant. Specifically, children from households with more livestock are at a higher risk of being exposed to harmful working conditions compared to children from households with fewer or no livestock. The average marginal effects of livestock numbers on children’s injury show that an increase in kind and number of livestock by one unit leads to a 0.1% and 0.2% decrease in children’s participation in very low/low, and moderate levels of children’s workplace harm, respectively. Conversely, the same change brings a 0.3% and 0.2% increase in children’s participation in household activities with high and very high levels of harm, respectively (see Table 8, Appendix A and Fig. 8).

Fig. 8
figure 8

Source: Own survey data, 2020/21

Predictive margins and average marginal effects of livestock number on children’s harm extents.

The study’s findings reflect real practices of child labour in the study areas, according to focus group discussants. They underlined that children often engage in labour-intensive livestock-related activities, such as herding, gathering and feeding animals, and cleaning animal houses. They also briefed that these activities expose children to hazardous working conditions, due to wild animal attacks, accidents, and exposure to harmful substances. The results of this study are consistent with the findings of a recent survey conducted by Shumetie and Mamo (2019).

The ordered probit regression highlights that family size inversely influences children’s tendency of exposure to workplace harms and is statistically significant at the 1%. Therefore, children from larger families have higher likelihood of participating in safe activities with very low, low, and moderate levels of harm than children from smaller families. The average marginal effects of family size on children’s degree of harm at work indicate that an increase in family size by one unit would, respectively, result in a 0.9% and 1.4% increase in the probability of children’s engagement in very low/low and moderately harmful activities, other things being equal. On the other hand, the same change in family size would decrease the probability of children’s engagement in high and very-high harmful works by 1.8 and 1.4 percentage points, respectively (Fig. 9).

Fig. 9
figure 9

Source: Own survey data, 2020/21

Predictive margins and average marginal effects of family size on children’s harm extents.

Two motives exist for this variation. First, in most of our study sites, child labour outside the family is culturally rejected and stigmatized, leading households with abundant children to be reluctant to allow their children to work for other households in the same community. Second, daily workers prefer living in small towns and working on daily contracts rather than long-term (yearly) contracts. This observation contests with empirical findings of Fetuga, et al. (2005), and Togunde Richardson (2006).

The positive coefficient of land fragmentation (0.456) in this study implies a direct impact on the degree of children’s workplace harm (Table 7). The average marginal effect of land fragmentation on children’s harm level indicates that, other things being static, a unit increase in the number of crop land locations would lead to a 4.6% and 3.7% increase in the likelihood of children’s exposure to high and very high levels of harm in household activities, respectively. Additionally, it would result in a 2.3%, 2.5%, and 3.5% decrease in children’s participation in very low, low, and moderately harmful activities, respectively (Table 8, Appendix A). Key informants of both districts explained that farmers use their land for various purposes, such as growing food and fodder, and constructing dwellings, which require a labour-intensive smallholder production system involving both human labour and animals. Consequently, as the number of cropland locations increases, more human labour is needed, which can lure children away from schooling (Fig. 10). This finding is substantiated by the previous of Markussen et al (2016).

Fig. 10
figure 10

Source: Own survey data, 2020/21

Predictive margins and average marginal effects of land fragmentation on children’s harm extents.

The negative coefficient (− 0.071) for household head education indicates that it has a contrasting effect on children’s level of harm while working (Table 7). The marginal effect shows that other factors held constant, and a 1-year increase in the schooling of family heads decreases children’s likelihood of participating in highly and very highly harming activities by approximately 0.9 and 0.7%, respectively. In contrast, there is an increase in the likelihood of children participating in activities with very low or low, and moderately harmful levels of harm by 0.5 and 0.7%, respectively (Fig. 11).

Fig. 11
figure 11

Source: Own survey results, 2020/21

Predictive margins and average marginal effects of household head education levels.

The qualitative information from key informants of both districts indicates that when household heads have more years of education, they tend to understand the negative consequences of child labour and recognize the importance of education for their children’s future. As a result, they are more likely to encourage their children to attend school instead of sending to harmful activities. This finding is supported by recent empirical findings of Abebe and Fikre (2021).

Remittances’ predictive value has been shown to have a direct impact on the likelihood of children’s workplace injuries. The likelihood of children being exposed to a serious workplace injury would increase by 0.449 for every unit increase in anticipated remittance amounts in Ethiopian Birr (Table 7). Furthermore, the predictive and marginal effects of projected remittance values on children’s degree of workplace injury demonstrate that an increase in remittance income would exacerbate the degree of harm to children, citrus paribus (Fig. 12). Both key informant interviewees and focus group participants from Raya-Kobo underscore two key reasons for such impact. First, remittance directly influences children’s exposure to work-related harms by withdrawing income to other investment purposes instead of investing in children’s education. Second, remittances also indirectly worsen children’s harm extent, since they withdraw the labour force from the study areas via migration to the Arab World. A study by Acosta supported the findings of this study, as he found that remittances induce increased in unpaid child labour (Acosta, 2011).

Fig. 12
figure 12

Source: Own survey results, 2020/21

Predictive margins and average marginal effects of remittance’s predictive values.

Conclusions and recommendations


This study has provided valuable insights into child labour profiles in Raya-Kobo and Angot Districts. The descriptive results highlighted the alarming prevalence of child labour, with children working an average of 49.9 h per week. It was also observed that children were involved in household tasks from an early age and that gender-based roles existed for children. This study also revealed that important issues including conflict, inflation, and different socio-cultural elements significantly exacerbated child labour and caused millions of children to drop out of early education. The study also employed an ordered probit regression to analyze the factors influencing the degree of harm experienced by children while working. Thirteen explanatory factors were examined, including household age, income diversification, land fragmentation, livestock density, family size, household head education level, cart ownership, rural transportation, remittances instrumented, and social shocks.

The statistical analysis revealed that all of these variables, except for wealth index, economic shock, and sex of home heads, had a significant influence on the level of injury children experienced at their workplaces. Factors, such as household age, income diversification, land fragmentation, livestock number, and social shocks, were found to directly exacerbate children’s levels of harm at work. On the other hand, family size, cart ownership, household head education, and access to rural transportation were identified as factors that can improve children’s working conditions by reducing the extent of damage they are exposed to.

These findings highlight the multifaceted nature of child labour and the various factors that contribute to the vulnerability of children in these districts. The study provides important insights for policy-makers and stakeholders in devising interventions and strategies to address child labour and improve the working conditions of children. However, it is essential to note that this study has certain limitations. First, the research only focused on two districts, which may limit generalizability to other regions. Additionally, the study relied on self-reported data, which may introduce bias or inaccuracies. Future research could overcome these limitations by expanding the scope to include a more diverse sample and utilizing more objective measures of child labour and its impact.

Overall, this study sheds light on the problematic issue of child labour in Raya-Kobo and Angot Districts and provides valuable insights into the factors influencing the degree of harm experienced by children. It is hoped that these findings will inform effective interventions and policies to combat child labour and promote the well-being of children in these communities.


Based on the study’s findings, it is recommended that various stakeholders work together to develop and implement policies and strategies that prioritize the well-being of children, including the following:

  • Government actors at all levels should design policies and strategies that protect young children from hazardous household activities by taking into account their age and vulnerability.

  • Educational institutions (mainly Universities) should involve in raising awareness among communities about the negative impacts of child labour, and the importance of children’s early education through various means such as farmers’ field schools and public assemblies, to eliminate harmful child labour practices.

  • NGOs should engage in providing training and capacity building for local agricultural stakeholders, such as farmers, extension agents, religious leaders, and politicians, with a focus on child labour, its detrimental effects, and the significance of early education for children.

  • Transport service provision organizations at all levels need to improve rural transportation infrastructure by repairing damaged roads, building new ones that connect rural areas to towns or main roads and farmers’ croplands to at least feeder roads.

  • Financial institutions should encourage rural households to invest in carts improved vehicles to facilitate the delivery of farm inputs and other goods providing credit. They should also support educational institutions to enhance their capacity for universal early education for all children in their school age.

Upcoming studies should investigate the obstacles to enhancing inclusive early education implementation systems and intervention strategies in Ethiopia and beyond.

Availability of data and materials

Data will be available upon formal request of the corresponding author.


  • AACAMO. (2022). Children : The Future Hope of Addis Ababa Early Childhood Development ( ECD ) Program Children : The Future Hope of Addis Ababa Early Childhood Development Program. (February)

  • Abebe, G., & Fikre, S. (2021). Individual, household, and community level factors of child labour in rural Ethiopia Individual, household, and community level factors of child labour in rural Ethiopia. Cogent Social Sciences.

    Article  Google Scholar 

  • Acosta, P. (2011). School attendance, child labour, and remittances from international migration in el salvador. Journal of Development Studies, 47(6), 913–936.

    Article  Google Scholar 

  • Adeoye, S. O., Agbonlahor, M. U., Ashaolu, O. F., & Ugalahi, U. B. (2017). Analysis of child labour diamenstions and causes in rural farm housheholds of Ogun State, Nigeria. African Journal of Food, Agriculture, Nutrition, and Development, 17(3), 12198–12213.

    Article  Google Scholar 

  • Afriyie, L. T., Saeed, B. I. I., & Alhassan, A. (2019). Determinants of child labour practices in Ghana. Journal of Public Health (germany), 27(2), 211–217.

    Article  Google Scholar 

  • Ahad, M. A., Chowdhury, M., Parry, Y. K., & Willis, E. (2021). Urban child labour in bangladesh: Determinants and its possible imacts on health and education. Social Sciences.

    Article  Google Scholar 

  • Basu, A., & Dimova, R. (2024). Household preferences and child labour in Rural Ethiopia. Journal of African Economies, 33(1), 20–45.

    Article  Google Scholar 

  • Becker, W. E., & Kennedy, P. E. (1992). A graphical exposition of the ordered probit. Econometric Theory, 8(1), 127–131.

    Article  Google Scholar 

  • Chowa, G., Ansong, D., & Masa, R. (2010). Assets and child well-being in developing countries: A research review. Children and Youth Services Review, 32(11), 1508–1519.

    Article  Google Scholar 

  • Creswel, J. W. (2009). Research design : Qualitative, quantitative, and mixed methods approaches (3rd ed.). SAGE Publications Inc.

    Google Scholar 

  • Crivello, G., & Pankhurst, A. (2015). Exploring children’s experiences of work in Ethiopia: A guide for child-focused research. University of Oxford.

    Google Scholar 

  • CSA, U. E. and C. (2020). Child Labour Analysis in Ethiopia 2020. 126.

  • De Carvalho Filho, I. E. (2012). Household income as a determinant of child labour and school enrollment in Brazil: Evidence from a social security reform. Economic Development and Cultural Change, 60(2), 399–435.

    Article  Google Scholar 

  • Emerson, P. M., & Knabb, S. D. (2006). Opportunity, inequality and the intergenerational transmission of child labour. Economica, 73(291), 413–434.

    Article  Google Scholar 

  • Eurofound. (2022). Working conditions in the time of COVID-19: Implications for the future, European Working Conditions Telephone Survey 2021 series.

  • FAO. (2020). FAO framework on ending child labour in agriculture. FAO Framework on Ending Child Labour in Agriculture.

    Article  Google Scholar 

  • Fetuga, B. M., Njokama, F. O., & Olowu, A. O. (2005). Prevalence, types and demographic features of child labour among school children in Nigeria. BMC International Health and Human Rights, 5, 1–7.

    Article  Google Scholar 

  • Hyder Baloch, A., & Behrman, J. R. (2016). Climatic shocks and child human capital: Evidence from Ethiopia. Journal of Asian and African Studies, 51(5), 594–618.

    Article  Google Scholar 

  • Ibrahim, A., Abdalla, S. M., Jafer, M., & Abdelgadir, J. (2018). Child labour and health : a systematic literature review of the impacts of child labour on child ‘ s health in low- and middle-income countries. 41(1), 18–26.

  • IFAD. (2022). The International Partnership for Cooperation on Child Labour in Agriculture (IPCCLA) Statement for the 5th Global Conference on the Elimination of Child Labour. Retrieved from

  • ILO. (2017a). Ending child labour by 2025: A review of policies and programmes. International Labour Organization., 46(5), 477–491.

    Article  Google Scholar 

  • ILO. (2017b). Global estimates of child labour: Results and trends 2012–2016. In Ilo.

    Google Scholar 

  • ILO and UNICEF. (2022). Methodology of the 2020 ILO-UNICEF Global Estimates of Child Labour. ILO and UNICEF.

    Google Scholar 

  • Jackman, S. (2000). Models for ordered outcomes, political science 200c. Researchgate.

    Google Scholar 

  • John, S. P., & Murugan, K. R. (2021). The cause and impact of child labour on a child’s short and long-term health. World Journal of Research and Review, 13(4), 7–13.

    Article  Google Scholar 

  • Kaur, N., & Byard, R. W. (2021). Prevalence and potential consequences of child labour in India and the possible impact of COVID-19—a contemporary overview. Medicine, Science and the Law, 61(3), 208–214.

    Article  Google Scholar 

  • Markussen, T., Tarp, F., Thiep, D. H., Do, N., & Tuan, A. (2016). Inter- and intra-farm land fragmentation in Viet Nam (No. 11; Vol. 11).

  • Mbebi, O. E. (2018). Determinants of child labour in Cameroon: The role of area of residence and gender. Revue D’economie Du Developpement.

    Article  Google Scholar 

  • Mendolia, S., Nguyen, N., & Yerokhin, O. (2019). The impact of parental illness on children’s schooling and labour force participation: Evidence from Vietnam. Review of Economics of the Household, 17(2), 469–492.

    Article  Google Scholar 

  • Mengstu, K. A. (2017). The effect of drought on child labour in Ethiopia (Vol. 88). UNiversity of Oslo.

    Google Scholar 

  • Mergia, A. T. (2020). The practice of inclusive education in Ethiopia. Programme and Meeting Document, 1–25.

  • Morton, J. F. (2007). The impact of climate change on smallholder and subsistence agriculture. Proceedings of the National Academy of Sciences of the United States of America, 104(50), 19680–19685.

    Article  Google Scholar 

  • Pankhurst, A., Crivello, G., Chuta, N., Tiumelissan, A., Valadez, L., Wilson, E., & Medhin, L. W. (2015). Child Work and Labour in Ethiopia: Literature Review and Policy Implications. Proceedings of the 10th Annual Conference of the Ethiopian Society of Sociologists, Social Workers and Anthropologists, (June 2013).

  • Puurbalanta, R., & Adebanji, A. O. (2016). Household poverty-risk analysis and prediction using Bayesian ordinal probit models. International Journal of Statistics and Applications, 6(6), 399–407.

    Article  Google Scholar 

  • Roggero, P., Mangiaterra, V., Bustreo, F., & Rosati, F. (2007). The health impact of child labour in developing countries: Evidence from cross-country data. American Journal of Public Health, 97(2), 271–275.

    Article  Google Scholar 

  • Sanusi, W. A., & Akinnniram, N. (2013). Effect of households poverty level on child labour participation among households in Nigeria. Developing Country Studies., 3(7), 97–102.

    Google Scholar 

  • Shumetie, A., & Mamo, K. (2019). Effect of cropland and livestock ownership on child labour in eastern Ethiopia: empirical examination of the Wealth Paradox. International Journal of Child Care and Education Policy.

    Article  Google Scholar 

  • Thévenon, O., & Edmonds, E. (2019). Child labour: Causes, consequences and policies to tackle it. OECD Social, Employment, and Migration Working Papers, (235), 4–82.

  • Togunde, D., & Richardson, S. (2006). Household size and composition as correlates of child labour in Urban Nigeria. Africa Development, 31(1), 50.

    Article  Google Scholar 

  • UNICEF. (2020). UNICEF Ethiopia Annual Report 2019. UNICEF.

    Google Scholar 

  • UNICEF, & UNESCO Institute for Statistics. (2015). Global Out-of-School Children Initiative. UNICEF, & UNESCO Institute for Statistics.

    Google Scholar 

  • WHO. (2020). World Health Organization annual report 2019 WHO Country Office Lebanon: Health for all. World Health Organization.

    Google Scholar 

  • Wondimu, H. (2022). Factors leading to child labour exploitation and the involvement of non-governmental organizations (NGOs) in Addis Abeba Ethiopia factors leading to child labour exploitation and the involvement of non-governmental. Cogent Social Sciences.

    Article  Google Scholar 

  • World Meteorological Organization. (2012). The Global climate 2001–2010, a decade of climate extremes, summary report. World Meteorological Organization.

    Google Scholar 

  • Yamane, T. (1973). Statistics: An introductory analysis. Haroper & Row, Evanston & Londol, and John Weatherhill, JNC.

    Google Scholar 

Download references


All stakeholders involved in providing necessary technical and administrative support are acknowledged.


This research received no external funding.

Author information

Authors and Affiliations



Mengistu Weldeyesus: “Conceptualization, methodology, software, formal analysis, investigation, data curation, writing original draft preparation, writing review and editing, and visualization. Bamlaku Alemu: Validation, software, supervision, and editing.

Corresponding author

Correspondence to Mengistu Abate Weldeyesus.

Ethics declarations

Competing interests

There are no competing interests as all authors have read published version of the manuscript and agreed upon.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix A

Appendix A

See Table 8

Table 8 Average marginal effects of explanatory variables

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Weldeyesus, M.A., Alemu, B.A. Profiles of child labour: prevalence, activities, contexts, and children’s harm extent in Raya-Kobo and Angot Districts, Ethiopia. ICEP 18, 4 (2024).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: