Participation in artisanal diamond mining and food security: A case study of Kasai Oriental in DR Congo

1.1 Artisanal mining overview

Artisanal and small-scale mining is emerging as a significant non-farm activity in developing countries. Essentially informal, characterized by low entry costs, high labor intensity, and a lack of mechanization, artisanal mining encompasses various individuals or small-group mining activities, ranging from precious stones like diamonds to industrial minerals such as tantalum, etc. [1].

The rapid growth of the population, coupled with seasonality and productivity decline in the farm sector, acute rural poverty, and poor development and integration of both secondary and tertiary sectors have made artisanal mining emerge as a crucial economic pole, particularly in sub-Saharan Africa. It serves as a vital livelihood for millions, especially those in remote rural areas [1]. By absorbing excess rural labor, it regulates rural migration, generates millions of direct jobs, and facilitates millions of other indirect jobs upstream and downstream of the mining supply chain, enabling miners and their families to meet their basic needs [2,3,4,5,6]. This catalytic role of the sector creates induced and inductive effects on the activities of local rural economies, affecting farm production, local consumption, small businesses, catering services, reinforcing then human, goods, and services mobility, etc. The presence of induced activities, in turn, becomes an attractive factor for new economic actors around mining sites, leading to a cumulative development effect through consumption [7,8,9].

Studies conducted in countries such as Angola, Sierra Leone, the Central African Republic, and Democratic Republic of the Congo (DRC) reinforce the idea that artisanal mining has become a structural pillar, allowing many rural households to connect to local and urban markets since it provides a significant number of jobs albeit precarious. In DRC, e.g., the sector employs a little over 700,000 direct artisanal miners, with over 1,200,000 dependents on these created jobs. It contributes significantly to the country’s gross domestic product [1,6,10].

Many rural residents participate in artisanal mining in developing countries in the search for minerals in rocky terrains, waterways, and both active and abandoned mine wells mainly during the dry season. The income generated from this activity has not only improved the living standards of participants but has also funded essential government services such as health, education, and development through taxes and other levies while supporting small businesses and household farm activities. If well-directed, then, diamond revenues can greatly benefit the economies of the regions where the stones are extracted such as in Botswana, Namibia, etc. [4,11,12,13].

However, artisanal mining raises serious socioeconomic and environmental concerns in developing countries. The massive influx of immigrants into mining zones often leads to intense land pressure, farmland expropriation, anarchic land use, and unsuitable soil conditions post-exploitation [14]. Job migration frequently results in shifts in labor allocation within rural households, with young men favoring mining over agriculture. This shift can decrease farm output, and coupled with the massive influx of miners, it increases food prices, challenging food security for rural households [15,16].

Recent studies in Ghana, Liberia, and Sierra Leone indicate that while artisanal mining provides additional income and plays an important role in rural communities, it frequently causes land degradation, water pollution, and a shift of labor from agriculture to mining. These effects create vulnerabilities in farming-dependent rural economies and potentially undermine sustainable rural development, highlighting the need for sustainable practices within the sector [17,18].

Environmental degradation is a critical issue of mining activities. Artisanal mining, including digging, washing, and sorting gravel, contributes to air and water pollution, rendering soil unsuitable for farming. Deforestation and community displacement further impact local agriculture and biodiversity. These factors significantly limit subsistence farming and agricultural productivity among local communities [19]. The negative effects are particularly concerning as smallholder farms are essential for building food security for rural households in many African countries. Research on abandoned mines in the DRC raises environmental challenges such as soil and water contamination, deforestation, and increased local conflicts, emphasizing the need for sustainable mining practices [20,21].

Additionally, artisanal mining is associated with severe issues such as child labor, moral degradation, gender inequalities, prostitution, drug-related problems, health issues, and fatal accidents. These problems arise from difficult working conditions, lack of social security, illegality, and informality [13,22,23,24].

1.2 Artisanal mining: Food security and nutrition

As highlighted in Section 1.1, artisanal mining is an undeniable source of livelihood for rural households in the Global South. Significant empirical works have flourished on this theme in recent years, since artisanal mining activities have been seen by many scholars as a complementary strategy to alleviate poverty [7,8,20,21]. This is particularly relevant given the challenges posed by rudimentary, seasonal, and low-productivity agriculture in most sub-Saharan countries [10].

The artisanal mining’s ability to create jobs, even if they are frequently precarious, makes it an alternative income-generating source that can significantly influence household poverty and compensate for the lack of opportunities for jobs in rural areas. With relatively higher returns compared to the agricultural sector in most developing countries, artisanal mining becomes a boon that attracts most young men to the extent that the volume of jobs created, and the induced external economies have a significant impact on the well-being of rural households to some extent at the expense of farm activities [15,22].

For instance, it has been in the literature observed that artisanal mining creates additional sources of income by providing seasonal employment for local communities mostly among unskilled rural young men. This additional income helps alleviate the burden of poverty and facilitates access to markets and essential services such as healthcare and education [1,16].

These studies primarily focus on the link between mining activities and poverty in communities reliant on them. However, they do not delve into the relationship between mining activities and food security.

Conversely, the study by Wegenast and Beck, dynamically addressed the intricate relationship between mining, rural livelihoods, and food security in Sub-Saharan Africa. The authors distinguish between aspects related to the distribution of property among national and international actors in the mining sector, as well as their respective impacts on food security parameters [25].

They revealed that mining activities, whether owned by national or international actors, can have both positive and negative impacts on food security, employment, and local economies. They not only highlight the potential benefits of mining, such as increased employment and infrastructure development, but also negative effects such as land dispossession, loss of access to natural resources, migratory influx, and disruptions in labor allocation within households. The combination of these elements can lead to increased living costs and make access to food difficult for local communities. In this instance, the authors found that mining activities improve access to food, particularly for men, while, on the other hand, they adversely affect child nutrition and women’s food security.

So, while beneficial to some extent, artisanal mining can be the potential channel to perpetuate transgenerational poverty, fostering a cycle of generational poverty and thus food insecurity among rural households [26].

However, the Wegenast and Beck study does not address the effects of mining at the household and local community levels, which can vary considerably depending on local and regional contexts. In the local context of Kasai Oriental, e.g., miners are marginalized in the diamond trade chain due to the informality of their work. This allows traders, traditional authorities, security services, gangs, and landowners to dominate and reap the most gains from artisanal mining. These factors severely limit the socioeconomic impact of artisanal mining on communities and households. This situation may contrast with other regions where artisanal mining is more structured and formalized [8,9,27]. Indeed, household participation in this industry can significantly affect their standard of living, access to resources, and family and community structures [9,15].

In this regard, the aim of our study is to fill this gap by specifically focusing on artisanal diamond mining activities due to their central role in the economy of Kasai Oriental since the sector’s liberalization in the early 1980s. Diamond mining is the only mineral resource exploited at both industrial and artisanal levels, providing a crucial livelihood for many households, especially in rural areas [28]. Given the sector’s socioeconomic importance and the environmental concerns associated with artisanal mining, targeted studies are necessary to explore sustainable development options for the region.

This study is grounded in the Sustainable livelihoods framework (SLF), which offers a holistic approach to understanding how various factors influence livelihoods, food security, and nutrition. The SLF allows for flexible research applications, highlighting specific aspects while considering the broader context [29,30]. Constrained in credit, time, and technology, rural households in Kasai Oriental attempt to adapt to agricultural shocks and risks by engaging in artisanal mining to earn the additional income necessary to improve their food and nutritional security as well as that of their children. Relying on this framework can allow us to understand the conditions under which household diversification into artisanal mining can or cannot lead to a sustainable development strategy in the rural areas of Kasai Oriental.

The focus of this study is on how rural households in Kasai Oriental employ their livelihood strategies to achieve well-being. Specifically, it analyzes the factors influencing household participation in diamond-based livelihoods and examines their impacts on food security and children’s nutrition. By doing so, this research seeks to enhance our understanding of the artisanal mining sector and its effects on household food security.

Our analysis provides localized insights into artisanal mining activities in this remote region and aims to fill a gap in the literature by examining the specific effects of artisanal diamond mining on household food security and children’s nutrition.

The remainder of the study is structured as follows: in Section 2, we will briefly look at the context of artisanal mining activities in Kasai Oriental. Section 3 discusses the methods and endogeneity issues. Section 4 presents the data used in the empirical analysis, Section 5 discusses the empirical results, and the last section provides some concluding remarks based on the findings.

3.1 Study design, sample size, and sampling frame

Among the 26 regions in DRC, Kasai Oriental stands out as one of the most impoverished and food-insecure areas. Located in the central part of DRC, its unfavorable geography makes it a landlocked region, with about 70% of the population residing in rural areas [35]. For this study, the data were collected through a cross-sectional household-based survey conducted in Kasai Oriental from November 17 to December 20, 2022, a rain and pre-harvest period characterized by hikes in food prices in the market. The study included variables at both the household and individual levels in rural areas, encompassing 310 households with 233 identified children to ensure a regionally representative sample.

Thus, we used a multi-stage sampling approach, combining purposive and simple random sampling methods, to ensure a representative sample across rural Kasai Oriental. Our sampling strategy comprised four levels. Initially, we purposively selected five rural health zones out of the nine in Kasai Oriental, ensuring each zone represented diverse administrative territories, geographic regions, and demographic characteristics. Within each selected health zone, we purposively chose 6–7 health areas from the 147 rural health areas, proportional to their demographic and geographic size to ensure representativeness. This stratified sampling approach included both densely and sparsely populated areas.

A simple random sampling technique using random numbers generated in Excel software 16 was employed to select villages and households for the survey from a list provided by local authorities and community leaders to minimize selection bias. However, we acknowledged that recent migrants or informal settlers might not be included in the village lists. If a selected household was unavailable or unwilling to participate, we used a replacement strategy, selecting the next household on the list. To circumvent this, we cross-referenced official lists with information from community health workers and local informants for better coverage.

The selection of 10 households per village was based on our sample size calculation, aiming for a cluster sample design of 310 households. This calculation was based on a food insecurity rate of 72% in Kasai Oriental, with Zα (the critical value for a two-tailed test at a 95% confidence level) of 1.96, and a margin of error (m) of 0.5, resulting in a sample size of 309, rounded up to 310. Consequently, we selected 10 households in each of the 31 villages.

3.2 Data collection and field procedure

The survey questions were translated into Tshiluba, the vernacular language spoken in Kasai Oriental, and pre-tested to ensure consistency and comprehensibility. During the survey, all questions were administered in Tshiluba using the KoboCollect tool phone-based survey. The data collection process involved three teams, each comprising two surveyors, making a total of six investigators. All of them were well-trained persons with prior experience in data collection with Catholic Relief Services and Action Contre la Faim in the Kasai Oriental.

Before the survey, all investigators received a comprehensive 5 days training in Mbuji Mayi, which included testing the questionnaires. Prior starting the survey, we discussed and obtained informed consent from all household representatives. We explained the study's purpose, their rights as participants, and how we would keep their information confidential to ensure they fully understood and agreed to participate. This approach helped us uphold ethical standards and establish trust with the participants before gathering any data. The collected data cover various aspects, including household demographics, socio-economic status, livelihood activities, and institutional circumstances. Specific information was gathered on household size, education level, and profession of the head of the household, sources of income, food security, access to markets and land, participation in mining activities, resource management, anthropometry, and child nutrition. Anthropometric data were collected from children aged 0–59 months, focusing on their height (in cm), weight (in kg), gender, and age. Mechanical weighing scales with an accuracy of 0.1 kg, and local wood infantometer measures with an accuracy of 0.1 cm, were used to gather these data. Children’s ages and birthdates were determined through memory recall and, when available, cross-verified with health, Christian, or administrative records. Data analysis was conducted using STATA 14 software. The data were summarized through descriptive statistics, including mean differences, cross-tabulations, linear fixed effects, and instrumental variable (IV) fixed effects models, to establish relationships between outcome variables and the variable of interest.

1. Ethical approval: The standard survey protocols, questionnaires, and anthropometric procedures related to children under 5 years old were reviewed and approved by the Medical Research Ethics Committee at the University of Mbuji Mayi in the Democratic Republic of Congo. The trial was officially registered under the reference number N/Ref:01/MREC/UM/NKL/2022 on September 14th, 2022.

2. Informed consent: Informed consent has been obtained from all household representatives.

3.3 Study outcome variables

In this study, we utilized the 9-item Household Food Insecurity Access Scale (HFIAS) module to assess households’ food security. This module is widely adopted as a standard measure of household food security in many African countries [28,29,30]. The HFIAS 9-item involves assigning raw scores based on specific household answers. Each question has four alternative codes: “0” for a “No” answer indicating no occurrence, “1” for rare, “2” for sometimes, and “3” for often.

The possible raw scores for any household range from 0 to 27, where 0 characterizes households as food secure and 27 as extremely food insecure [36]. Before using these raw scores, we checked for the internal consistency and reliability analysis of HFIAS questions. In this respect, Cronbach’s alpha score of 93.8% proved that the HFIAS 9-item had good internal consistency so that the data collected from them are reliable to analyze the food security in Kasai Oriental (see Appendix Table A1 for detail).

In another chapter, for assessing children’s nutritional status, we followed the World Health Organization (WHO) child growth reference and used anthropometric measures to establish three outcomes: height-for-age (HAZ) and weight-for-age (WAZ) indicators. We employed the Stata package zscore06 to generate z-scores based on the WHO’s growth standards [37]. The HAZ serves as an index to identify past undernutrition or chronic malnutrition. The WAZ is a composite indicator that provides information on changes in the prevalence of malnutrition over time. Children with HAZ and WAZ below minus two standard deviations (−2SD) from the WHO child growth reference are stunted and underweighted, respectively [37]. Missing data on height, weight, age, and gender, as well as measurement or coding errors, were excluded from our analyses. To achieve this, we applied WHO cut-off points for each child’s nutrition outcome for HAZ and WAZ, respectively. As a result, children with values falling within the following intervals were excluded: HAZ, ≤−6 or ≥6 SD; and WAZ, ≤−6 or ≥5 SD.

Thus, HFIAS raw scores and anthropometric z-scores are used in this study as continuous outcome variables.

3.4 Variable of interest and other variables

This study focuses on participation in artisanal diamond activities as the main variable of interest. It is coded as “1” for households with at least one adult who participated in artisanal mining during the last 30 days and “0” for all others. For an in-depth analysis, we also systematically control certain other variables. These additional factors are carefully selected considering their relevance to household food insecurity and poverty in many empirical studies [31,38,39]. They include household size and household head characteristics such as gender, level of education, marital status, and age. In addition, we include household dependency ratio, access to land, road conditions, livestock ownership, and household resource management. We also include other individual variables on children (gender, age) and the mother’s age. Since early marriages and births are still common in rural areas, young mothers often lack childcare experience, which can affect their children’s nutrition [40]. In regression analysis, this is especially important to consider in children under five because women play a primary role in childcare in this Kasai Oriental [10,37].

3.5 Empirical and estimation procedures

Given that the link between involvement in artisanal mining activities, food security, and children’s nutrition status remains little explored in empirical studies in Kasai Oriental, we employ village fixed effects along with instrumental variables models to capture the types of associations that exist in those variables while attempting to address the individual heterogeneity and the potential endogeneity of engagement in these activities. Household participation in artisanal diamond mining in Kasai Oriental is voluntary so it depends on various decision-making factors specific to each household, and thus an endogenous variable that ideally should be instrumented. Finding valid and appropriate instruments that meet relevance corr ( Z i , X i ) ≠ 0 and exogeneity corr ( Z i , U i ) = 0 criteria is often challenging in practice and involves ad hoc selection of variables [41,42]. However, in the literature, many researchers adopt the practical approach consisting of selecting instruments that are supposed to be exclusively related to endogenous variables. For instance, financial availability for equipment purchase, poverty, and transportation costs have been identified as factors motivating household members to engage in artisanal mining activities in Taita Taveta County in Tanzania [43].

Considering the approach, we identify and select potential instruments to be included in the regression. These are artisanal mining sites’ presence in sub-areas and the unemployment rate by territory. The presence of mining sites in sub-areas or not is expected to be a motivating factor for artisanal diamond participation. The unemployment rate reflects differences in unemployment distribution across territories, potentially leading individuals without jobs to engage in artisanal diamond activities. The instrumental variables were collected through the Division of Mines and Hydrocarbons and the Division of Labor and Social Foresight [37,44,45].

We applied the standard heuristic relevance threshold criterion (F > 10) to identify statistically significant associations between instruments and our endogenous variable. The results revealed the relevance of our instruments and are the motivating factor of participation in artisanal diamond mining. To assess under-identification, instrument strengths, and over-identification, respectively, we employed the Anderson–Rubin Wald test and Stock–Wright LM S statistic joint tests for robust inference with weak instruments. We also employed the Cragg–Donald and Kleibergen–Paap (1993) statistics based on the Stock and Yogo (2005) criteria for assessing instrument weakness, along with the over-identification test based on the Sargan–Hansen (1982) J-statistic. The results of these tests confirm the strength and reliability of our instruments (Table 6).

Following the work of Colin Cameron and Miller to account for possible heteroscedasticity and autocorrelation, robust standard errors clustered at village levels have been employed to improve the accuracy and reliability of associations that exist between engagement in artisanal mining activities, household food security, and children’s health conditions [46]. The key equations used in this study are as follows:

Participation equation

Food insecurity equation

Children’s nutritional equation

where z i represents a series of identified instruments; x i is a vector of household characteristics and covariate variables including participation in artisanal mining, household size, household dependency ratio, household head characteristics (gender, level of education, marital status, and also age), access to land, road conditions, farmland size, corn yields, livestock ownership, and household resource management, and other individual variables on children (gender, age) and the mother’s age; Y i represents the food insecurity variable (continuous); and C i denotes the continuous variable for children’s nutritional status.

So, the statistical models in this study are implemented using a robust methodological framework that combines linear village fixed effects and instrumental variables to address endogeneity and heterogeneity. The Instrumental variable (IV) estimation is implemented in two steps. First, using the selected instruments, a binary response model is used to fit the participation in artisanal diamond mining. In the second step, the values of food security and children’s anthropometry outcomes are regressed on participation in artisanal diamond mining and other household characteristics. Standard errors are clustered at the village level to correct for possible heteroscedasticity and autocorrelation, enhancing the reliability of the results. The IV models address endogeneity issues while fixed effects are included to control for unobserved time-invariant heterogeneity [47,48,49]. Combining these methods allows for a more robust analysis by addressing both types of biases. Data treatment and estimation methods were implemented using Excel 16 and STATA 14.2 software (StataCorp, USA) (Table 1).

4.1 Descriptive statistics

In total, 308 households were analyzed based on the HFIAS questionnaire. A significant proportion of households in Kasai Oriental have experienced food insecurity in all conditions (Table 2). In the first domain, 4.33% of households answered “no” in the full sample, indicating that 95.77% of sampled households felt anxious and uncertain about food availability. This proportion is similar for both participants and non-participants in artisanal diamond mining. In the second domain, the average proportion of “no” responses was 14%, suggesting that 86% of households in the full sample have insufficient food quality. However, participants have relatively better access to quality food, with 15.2% of them having access compared to 13.19% of non-participants. In the last domain, the average proportion of “no” responses was 30.58%, indicating that about 69.42% of households in the full sample have inadequate food quantity intake. Participants, on the other hand, have a slightly higher food intake at 31.36% compared to 29.93% for non-participants.

In Table 3, we can observe that the distribution of HAZ and WAZ dimensions between non-participant and participant households shows a slightly similar pattern. The mean difference in both dimensions of children’s health is not statistically significant, suggesting that there is likely no difference in the health conditions of children between participants and non-participants in artisanal diamond mining in rural households in Kasai Oriental.

Figures 2 and 3, and Table 4, respectively, present the overall distribution of HAZ and WAZ based on participation in artisanal diamond mining while disaggregating the extent of malnutrition in both a general and gender-specific breakdown. The analysis by gender reveals a predominance of boys experiencing stunting, underweight, and wasting. Overall, in the examined sample, 56.21% of children exhibit stunting (HAZ < −2 SD), and 44.79% are underweight (WAZ < −2 SD), these figures single out leftward-skewed distribution across, indicating the prevalence of malnutrition in the Kasai Oriental region (Figures 2 and 3).

Figure 2

Children’s anthropometric outcomes distribution.

Figure 3

Children’s anthropometric outcome distributions by artisanal diamond participation.

Table 5 displays descriptive statistics of variables categorized by diamond participation in the econometric models. We find significant differences between those involved in artisanal diamond activities and those who are not. On average, participants experience lower levels of food insecurity, with a mean score of 15.82, compared to 17.15 for non-participants. This difference is statistically significant at the 1% level based on t-test results.

Furthermore, when it comes to educational attainment, non-participants have lower levels of illiteracy compared to participants, with a significant mean difference at the 5% level. On the other hand, participants are more likely to be engaged in livestock activity than non-participants, with a significant mean difference at the 5% level. While these descriptive variations do not imply causation, they suggest that there might be structural differences in food security, access to land and agricultural inputs, livestock possession, and educational opportunities between participants and non-participants.

4.2 Regression results